diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-1152.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-1152.json new file mode 100644 index 0000000000000000000000000000000000000000..227e3a8cde2829d80bc774c4480cb5c8270fa709 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-1152.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-1152", + "checkpoint": "checkpoint-1152", + "global_step": 1152, + "tokens_seen": 1207959552, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25144393547102173, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25144393547102173 + ], + "acc": 0.2545309699263095, + "acc_ci": 0.0, + "acc_runs": [ + 0.2545309699263095 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043295650165269565, + "centered_acc": 0.0019252472946956385, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0019252472946956385 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5168661588683352, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5168661588683352 + ], + "acc": 0.5310119695321001, + "acc_ci": 0.0, + "acc_runs": [ + 0.5310119695321001 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011659185184878962, + "centered_acc": 0.0337323177366704, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0337323177366704 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2647306397306397, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2647306397306397 + ], + "acc": 0.26641414141414144, + "acc_ci": 0.0, + "acc_runs": [ + 0.26641414141414144 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009053021086174166, + "centered_acc": 0.019640852974186274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019640852974186274 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25 + ], + "acc": 0.197098976109215, + "acc_ci": 0.0, + "acc_runs": [ + 0.197098976109215 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012653835621466646, + "centered_acc": 0.0, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.276, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.276 + ], + "acc": 0.166, + "acc_ci": 0.0, + "acc_runs": [ + 0.166 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02001121929807355, + "centered_acc": 0.0346666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0346666666666667 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5059194948697711, + "acc_ci": 0.0, + "acc_runs": [ + 0.5059194948697711 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051500838485923, + "centered_acc": 0.011838989739542116, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011838989739542116 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.46, + "acc_ci": 0.0, + "acc_runs": [ + 0.46 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620332, + "centered_acc": -0.07999999999999996, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07999999999999996 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 5518097.083106795, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 5518097.083106795 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.18, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.18 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012155153135512024, + "centered_acc": -0.09333333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09333333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2898401223449994, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28062630136252376, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.03499492615268573, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2898401223449994 + ], + "mean_acc_runs": [ + 0.28062630136252376 + ], + "mean_centered_acc_runs": [ + -0.03499492615268573 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-1440.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-1440.json new file mode 100644 index 0000000000000000000000000000000000000000..7d0fe1d1719e44bae5d6ce5fff17839531707a81 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-1440.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-1440", + "checkpoint": "checkpoint-1440", + "global_step": 1440, + "tokens_seen": 1509949440, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2533359888468433, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2533359888468433 + ], + "acc": 0.2531368253335989, + "acc_ci": 0.0, + "acc_runs": [ + 0.2531368253335989 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004340328204135388, + "centered_acc": 0.0044479851291243655, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0044479851291243655 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5114254624591947, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5114254624591947 + ], + "acc": 0.5342763873775843, + "acc_ci": 0.0, + "acc_runs": [ + 0.5342763873775843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011662778026451529, + "centered_acc": 0.02285092491838947, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02285092491838947 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2702020202020202, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2702020202020202 + ], + "acc": 0.26052188552188554, + "acc_ci": 0.0, + "acc_runs": [ + 0.26052188552188554 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009112002229119622, + "centered_acc": 0.026936026936026963, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.026936026936026963 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2645051194539249, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2645051194539249 + ], + "acc": 0.2150170648464164, + "acc_ci": 0.0, + "acc_runs": [ + 0.2150170648464164 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012889272949313479, + "centered_acc": 0.019340159271899877, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019340159271899877 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.168, + "acc_ci": 0.0, + "acc_runs": [ + 0.168 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5082872928176796, + "acc_ci": 0.0, + "acc_runs": [ + 0.5082872928176796 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405055532282417, + "centered_acc": 0.016574585635359185, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.016574585635359185 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.52, + "acc_ci": 0.0, + "acc_runs": [ + 0.52 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05021167315686783, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 8775376.900286412, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 8775376.900286412 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.179, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.179 + ], + "acc": 0.049, + "acc_ci": 0.0, + "acc_runs": [ + 0.049 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012128730605719054, + "centered_acc": -0.09466666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09466666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2934114318269972, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2886526917670865, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.018363932729088834, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2934114318269972 + ], + "mean_acc_runs": [ + 0.2886526917670865 + ], + "mean_centered_acc_runs": [ + -0.018363932729088834 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-1728.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-1728.json new file mode 100644 index 0000000000000000000000000000000000000000..bb4ab45f3e98e5deaec99a183549fd1f54c42155 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-1728.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-1728", + "checkpoint": "checkpoint-1728", + "global_step": 1728, + "tokens_seen": 1811939328, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2492531368253336, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2492531368253336 + ], + "acc": 0.25253933479386576, + "acc_ci": 0.0, + "acc_runs": [ + 0.25253933479386576 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00431696567867478, + "centered_acc": -0.0009958175662218756, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0009958175662218756 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.515778019586507, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.515778019586507 + ], + "acc": 0.5331882480957563, + "acc_ci": 0.0, + "acc_runs": [ + 0.5331882480957563 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011660014400426034, + "centered_acc": 0.03155603917301408, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03155603917301408 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26936026936026936, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26936026936026936 + ], + "acc": 0.26304713804713803, + "acc_ci": 0.0, + "acc_runs": [ + 0.26304713804713803 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009103043207756808, + "centered_acc": 0.02581369248035914, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02581369248035914 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26023890784982934, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26023890784982934 + ], + "acc": 0.20648464163822525, + "acc_ci": 0.0, + "acc_runs": [ + 0.20648464163822525 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012821930225112672, + "centered_acc": 0.013651877133105783, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013651877133105783 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.296, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.296 + ], + "acc": 0.17, + "acc_ci": 0.0, + "acc_runs": [ + 0.17 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020435342091896222, + "centered_acc": 0.061333333333333316, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.061333333333333316 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.510655090765588, + "acc_ci": 0.0, + "acc_runs": [ + 0.510655090765588 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014049294536290372, + "centered_acc": 0.02131018153117603, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02131018153117603 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.51, + "acc_ci": 0.0, + "acc_runs": [ + 0.51 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05024183937956913, + "centered_acc": 0.020000000000000018, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020000000000000018 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 15918475.055290213, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 15918475.055290213 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.182, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.182 + ], + "acc": 0.05, + "acc_ci": 0.0, + "acc_runs": [ + 0.05 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0122075806376622, + "centered_acc": -0.09066666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09066666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29543838893698987, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28742019151142734, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.017935826337166616, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29543838893698987 + ], + "mean_acc_runs": [ + 0.28742019151142734 + ], + "mean_centered_acc_runs": [ + -0.017935826337166616 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-2016.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-2016.json new file mode 100644 index 0000000000000000000000000000000000000000..1837579ba9162b0ccda127c5f16c4a43af57a377 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-2016.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-2016", + "checkpoint": "checkpoint-2016", + "global_step": 2016, + "tokens_seen": 2113929216, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25492929695279826, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25492929695279826 + ], + "acc": 0.25363473411670984, + "acc_ci": 0.0, + "acc_runs": [ + 0.25363473411670984 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043493077027353935, + "centered_acc": 0.006572395937064342, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006572395937064342 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5228509249183896, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5228509249183896 + ], + "acc": 0.5331882480957563, + "acc_ci": 0.0, + "acc_runs": [ + 0.5331882480957563 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011653634832401168, + "centered_acc": 0.04570184983677916, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04570184983677916 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26515151515151514, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26515151515151514 + ], + "acc": 0.25673400673400676, + "acc_ci": 0.0, + "acc_runs": [ + 0.25673400673400676 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009057621139172679, + "centered_acc": 0.020202020202020183, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020202020202020183 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2627986348122867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2627986348122867 + ], + "acc": 0.2090443686006826, + "acc_ci": 0.0, + "acc_runs": [ + 0.2090443686006826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012862523175351397, + "centered_acc": 0.017064846416382284, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017064846416382284 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.284, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.284 + ], + "acc": 0.172, + "acc_ci": 0.0, + "acc_runs": [ + 0.172 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020186703693570774, + "centered_acc": 0.0453333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0453333333333333 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5098658247829518, + "acc_ci": 0.0, + "acc_runs": [ + 0.5098658247829518 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014049749833367674, + "centered_acc": 0.019731649565903675, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019731649565903675 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5, + "acc_ci": 0.0, + "acc_runs": [ + 0.5 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.050251890762960605, + "centered_acc": 0.0, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 29474969.753621306, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 29474969.753621306 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.041, + "acc_ci": 0.0, + "acc_runs": [ + 0.041 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2951217286391649, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28537546441038075, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.020090997906790715, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2951217286391649 + ], + "mean_acc_runs": [ + 0.28537546441038075 + ], + "mean_centered_acc_runs": [ + -0.020090997906790715 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-2304.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-2304.json new file mode 100644 index 0000000000000000000000000000000000000000..86431d13261b47236825476f180ceb44a5cab5b0 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-2304.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-2304", + "checkpoint": "checkpoint-2304", + "global_step": 2304, + "tokens_seen": 2415919104, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25423222465644296, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25423222465644296 + ], + "acc": 0.25562636924915355, + "acc_ci": 0.0, + "acc_runs": [ + 0.25562636924915355 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00434538861452022, + "centered_acc": 0.005642966208590616, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005642966208590616 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5223068552774756, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5223068552774756 + ], + "acc": 0.5391730141458106, + "acc_ci": 0.0, + "acc_runs": [ + 0.5391730141458106 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011654208652596384, + "centered_acc": 0.044613710554951114, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.044613710554951114 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.273989898989899, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.273989898989899 + ], + "acc": 0.2702020202020202, + "acc_ci": 0.0, + "acc_runs": [ + 0.2702020202020202 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00915180590154416, + "centered_acc": 0.03198653198653201, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03198653198653201 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2593856655290102, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2593856655290102 + ], + "acc": 0.2167235494880546, + "acc_ci": 0.0, + "acc_runs": [ + 0.2167235494880546 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012808273573927224, + "centered_acc": 0.012514220705346949, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012514220705346949 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.18, + "acc_ci": 0.0, + "acc_runs": [ + 0.18 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.500394632991318, + "acc_ci": 0.0, + "acc_runs": [ + 0.500394632991318 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405248130604961, + "centered_acc": 0.0007892659826360671, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0007892659826360671 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.53, + "acc_ci": 0.0, + "acc_runs": [ + 0.53 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05016135580465919, + "centered_acc": 0.06000000000000005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06000000000000005 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 41786572.40411945, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 41786572.40411945 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2941524407421379, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29174070478500574, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.01643093122383806, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2941524407421379 + ], + "mean_acc_runs": [ + 0.29174070478500574 + ], + "mean_centered_acc_runs": [ + -0.01643093122383806 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-2592.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-2592.json new file mode 100644 index 0000000000000000000000000000000000000000..3813f2e51bd418fff40612de30983b29d7e19647 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-2592.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-2592", + "checkpoint": "checkpoint-2592", + "global_step": 2592, + "tokens_seen": 2717908992, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25413264289982074, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25413264289982074 + ], + "acc": 0.255327623979287, + "acc_ci": 0.0, + "acc_runs": [ + 0.255327623979287 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004344827546976306, + "centered_acc": 0.005510190533094317, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005510190533094317 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5038084874863983, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5038084874863983 + ], + "acc": 0.5261153427638737, + "acc_ci": 0.0, + "acc_runs": [ + 0.5261153427638737 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665485744747022, + "centered_acc": 0.007616974972796564, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007616974972796564 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26515151515151514, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26515151515151514 + ], + "acc": 0.26304713804713803, + "acc_ci": 0.0, + "acc_runs": [ + 0.26304713804713803 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009057621139172679, + "centered_acc": 0.020202020202020183, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020202020202020183 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2645051194539249, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2645051194539249 + ], + "acc": 0.2235494880546075, + "acc_ci": 0.0, + "acc_runs": [ + 0.2235494880546075 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012889272949313479, + "centered_acc": 0.019340159271899877, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019340159271899877 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.18, + "acc_ci": 0.0, + "acc_runs": [ + 0.18 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.510655090765588, + "acc_ci": 0.0, + "acc_runs": [ + 0.510655090765588 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014049294536290372, + "centered_acc": 0.02131018153117603, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02131018153117603 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.51, + "acc_ci": 0.0, + "acc_runs": [ + 0.51 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05024183937956913, + "centered_acc": 0.020000000000000018, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020000000000000018 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 81037210.12118527, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 81037210.12118527 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.182, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.182 + ], + "acc": 0.04, + "acc_ci": 0.0, + "acc_runs": [ + 0.04 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0122075806376622, + "centered_acc": -0.09066666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09066666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29059962749860985, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28869821453841943, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.023123579623142113, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29059962749860985 + ], + "mean_acc_runs": [ + 0.28869821453841943 + ], + "mean_centered_acc_runs": [ + -0.023123579623142113 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-288.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-288.json new file mode 100644 index 0000000000000000000000000000000000000000..13a7e2d6ccf4021687cc61e3126086be6ce9d721 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-288.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-288", + "checkpoint": "checkpoint-288", + "global_step": 288, + "tokens_seen": 301989888, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25562636924915355, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25562636924915355 + ], + "acc": 0.2548297151961761, + "acc_ci": 0.0, + "acc_runs": [ + 0.2548297151961761 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004353212146198103, + "centered_acc": 0.007501825665538068, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007501825665538068 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5021762785636561, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5021762785636561 + ], + "acc": 0.5239390642002176, + "acc_ci": 0.0, + "acc_runs": [ + 0.5239390642002176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665713661738719, + "centered_acc": 0.0043525571273121955, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0043525571273121955 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2558922558922559, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2558922558922559 + ], + "acc": 0.24831649831649832, + "acc_ci": 0.0, + "acc_runs": [ + 0.24831649831649832 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008953950243013958, + "centered_acc": 0.007856341189674524, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007856341189674524 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2627986348122867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2627986348122867 + ], + "acc": 0.2030716723549488, + "acc_ci": 0.0, + "acc_runs": [ + 0.2030716723549488 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012862523175351397, + "centered_acc": 0.017064846416382284, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017064846416382284 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.284, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.284 + ], + "acc": 0.162, + "acc_ci": 0.0, + "acc_runs": [ + 0.162 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020186703693570774, + "centered_acc": 0.0453333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0453333333333333 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5201262825572218, + "acc_ci": 0.0, + "acc_runs": [ + 0.5201262825572218 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014041096664344348, + "centered_acc": 0.04025256511444364, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04025256511444364 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.6, + "acc_ci": 0.0, + "acc_runs": [ + 0.6 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0492365963917331, + "centered_acc": 0.19999999999999996, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.19999999999999996 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 335678.3049185522, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 335678.3049185522 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.173, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.173 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011967214137559927, + "centered_acc": -0.10266666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10266666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.28891558975289205, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29355706943987625, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0026366971413980087, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.28891558975289205 + ], + "mean_acc_runs": [ + 0.29355706943987625 + ], + "mean_centered_acc_runs": [ + -0.0026366971413980087 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-2880.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-2880.json new file mode 100644 index 0000000000000000000000000000000000000000..9bb78f3cc7e048905ede3235b2aa9ed48265503a --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-2880.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-2880", + "checkpoint": "checkpoint-2880", + "global_step": 2880, + "tokens_seen": 3019898880, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2522405895239992, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2522405895239992 + ], + "acc": 0.25323640709022105, + "acc_ci": 0.0, + "acc_runs": [ + 0.25323640709022105 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004334110169012113, + "centered_acc": 0.00298745269866559, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.00298745269866559 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5103373231773667, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5103373231773667 + ], + "acc": 0.529923830250272, + "acc_ci": 0.0, + "acc_runs": [ + 0.529923830250272 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01166333067307569, + "centered_acc": 0.020674646354733373, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020674646354733373 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2668350168350168, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2668350168350168 + ], + "acc": 0.26052188552188554, + "acc_ci": 0.0, + "acc_runs": [ + 0.26052188552188554 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009075915859266983, + "centered_acc": 0.02244668911335575, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02244668911335575 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.27303754266211605, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27303754266211605 + ], + "acc": 0.22098976109215018, + "acc_ci": 0.0, + "acc_runs": [ + 0.22098976109215018 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01301933276263586, + "centered_acc": 0.030716723549488067, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.030716723549488067 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.292, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.292 + ], + "acc": 0.184, + "acc_ci": 0.0, + "acc_runs": [ + 0.184 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020354375480530096, + "centered_acc": 0.05599999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05599999999999997 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.500394632991318, + "acc_ci": 0.0, + "acc_runs": [ + 0.500394632991318 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405248130604961, + "centered_acc": 0.0007892659826360671, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0007892659826360671 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.52, + "acc_ci": 0.0, + "acc_runs": [ + 0.52 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05021167315686783, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 132665039.98033264, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 132665039.98033264 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.178, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.178 + ], + "acc": 0.042, + "acc_ci": 0.0, + "acc_runs": [ + 0.042 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012102167676183571, + "centered_acc": -0.09600000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09600000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29540841203308316, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2889353978719547, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.01842336652819117, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29540841203308316 + ], + "mean_acc_runs": [ + 0.2889353978719547 + ], + "mean_centered_acc_runs": [ + -0.01842336652819117 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-3168.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-3168.json new file mode 100644 index 0000000000000000000000000000000000000000..06d33199f2824f135690c66e8e6122d725925f8a --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-3168.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-3168", + "checkpoint": "checkpoint-3168", + "global_step": 3168, + "tokens_seen": 3321888768, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25423222465644296, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25423222465644296 + ], + "acc": 0.25124477195777734, + "acc_ci": 0.0, + "acc_runs": [ + 0.25124477195777734 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00434538861452022, + "centered_acc": 0.005642966208590616, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005642966208590616 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.514689880304679, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.514689880304679 + ], + "acc": 0.5239390642002176, + "acc_ci": 0.0, + "acc_runs": [ + 0.5239390642002176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011660788281735552, + "centered_acc": 0.029379760609357986, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.029379760609357986 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26641414141414144, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26641414141414144 + ], + "acc": 0.26136363636363635, + "acc_ci": 0.0, + "acc_runs": [ + 0.26136363636363635 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009071357971078843, + "centered_acc": 0.021885521885521914, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021885521885521914 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2636518771331058, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2636518771331058 + ], + "acc": 0.2150170648464164, + "acc_ci": 0.0, + "acc_runs": [ + 0.2150170648464164 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012875929151296968, + "centered_acc": 0.018202502844141044, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018202502844141044 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.302, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.302 + ], + "acc": 0.18, + "acc_ci": 0.0, + "acc_runs": [ + 0.18 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02055326917420912, + "centered_acc": 0.06933333333333332, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06933333333333332 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.494869771112865, + "acc_ci": 0.0, + "acc_runs": [ + 0.494869771112865 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051745961790525, + "centered_acc": -0.010260457774269982, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.010260457774269982 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.49, + "acc_ci": 0.0, + "acc_runs": [ + 0.49 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05024183937956913, + "centered_acc": -0.020000000000000018, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.020000000000000018 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 198677019.72047818, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 198677019.72047818 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.177, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.177 + ], + "acc": 0.048, + "acc_ci": 0.0, + "acc_runs": [ + 0.048 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012075463420375087, + "centered_acc": -0.09733333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09733333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29633135391806154, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2842721770254613, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.02517497585325087, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29633135391806154 + ], + "mean_acc_runs": [ + 0.2842721770254613 + ], + "mean_centered_acc_runs": [ + -0.02517497585325087 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-3456.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-3456.json new file mode 100644 index 0000000000000000000000000000000000000000..911597657c7e4a0b688b7af08071e6b1887194cf --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-3456.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-3456", + "checkpoint": "checkpoint-3456", + "global_step": 3456, + "tokens_seen": 3623878656, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2529376618203545, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2529376618203545 + ], + "acc": 0.2533359888468433, + "acc_ci": 0.0, + "acc_runs": [ + 0.2533359888468433 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043380713189125845, + "centered_acc": 0.003916882427139316, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.003916882427139316 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5097932535364527, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5097932535364527 + ], + "acc": 0.5206746463547334, + "acc_ci": 0.0, + "acc_runs": [ + 0.5206746463547334 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011663586263283307, + "centered_acc": 0.019586507072905324, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019586507072905324 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26936026936026936, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26936026936026936 + ], + "acc": 0.26304713804713803, + "acc_ci": 0.0, + "acc_runs": [ + 0.26304713804713803 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009103043207756808, + "centered_acc": 0.02581369248035914, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02581369248035914 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26535836177474403, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26535836177474403 + ], + "acc": 0.22013651877133106, + "acc_ci": 0.0, + "acc_runs": [ + 0.22013651877133106 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012902554762313832, + "centered_acc": 0.020477815699658713, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020477815699658713 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.178, + "acc_ci": 0.0, + "acc_runs": [ + 0.178 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5059194948697711, + "acc_ci": 0.0, + "acc_runs": [ + 0.5059194948697711 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051500838485923, + "centered_acc": 0.011838989739542116, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011838989739542116 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 284454300.4646191, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 284454300.4646191 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.18, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.18 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012155153135512024, + "centered_acc": -0.09333333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09333333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29324159108197007, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29054012486635167, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0147175395221846, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29324159108197007 + ], + "mean_acc_runs": [ + 0.29054012486635167 + ], + "mean_centered_acc_runs": [ + -0.0147175395221846 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-3744.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-3744.json new file mode 100644 index 0000000000000000000000000000000000000000..6c01ff4eb1fa3ab64ec0327a257994faec88acf8 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-3744.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-3744", + "checkpoint": "checkpoint-3744", + "global_step": 3744, + "tokens_seen": 3925868544, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2517426807408883, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2517426807408883 + ], + "acc": 0.24995020912168892, + "acc_ci": 0.0, + "acc_runs": [ + 0.24995020912168892 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004331271717773648, + "centered_acc": 0.0023235743211843887, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0023235743211843887 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5119695321001088, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5119695321001088 + ], + "acc": 0.5217627856365615, + "acc_ci": 0.0, + "acc_runs": [ + 0.5217627856365615 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011662480968070035, + "centered_acc": 0.02393906420021752, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02393906420021752 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2622053872053872, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2622053872053872 + ], + "acc": 0.25252525252525254, + "acc_ci": 0.0, + "acc_runs": [ + 0.25252525252525254 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009025197991724937, + "centered_acc": 0.01627384960718296, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01627384960718296 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2636518771331058, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2636518771331058 + ], + "acc": 0.22013651877133106, + "acc_ci": 0.0, + "acc_runs": [ + 0.22013651877133106 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012875929151296968, + "centered_acc": 0.018202502844141044, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018202502844141044 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.294, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.294 + ], + "acc": 0.176, + "acc_ci": 0.0, + "acc_runs": [ + 0.176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02039509548493655, + "centered_acc": 0.058666666666666645, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.058666666666666645 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5011838989739542, + "acc_ci": 0.0, + "acc_runs": [ + 0.5011838989739542 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052446290528871, + "centered_acc": 0.0023677979479084232, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0023677979479084232 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 371243385.9219187, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 371243385.9219187 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.175, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.175 + ], + "acc": 0.042, + "acc_ci": 0.0, + "acc_runs": [ + 0.042 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012021627157731998, + "centered_acc": -0.10000000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10000000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29309491286324835, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2891846126802488, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.013516846762810923, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29309491286324835 + ], + "mean_acc_runs": [ + 0.2891846126802488 + ], + "mean_centered_acc_runs": [ + -0.013516846762810923 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-4032.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-4032.json new file mode 100644 index 0000000000000000000000000000000000000000..ce70f55876454437baadb90720e6666dd975fe94 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-4032.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-4032", + "checkpoint": "checkpoint-4032", + "global_step": 4032, + "tokens_seen": 4227858432, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25363473411670984, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25363473411670984 + ], + "acc": 0.24995020912168892, + "acc_ci": 0.0, + "acc_runs": [ + 0.24995020912168892 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004342017709967513, + "centered_acc": 0.0048463121556131155, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0048463121556131155 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.514145810663765, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.514145810663765 + ], + "acc": 0.5408052230685527, + "acc_ci": 0.0, + "acc_runs": [ + 0.5408052230685527 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011661154475524689, + "centered_acc": 0.028291621327529937, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.028291621327529937 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26262626262626265, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26262626262626265 + ], + "acc": 0.26136363636363635, + "acc_ci": 0.0, + "acc_runs": [ + 0.26136363636363635 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009029861776763957, + "centered_acc": 0.01683501683501687, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01683501683501687 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2551194539249147, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2551194539249147 + ], + "acc": 0.21245733788395904, + "acc_ci": 0.0, + "acc_runs": [ + 0.21245733788395904 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012739038695202038, + "centered_acc": 0.006825938566552929, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006825938566552929 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.18, + "acc_ci": 0.0, + "acc_runs": [ + 0.18 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48855564325177586, + "acc_ci": 0.0, + "acc_runs": [ + 0.48855564325177586 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014048804199859391, + "centered_acc": -0.022888713496448276, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.022888713496448276 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.53, + "acc_ci": 0.0, + "acc_runs": [ + 0.53 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05016135580465919, + "centered_acc": 0.06000000000000005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06000000000000005 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 540085106.0808784, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 540085106.0808784 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.179, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.179 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012128730605719054, + "centered_acc": -0.09466666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09466666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2917543768886087, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2885419511463313, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.021797951970111272, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2917543768886087 + ], + "mean_acc_runs": [ + 0.2885419511463313 + ], + "mean_centered_acc_runs": [ + -0.021797951970111272 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-4320.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-4320.json new file mode 100644 index 0000000000000000000000000000000000000000..367293a83e3c4e0f9f2271ee82fb0860d3c751f7 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-4320.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-4320", + "checkpoint": "checkpoint-4320", + "global_step": 4320, + "tokens_seen": 4529848320, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.252141007767377, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.252141007767377 + ], + "acc": 0.2515435172276439, + "acc_ci": 0.0, + "acc_runs": [ + 0.2515435172276439 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004333543083293202, + "centered_acc": 0.0028546770231693643, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0028546770231693643 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5239390642002176, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5239390642002176 + ], + "acc": 0.5369967355821545, + "acc_ci": 0.0, + "acc_runs": [ + 0.5369967355821545 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165244562107948, + "centered_acc": 0.04787812840043526, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04787812840043526 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2668350168350168, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2668350168350168 + ], + "acc": 0.2617845117845118, + "acc_ci": 0.0, + "acc_runs": [ + 0.2617845117845118 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009075915859266983, + "centered_acc": 0.02244668911335575, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02244668911335575 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26109215017064846, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26109215017064846 + ], + "acc": 0.2090443686006826, + "acc_ci": 0.0, + "acc_runs": [ + 0.2090443686006826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012835523909473708, + "centered_acc": 0.014789533560864617, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014789533560864617 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.29, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.29 + ], + "acc": 0.176, + "acc_ci": 0.0, + "acc_runs": [ + 0.176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020313179231745138, + "centered_acc": 0.05333333333333331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05333333333333331 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5098658247829518, + "acc_ci": 0.0, + "acc_runs": [ + 0.5098658247829518 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014049749833367674, + "centered_acc": 0.019731649565903675, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019731649565903675 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 879369525.3834081, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 879369525.3834081 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.175, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.175 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012021627157731998, + "centered_acc": -0.10000000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10000000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2948345398288767, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2917522419751645, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.009154562828393034, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2948345398288767 + ], + "mean_acc_runs": [ + 0.2917522419751645 + ], + "mean_centered_acc_runs": [ + -0.009154562828393034 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-4608.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-4608.json new file mode 100644 index 0000000000000000000000000000000000000000..5056cfdb6c7addb4e269d8222243d549c68ddf64 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-4608.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-4608", + "checkpoint": "checkpoint-4608", + "global_step": 4608, + "tokens_seen": 4831838208, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2513443537143995, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2513443537143995 + ], + "acc": 0.2526389165504879, + "acc_ci": 0.0, + "acc_runs": [ + 0.2526389165504879 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043289955103121775, + "centered_acc": 0.001792471619199339, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.001792471619199339 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5174102285092492, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5174102285092492 + ], + "acc": 0.5201305767138193, + "acc_ci": 0.0, + "acc_runs": [ + 0.5201305767138193 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011658749823107507, + "centered_acc": 0.03482045701849845, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03482045701849845 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26515151515151514, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26515151515151514 + ], + "acc": 0.26430976430976433, + "acc_ci": 0.0, + "acc_runs": [ + 0.26430976430976433 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009057621139172679, + "centered_acc": 0.020202020202020183, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020202020202020183 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2551194539249147, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2551194539249147 + ], + "acc": 0.2030716723549488, + "acc_ci": 0.0, + "acc_runs": [ + 0.2030716723549488 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012739038695202038, + "centered_acc": 0.006825938566552929, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006825938566552929 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.29, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.29 + ], + "acc": 0.178, + "acc_ci": 0.0, + "acc_runs": [ + 0.178 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020313179231745138, + "centered_acc": 0.05333333333333331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05333333333333331 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5043409629044988, + "acc_ci": 0.0, + "acc_runs": [ + 0.5043409629044988 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051956064076884, + "centered_acc": 0.008681925808997626, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008681925808997626 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 1265625603.7055895, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 1265625603.7055895 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.177, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.177 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012075463420375087, + "centered_acc": -0.09733333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09733333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29267092521667976, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.288377935460722, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.01501136258192564, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29267092521667976 + ], + "mean_acc_runs": [ + 0.288377935460722 + ], + "mean_centered_acc_runs": [ + -0.01501136258192564 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-4896.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-4896.json new file mode 100644 index 0000000000000000000000000000000000000000..92cccd5bc30f8465bf2718bf0fd1b54376b4be14 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-4896.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-4896", + "checkpoint": "checkpoint-4896", + "global_step": 4896, + "tokens_seen": 5133828096, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25363473411670984, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25363473411670984 + ], + "acc": 0.2500497908783111, + "acc_ci": 0.0, + "acc_runs": [ + 0.2500497908783111 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004342017709967513, + "centered_acc": 0.0048463121556131155, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0048463121556131155 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.49836779107725787, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.49836779107725787 + ], + "acc": 0.5255712731229597, + "acc_ci": 0.0, + "acc_runs": [ + 0.5255712731229597 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665762007194807, + "centered_acc": -0.0032644178454842576, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0032644178454842576 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2622053872053872, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2622053872053872 + ], + "acc": 0.2558922558922559, + "acc_ci": 0.0, + "acc_runs": [ + 0.2558922558922559 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009025197991724937, + "centered_acc": 0.01627384960718296, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01627384960718296 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2508532423208191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2508532423208191 + ], + "acc": 0.20648464163822525, + "acc_ci": 0.0, + "acc_runs": [ + 0.20648464163822525 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012668198621315347, + "centered_acc": 0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.172, + "acc_ci": 0.0, + "acc_runs": [ + 0.172 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4980268350434096, + "acc_ci": 0.0, + "acc_runs": [ + 0.4980268350434096 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052376259225643, + "centered_acc": -0.003946329913180779, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.003946329913180779 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 1603216289.6871345, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 1603216289.6871345 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.172, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.172 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011939788882495308, + "centered_acc": -0.10400000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10400000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.28617685912002905, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2870312258348862, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.023597556224523267, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.28617685912002905 + ], + "mean_acc_runs": [ + 0.2870312258348862 + ], + "mean_centered_acc_runs": [ + -0.023597556224523267 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-5184.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-5184.json new file mode 100644 index 0000000000000000000000000000000000000000..5693ad118526d35b842bef00fcb6e7dcf481cc50 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-5184.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-5184", + "checkpoint": "checkpoint-5184", + "global_step": 5184, + "tokens_seen": 5435817984, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25104560844453294, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25104560844453294 + ], + "acc": 0.24805815574586737, + "acc_ci": 0.0, + "acc_runs": [ + 0.24805815574586737 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004327285172595903, + "centered_acc": 0.0013941445927105889, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0013941445927105889 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.49782372143634385, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.49782372143634385 + ], + "acc": 0.5195865070729053, + "acc_ci": 0.0, + "acc_runs": [ + 0.5195865070729053 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665713661738719, + "centered_acc": -0.0043525571273123065, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0043525571273123065 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2609427609427609, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2609427609427609 + ], + "acc": 0.2588383838383838, + "acc_ci": 0.0, + "acc_runs": [ + 0.2588383838383838 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009011142493235922, + "centered_acc": 0.014590347923681227, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014590347923681227 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.257679180887372, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.257679180887372 + ], + "acc": 0.2175767918088737, + "acc_ci": 0.0, + "acc_runs": [ + 0.2175767918088737 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012780770562768334, + "centered_acc": 0.010238907849829356, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.010238907849829356 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.278, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.278 + ], + "acc": 0.172, + "acc_ci": 0.0, + "acc_runs": [ + 0.172 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020055833888070865, + "centered_acc": 0.037333333333333364, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037333333333333364 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5272296764009471, + "acc_ci": 0.0, + "acc_runs": [ + 0.5272296764009471 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01403163162982771, + "centered_acc": 0.05445935280189418, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05445935280189418 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 2223016723.1563606, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 2223016723.1563606 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.173, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.173 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011967214137559927, + "centered_acc": -0.10266666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10266666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.28641521195183495, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2917576976640678, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.014714245971681063, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.28641521195183495 + ], + "mean_acc_runs": [ + 0.2917576976640678 + ], + "mean_centered_acc_runs": [ + -0.014714245971681063 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-5472.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-5472.json new file mode 100644 index 0000000000000000000000000000000000000000..a35e592ece2917d91d70b9d9b5edc7719a0b2245 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-5472.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-5472", + "checkpoint": "checkpoint-5472", + "global_step": 5472, + "tokens_seen": 5737807872, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.24965146385182235, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24965146385182235 + ], + "acc": 0.2484564827723561, + "acc_ci": 0.0, + "acc_runs": [ + 0.2484564827723561 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00431926743246105, + "centered_acc": -0.000464714864236863, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.000464714864236863 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5087051142546246, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5087051142546246 + ], + "acc": 0.5163220892274211, + "acc_ci": 0.0, + "acc_runs": [ + 0.5163220892274211 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011664055982032772, + "centered_acc": 0.017410228509249226, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017410228509249226 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2647306397306397, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2647306397306397 + ], + "acc": 0.26304713804713803, + "acc_ci": 0.0, + "acc_runs": [ + 0.26304713804713803 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009053021086174166, + "centered_acc": 0.019640852974186274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019640852974186274 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25597269624573377, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25597269624573377 + ], + "acc": 0.21928327645051193, + "acc_ci": 0.0, + "acc_runs": [ + 0.21928327645051193 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012753013241244613, + "centered_acc": 0.007963594994311688, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007963594994311688 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.298, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.298 + ], + "acc": 0.17, + "acc_ci": 0.0, + "acc_runs": [ + 0.17 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020475118092988954, + "centered_acc": 0.06399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06399999999999999 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5367008681925809, + "acc_ci": 0.0, + "acc_runs": [ + 0.5367008681925809 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014014578458843366, + "centered_acc": 0.07340173638516179, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.07340173638516179 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 3581626111.034089, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 3581626111.034089 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.176, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.176 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012048616898597498, + "centered_acc": -0.09866666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09866666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29217665234713674, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2928097316463708, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0066822272356215386, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29217665234713674 + ], + "mean_acc_runs": [ + 0.2928097316463708 + ], + "mean_centered_acc_runs": [ + -0.0066822272356215386 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-576.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-576.json new file mode 100644 index 0000000000000000000000000000000000000000..933ff04049dc63f5c1f9d420029adefbc1ba6610 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-576.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-576", + "checkpoint": "checkpoint-576", + "global_step": 576, + "tokens_seen": 603979776, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25542720573590916, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25542720573590916 + ], + "acc": 0.2539334793865764, + "acc_ci": 0.0, + "acc_runs": [ + 0.2539334793865764 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004352098082984506, + "centered_acc": 0.007236274314545543, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007236274314545543 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5097932535364527, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5097932535364527 + ], + "acc": 0.530467899891186, + "acc_ci": 0.0, + "acc_runs": [ + 0.530467899891186 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011663586263283307, + "centered_acc": 0.019586507072905324, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019586507072905324 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26641414141414144, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26641414141414144 + ], + "acc": 0.25252525252525254, + "acc_ci": 0.0, + "acc_runs": [ + 0.25252525252525254 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009071357971078843, + "centered_acc": 0.021885521885521914, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021885521885521914 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25853242320819114, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25853242320819114 + ], + "acc": 0.21331058020477817, + "acc_ci": 0.0, + "acc_runs": [ + 0.21331058020477817 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012794553754288777, + "centered_acc": 0.011376564277588189, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011376564277588189 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.302, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.302 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02055326917420912, + "centered_acc": 0.06933333333333332, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06933333333333332 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5082872928176796, + "acc_ci": 0.0, + "acc_runs": [ + 0.5082872928176796 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405055532282417, + "centered_acc": 0.016574585635359185, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.016574585635359185 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.52, + "acc_ci": 0.0, + "acc_runs": [ + 0.52 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05021167315686783, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 1182988.0340544116, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 1182988.0340544116 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.176, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.176 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012048616898597498, + "centered_acc": -0.09866666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09866666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29469450398244906, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2857811966599173, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.017344328511112506, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29469450398244906 + ], + "mean_acc_runs": [ + 0.2857811966599173 + ], + "mean_centered_acc_runs": [ + -0.017344328511112506 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-5760.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-5760.json new file mode 100644 index 0000000000000000000000000000000000000000..697c538ccdde6e7592357b99a24101b22fe6690c --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-5760.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-5760", + "checkpoint": "checkpoint-5760", + "global_step": 5760, + "tokens_seen": 6039797760, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2548297151961761, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2548297151961761 + ], + "acc": 0.25104560844453294, + "acc_ci": 0.0, + "acc_runs": [ + 0.25104560844453294 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004348748730529728, + "centered_acc": 0.006439620261568117, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006439620261568117 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5027203482045702, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5027203482045702 + ], + "acc": 0.5233949945593036, + "acc_ci": 0.0, + "acc_runs": [ + 0.5233949945593036 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665651503000776, + "centered_acc": 0.005440696409140466, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005440696409140466 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2647306397306397, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2647306397306397 + ], + "acc": 0.26304713804713803, + "acc_ci": 0.0, + "acc_runs": [ + 0.26304713804713803 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009053021086174166, + "centered_acc": 0.019640852974186274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019640852974186274 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2508532423208191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2508532423208191 + ], + "acc": 0.21245733788395904, + "acc_ci": 0.0, + "acc_runs": [ + 0.21245733788395904 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012668198621315347, + "centered_acc": 0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.29, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.29 + ], + "acc": 0.17, + "acc_ci": 0.0, + "acc_runs": [ + 0.17 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020313179231745138, + "centered_acc": 0.05333333333333331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05333333333333331 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4877663772691397, + "acc_ci": 0.0, + "acc_runs": [ + 0.4877663772691397 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01404827882040573, + "centered_acc": -0.024467245461720633, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.024467245461720633 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 4085138069.983313, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 4085138069.983313 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.167, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.167 + ], + "acc": 0.05, + "acc_ci": 0.0, + "acc_runs": [ + 0.05 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011800434324644591, + "centered_acc": -0.11066666666666665, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.11066666666666665 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2883556575753676, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2886192979689398, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.02139631435277773, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2883556575753676 + ], + "mean_acc_runs": [ + 0.2886192979689398 + ], + "mean_centered_acc_runs": [ + -0.02139631435277773 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-6048.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-6048.json new file mode 100644 index 0000000000000000000000000000000000000000..e39d58c1efa8b2e0e1d9f94f2379f969a05990ff --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-6048.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-6048", + "checkpoint": "checkpoint-6048", + "global_step": 6048, + "tokens_seen": 6341787648, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25473013343955386, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25473013343955386 + ], + "acc": 0.24915355506871142, + "acc_ci": 0.0, + "acc_runs": [ + 0.24915355506871142 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043481894593367845, + "centered_acc": 0.006306844586071818, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006306844586071818 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5070729053318824, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5070729053318824 + ], + "acc": 0.5038084874863983, + "acc_ci": 0.0, + "acc_runs": [ + 0.5038084874863983 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011664656918146136, + "centered_acc": 0.014145810663764857, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014145810663764857 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2680976430976431, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2680976430976431 + ], + "acc": 0.2617845117845118, + "acc_ci": 0.0, + "acc_runs": [ + 0.2617845117845118 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009089526578213825, + "centered_acc": 0.024130190796857482, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.024130190796857482 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25170648464163825, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25170648464163825 + ], + "acc": 0.21331058020477817, + "acc_ci": 0.0, + "acc_runs": [ + 0.21331058020477817 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012682496334042911, + "centered_acc": 0.0022753128555176674, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0022753128555176674 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.304, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.304 + ], + "acc": 0.184, + "acc_ci": 0.0, + "acc_runs": [ + 0.184 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02059164957122494, + "centered_acc": 0.072, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.072 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5082872928176796, + "acc_ci": 0.0, + "acc_runs": [ + 0.5082872928176796 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405055532282417, + "centered_acc": 0.016574585635359185, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.016574585635359185 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 5348042468.434231, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 5348042468.434231 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.176, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.176 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012048616898597498, + "centered_acc": -0.09866666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09866666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29360119441845295, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28956318891357796, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.01185099984241055, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29360119441845295 + ], + "mean_acc_runs": [ + 0.28956318891357796 + ], + "mean_centered_acc_runs": [ + -0.01185099984241055 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-6336.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-6336.json new file mode 100644 index 0000000000000000000000000000000000000000..1bda8a3780baf1ee813e3471b78d919304a5a67b --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-6336.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-6336", + "checkpoint": "checkpoint-6336", + "global_step": 6336, + "tokens_seen": 6643777536, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25672176857199763, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25672176857199763 + ], + "acc": 0.24975104560844452, + "acc_ci": 0.0, + "acc_runs": [ + 0.24975104560844452 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043593182064291, + "centered_acc": 0.008962358095996844, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008962358095996844 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5048966267682263, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5048966267682263 + ], + "acc": 0.5239390642002176, + "acc_ci": 0.0, + "acc_runs": [ + 0.5239390642002176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665264730078137, + "centered_acc": 0.009793253536452662, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009793253536452662 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26430976430976433, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26430976430976433 + ], + "acc": 0.26641414141414144, + "acc_ci": 0.0, + "acc_runs": [ + 0.26641414141414144 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009048410451862887, + "centered_acc": 0.019079685746352437, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019079685746352437 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2508532423208191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2508532423208191 + ], + "acc": 0.2158703071672355, + "acc_ci": 0.0, + "acc_runs": [ + 0.2158703071672355 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012668198621315347, + "centered_acc": 0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.29, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.29 + ], + "acc": 0.184, + "acc_ci": 0.0, + "acc_runs": [ + 0.184 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020313179231745138, + "centered_acc": 0.05333333333333331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05333333333333331 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.489344909234412, + "acc_ci": 0.0, + "acc_runs": [ + 0.489344909234412 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014049294536290372, + "centered_acc": -0.02131018153117603, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.02131018153117603 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48, + "acc_ci": 0.0, + "acc_runs": [ + 0.48 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05021167315686783, + "centered_acc": -0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.040000000000000036 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 6327859426.910668, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 6327859426.910668 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.174, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.174 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011994493230973449, + "centered_acc": -0.10133333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10133333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29013023366180124, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2834606929398152, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.03486247824191274, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29013023366180124 + ], + "mean_acc_runs": [ + 0.2834606929398152 + ], + "mean_centered_acc_runs": [ + -0.03486247824191274 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-6624.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-6624.json new file mode 100644 index 0000000000000000000000000000000000000000..34ecafc51b7f8a3816e0ffee176596ccd3d215f0 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-6624.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-6624", + "checkpoint": "checkpoint-6624", + "global_step": 6624, + "tokens_seen": 6945767424, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25273849830711015, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25273849830711015 + ], + "acc": 0.25054769966142204, + "acc_ci": 0.0, + "acc_runs": [ + 0.25054769966142204 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043369410695685015, + "centered_acc": 0.003651331076146865, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.003651331076146865 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.514145810663765, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.514145810663765 + ], + "acc": 0.5163220892274211, + "acc_ci": 0.0, + "acc_runs": [ + 0.5163220892274211 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011661154475524689, + "centered_acc": 0.028291621327529937, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.028291621327529937 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2706228956228956, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2706228956228956 + ], + "acc": 0.2680976430976431, + "acc_ci": 0.0, + "acc_runs": [ + 0.2680976430976431 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009116466166404035, + "centered_acc": 0.0274971941638608, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0274971941638608 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25853242320819114, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25853242320819114 + ], + "acc": 0.22013651877133106, + "acc_ci": 0.0, + "acc_runs": [ + 0.22013651877133106 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012794553754288777, + "centered_acc": 0.011376564277588189, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011376564277588189 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.174, + "acc_ci": 0.0, + "acc_runs": [ + 0.174 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5082872928176796, + "acc_ci": 0.0, + "acc_runs": [ + 0.5082872928176796 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405055532282417, + "centered_acc": 0.016574585635359185, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.016574585635359185 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.52, + "acc_ci": 0.0, + "acc_runs": [ + 0.52 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05021167315686783, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 10927725204.81932, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 10927725204.81932 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.174, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.174 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011994493230973449, + "centered_acc": -0.10133333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10133333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.291673271300327, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2882872767060822, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.019707457033938627, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.291673271300327 + ], + "mean_acc_runs": [ + 0.2882872767060822 + ], + "mean_centered_acc_runs": [ + -0.019707457033938627 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-6912.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-6912.json new file mode 100644 index 0000000000000000000000000000000000000000..18ade8497870fcf3e7a40e0ed60b5e81e8fa8c6c --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-6912.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-6912", + "checkpoint": "checkpoint-6912", + "global_step": 6912, + "tokens_seen": 7247757312, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25273849830711015, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25273849830711015 + ], + "acc": 0.24945230033857796, + "acc_ci": 0.0, + "acc_runs": [ + 0.24945230033857796 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043369410695685015, + "centered_acc": 0.003651331076146865, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.003651331076146865 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5163220892274211, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5163220892274211 + ], + "acc": 0.515233949945593, + "acc_ci": 0.0, + "acc_runs": [ + 0.515233949945593 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011659606710151597, + "centered_acc": 0.03264417845484213, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03264417845484213 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2601010101010101, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2601010101010101 + ], + "acc": 0.26304713804713803, + "acc_ci": 0.0, + "acc_runs": [ + 0.26304713804713803 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009001718541079735, + "centered_acc": 0.013468013468013481, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013468013468013481 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25597269624573377, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25597269624573377 + ], + "acc": 0.21928327645051193, + "acc_ci": 0.0, + "acc_runs": [ + 0.21928327645051193 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012753013241244613, + "centered_acc": 0.007963594994311688, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007963594994311688 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.176, + "acc_ci": 0.0, + "acc_runs": [ + 0.176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5011838989739542, + "acc_ci": 0.0, + "acc_runs": [ + 0.5011838989739542 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052446290528871, + "centered_acc": 0.0023677979479084232, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0023677979479084232 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.51, + "acc_ci": 0.0, + "acc_runs": [ + 0.51 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05024183937956913, + "centered_acc": 0.020000000000000018, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020000000000000018 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 12196371440.19413, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 12196371440.19413 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.174, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.174 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011994493230973449, + "centered_acc": -0.10133333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10133333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2898557156468792, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28584880255294753, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.02496261042719001, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2898557156468792 + ], + "mean_acc_runs": [ + 0.28584880255294753 + ], + "mean_centered_acc_runs": [ + -0.02496261042719001 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-72.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-72.json new file mode 100644 index 0000000000000000000000000000000000000000..fb2afeeec95ff72401f24334ee6641ef25b5b600 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-72.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-72", + "checkpoint": "checkpoint-72", + "global_step": 72, + "tokens_seen": 75497472, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2608046205935073, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2608046205935073 + ], + "acc": 0.25692093208524197, + "acc_ci": 0.0, + "acc_runs": [ + 0.25692093208524197 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00438176194155244, + "centered_acc": 0.014406160791343048, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014406160791343048 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5103373231773667, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5103373231773667 + ], + "acc": 0.5190424374319913, + "acc_ci": 0.0, + "acc_runs": [ + 0.5190424374319913 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01166333067307569, + "centered_acc": 0.020674646354733373, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020674646354733373 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2542087542087542, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2542087542087542 + ], + "acc": 0.24368686868686867, + "acc_ci": 0.0, + "acc_runs": [ + 0.24368686868686867 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008934537681141764, + "centered_acc": 0.005611672278338957, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005611672278338957 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.257679180887372, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.257679180887372 + ], + "acc": 0.20648464163822525, + "acc_ci": 0.0, + "acc_runs": [ + 0.20648464163822525 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012780770562768334, + "centered_acc": 0.010238907849829356, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.010238907849829356 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.278, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.278 + ], + "acc": 0.148, + "acc_ci": 0.0, + "acc_runs": [ + 0.148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020055833888070865, + "centered_acc": 0.037333333333333364, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037333333333333364 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.49171270718232046, + "acc_ci": 0.0, + "acc_runs": [ + 0.49171270718232046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405055532282417, + "centered_acc": -0.016574585635359074, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.016574585635359074 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 17089.967541159473, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 17089.967541159473 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.173, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.173 + ], + "acc": 0.04, + "acc_ci": 0.0, + "acc_runs": [ + 0.04 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011967214137559927, + "centered_acc": -0.10266666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10266666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.28900497981116674, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28241350487983474, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.02160017868300522, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.28900497981116674 + ], + "mean_acc_runs": [ + 0.28241350487983474 + ], + "mean_centered_acc_runs": [ + -0.02160017868300522 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-7200.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-7200.json new file mode 100644 index 0000000000000000000000000000000000000000..cc2b8b46428cbb2d8b17d535cbbaad1a92428823 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-7200.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-7200", + "checkpoint": "checkpoint-7200", + "global_step": 7200, + "tokens_seen": 7549747200, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2531368253335989, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2531368253335989 + ], + "acc": 0.24965146385182235, + "acc_ci": 0.0, + "acc_runs": [ + 0.24965146385182235 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004339200363454809, + "centered_acc": 0.00418243377813184, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.00418243377813184 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5136017410228509, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5136017410228509 + ], + "acc": 0.5212187159956474, + "acc_ci": 0.0, + "acc_runs": [ + 0.5212187159956474 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011661506839823792, + "centered_acc": 0.027203482045701888, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.027203482045701888 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2680976430976431, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2680976430976431 + ], + "acc": 0.2727272727272727, + "acc_ci": 0.0, + "acc_runs": [ + 0.2727272727272727 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009089526578213825, + "centered_acc": 0.024130190796857482, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.024130190796857482 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2636518771331058, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2636518771331058 + ], + "acc": 0.20563139931740615, + "acc_ci": 0.0, + "acc_runs": [ + 0.20563139931740615 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012875929151296968, + "centered_acc": 0.018202502844141044, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018202502844141044 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.174, + "acc_ci": 0.0, + "acc_runs": [ + 0.174 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4980268350434096, + "acc_ci": 0.0, + "acc_runs": [ + 0.4980268350434096 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052376259225643, + "centered_acc": -0.003946329913180779, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.003946329913180779 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 19251662182.07262, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 19251662182.07262 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.17, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.17 + ], + "acc": 0.04, + "acc_ci": 0.0, + "acc_runs": [ + 0.04 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011884495834541793, + "centered_acc": -0.10666666666666665, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10666666666666665 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29241468109786645, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2879543148709258, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.016924384840846057, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29241468109786645 + ], + "mean_acc_runs": [ + 0.2879543148709258 + ], + "mean_centered_acc_runs": [ + -0.016924384840846057 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-7488.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-7488.json new file mode 100644 index 0000000000000000000000000000000000000000..0bda57b27d2363cee33ad77b522ec72a525688f8 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-7488.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-7488", + "checkpoint": "checkpoint-7488", + "global_step": 7488, + "tokens_seen": 7851737088, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.252141007767377, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.252141007767377 + ], + "acc": 0.2526389165504879, + "acc_ci": 0.0, + "acc_runs": [ + 0.2526389165504879 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004333543083293202, + "centered_acc": 0.0028546770231693643, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0028546770231693643 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5136017410228509, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5136017410228509 + ], + "acc": 0.528835690968444, + "acc_ci": 0.0, + "acc_runs": [ + 0.528835690968444 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011661506839823792, + "centered_acc": 0.027203482045701888, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.027203482045701888 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.25925925925925924, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25925925925925924 + ], + "acc": 0.26262626262626265, + "acc_ci": 0.0, + "acc_runs": [ + 0.26262626262626265 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008992251535805688, + "centered_acc": 0.01234567901234566, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01234567901234566 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2568259385665529, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2568259385665529 + ], + "acc": 0.22098976109215018, + "acc_ci": 0.0, + "acc_runs": [ + 0.22098976109215018 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012766923794116735, + "centered_acc": 0.009101251422070522, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009101251422070522 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.288, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.288 + ], + "acc": 0.184, + "acc_ci": 0.0, + "acc_runs": [ + 0.184 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020271503835075255, + "centered_acc": 0.05066666666666664, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05066666666666664 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5185477505919495, + "acc_ci": 0.0, + "acc_runs": [ + 0.5185477505919495 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014042813708888347, + "centered_acc": 0.03709550118389893, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03709550118389893 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.53, + "acc_ci": 0.0, + "acc_runs": [ + 0.53 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05016135580465919, + "centered_acc": 0.06000000000000005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06000000000000005 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 25041298117.38943, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 25041298117.38943 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.175, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.175 + ], + "acc": 0.039, + "acc_ci": 0.0, + "acc_runs": [ + 0.039 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012021627157731998, + "centered_acc": -0.10000000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10000000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29080465776934, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29149258436029946, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.01601753545541626, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29080465776934 + ], + "mean_acc_runs": [ + 0.29149258436029946 + ], + "mean_centered_acc_runs": [ + -0.01601753545541626 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-7776.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-7776.json new file mode 100644 index 0000000000000000000000000000000000000000..d2d6d0bd1f9f67eb7222593a88dbe7697b3f6693 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-7776.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-7776", + "checkpoint": "checkpoint-7776", + "global_step": 7776, + "tokens_seen": 8153726976, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.253734315873332, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.253734315873332 + ], + "acc": 0.25253933479386576, + "acc_ci": 0.0, + "acc_runs": [ + 0.25253933479386576 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004342580277662561, + "centered_acc": 0.004979087831109341, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.004979087831109341 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5076169749727966, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5076169749727966 + ], + "acc": 0.5228509249183896, + "acc_ci": 0.0, + "acc_runs": [ + 0.5228509249183896 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011664470424044771, + "centered_acc": 0.015233949945593128, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015233949945593128 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2622053872053872, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2622053872053872 + ], + "acc": 0.2638888888888889, + "acc_ci": 0.0, + "acc_runs": [ + 0.2638888888888889 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009025197991724937, + "centered_acc": 0.01627384960718296, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01627384960718296 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2508532423208191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2508532423208191 + ], + "acc": 0.20136518771331058, + "acc_ci": 0.0, + "acc_runs": [ + 0.20136518771331058 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012668198621315347, + "centered_acc": 0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.176, + "acc_ci": 0.0, + "acc_runs": [ + 0.176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5067087608524072, + "acc_ci": 0.0, + "acc_runs": [ + 0.5067087608524072 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051220692330247, + "centered_acc": 0.013417521704814472, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013417521704814472 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.52, + "acc_ci": 0.0, + "acc_runs": [ + 0.52 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05021167315686783, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 26042077270.405506, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 26042077270.405506 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.172, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.172 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011939788882495308, + "centered_acc": -0.10400000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10400000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.28806832006205585, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28656405589405626, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.023746260474386, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.28806832006205585 + ], + "mean_acc_runs": [ + 0.28656405589405626 + ], + "mean_centered_acc_runs": [ + -0.023746260474386 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-8064.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-8064.json new file mode 100644 index 0000000000000000000000000000000000000000..7bc85d601c4ebbc932923d263e9b1296f2629a6f --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-8064.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-8064", + "checkpoint": "checkpoint-8064", + "global_step": 8064, + "tokens_seen": 8455716864, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25124477195777734, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25124477195777734 + ], + "acc": 0.24955188209520016, + "acc_ci": 0.0, + "acc_runs": [ + 0.24955188209520016 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004328425700998576, + "centered_acc": 0.0016596959437031138, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0016596959437031138 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5108813928182807, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5108813928182807 + ], + "acc": 0.5212187159956474, + "acc_ci": 0.0, + "acc_runs": [ + 0.5212187159956474 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011663061261117717, + "centered_acc": 0.02176278563656142, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02176278563656142 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2617845117845118, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2617845117845118 + ], + "acc": 0.27146464646464646, + "acc_ci": 0.0, + "acc_runs": [ + 0.27146464646464646 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009020523527210056, + "centered_acc": 0.015712682379349047, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015712682379349047 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2568259385665529, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2568259385665529 + ], + "acc": 0.22013651877133106, + "acc_ci": 0.0, + "acc_runs": [ + 0.22013651877133106 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012766923794116735, + "centered_acc": 0.009101251422070522, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009101251422070522 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.182, + "acc_ci": 0.0, + "acc_runs": [ + 0.182 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4972375690607735, + "acc_ci": 0.0, + "acc_runs": [ + 0.4972375690607735 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052271211616401, + "centered_acc": -0.0055248618784530246, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0055248618784530246 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.52, + "acc_ci": 0.0, + "acc_runs": [ + 0.52 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05021167315686783, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 34040228153.797745, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 34040228153.797745 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.169, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.169 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011856625977890152, + "centered_acc": -0.10799999999999998, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10799999999999998 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2886227691878538, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28828967941612993, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.025116317364744623, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2886227691878538 + ], + "mean_acc_runs": [ + 0.28828967941612993 + ], + "mean_centered_acc_runs": [ + -0.025116317364744623 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-8352.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-8352.json new file mode 100644 index 0000000000000000000000000000000000000000..36d91f18f6d4daef202435390203a41ad6abfe6a --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-8352.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-8352", + "checkpoint": "checkpoint-8352", + "global_step": 8352, + "tokens_seen": 8757706752, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2523401712806214, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2523401712806214 + ], + "acc": 0.24885480979884486, + "acc_ci": 0.0, + "acc_runs": [ + 0.24885480979884486 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004334676952703991, + "centered_acc": 0.0031202283741618895, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0031202283741618895 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.514689880304679, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.514689880304679 + ], + "acc": 0.5195865070729053, + "acc_ci": 0.0, + "acc_runs": [ + 0.5195865070729053 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011660788281735552, + "centered_acc": 0.029379760609357986, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.029379760609357986 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26725589225589225, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26725589225589225 + ], + "acc": 0.2638888888888889, + "acc_ci": 0.0, + "acc_runs": [ + 0.2638888888888889 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009080463246017426, + "centered_acc": 0.023007856341189663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023007856341189663 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.257679180887372, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.257679180887372 + ], + "acc": 0.21928327645051193, + "acc_ci": 0.0, + "acc_runs": [ + 0.21928327645051193 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012780770562768334, + "centered_acc": 0.010238907849829356, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.010238907849829356 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.18, + "acc_ci": 0.0, + "acc_runs": [ + 0.18 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5035516969218626, + "acc_ci": 0.0, + "acc_runs": [ + 0.5035516969218626 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052131146915852, + "centered_acc": 0.00710339384372527, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.00710339384372527 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.53, + "acc_ci": 0.0, + "acc_runs": [ + 0.53 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05016135580465919, + "centered_acc": 0.06000000000000005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06000000000000005 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 49483206131.061844, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 49483206131.061844 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.177, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.177 + ], + "acc": 0.049, + "acc_ci": 0.0, + "acc_runs": [ + 0.049 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012075463420375087, + "centered_acc": -0.09733333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09733333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2914941874547608, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2892452640906714, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.018656473640852052, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2914941874547608 + ], + "mean_acc_runs": [ + 0.2892452640906714 + ], + "mean_centered_acc_runs": [ + -0.018656473640852052 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-864.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-864.json new file mode 100644 index 0000000000000000000000000000000000000000..667fc44dc7a4ba472c2d3f62f204aba0446bc5fb --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-864.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-864", + "checkpoint": "checkpoint-864", + "global_step": 864, + "tokens_seen": 905969664, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25184226249751046, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25184226249751046 + ], + "acc": 0.25423222465644296, + "acc_ci": 0.0, + "acc_runs": [ + 0.25423222465644296 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004331840012787794, + "centered_acc": 0.0024563499966806144, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0024563499966806144 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5048966267682263, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5048966267682263 + ], + "acc": 0.5228509249183896, + "acc_ci": 0.0, + "acc_runs": [ + 0.5228509249183896 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665264730078137, + "centered_acc": 0.009793253536452662, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009793253536452662 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2718855218855219, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2718855218855219 + ], + "acc": 0.2680976430976431, + "acc_ci": 0.0, + "acc_runs": [ + 0.2680976430976431 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009129795867310605, + "centered_acc": 0.02918069584736253, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02918069584736253 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2636518771331058, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2636518771331058 + ], + "acc": 0.19965870307167236, + "acc_ci": 0.0, + "acc_runs": [ + 0.19965870307167236 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012875929151296968, + "centered_acc": 0.018202502844141044, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018202502844141044 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.3, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.3 + ], + "acc": 0.156, + "acc_ci": 0.0, + "acc_runs": [ + 0.156 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020514426225627987, + "centered_acc": 0.06666666666666665, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06666666666666665 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4972375690607735, + "acc_ci": 0.0, + "acc_runs": [ + 0.4972375690607735 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052271211616401, + "centered_acc": -0.0055248618784530246, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0055248618784530246 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.51, + "acc_ci": 0.0, + "acc_runs": [ + 0.51 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05024183937956913, + "centered_acc": 0.020000000000000018, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020000000000000018 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 3174241.2432426736, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 3174241.2432426736 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.175, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.175 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012021627157731998, + "centered_acc": -0.10000000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10000000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2945460480473941, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28333645265786217, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.022516718826638766, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2945460480473941 + ], + "mean_acc_runs": [ + 0.28333645265786217 + ], + "mean_centered_acc_runs": [ + -0.022516718826638766 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/checkpoint-8640.json b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-8640.json new file mode 100644 index 0000000000000000000000000000000000000000..63283d588b725643c270271e1756ba84f78f2b96 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/checkpoint-8640.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", + "run": "baseline-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-8640", + "checkpoint": "checkpoint-8640", + "global_step": 8640, + "tokens_seen": 9059696640, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25184226249751046, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25184226249751046 + ], + "acc": 0.2513443537143995, + "acc_ci": 0.0, + "acc_runs": [ + 0.2513443537143995 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004331840012787794, + "centered_acc": 0.0024563499966806144, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0024563499966806144 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5070729053318824, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5070729053318824 + ], + "acc": 0.5184983677910773, + "acc_ci": 0.0, + "acc_runs": [ + 0.5184983677910773 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011664656918146136, + "centered_acc": 0.014145810663764857, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014145810663764857 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26262626262626265, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26262626262626265 + ], + "acc": 0.26052188552188554, + "acc_ci": 0.0, + "acc_runs": [ + 0.26052188552188554 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009029861776763957, + "centered_acc": 0.01683501683501687, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01683501683501687 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24914675767918087, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24914675767918087 + ], + "acc": 0.21160409556313994, + "acc_ci": 0.0, + "acc_runs": [ + 0.21160409556313994 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012639407111926505, + "centered_acc": -0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.264, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.264 + ], + "acc": 0.168, + "acc_ci": 0.0, + "acc_runs": [ + 0.168 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01973288558592215, + "centered_acc": 0.018666666666666682, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018666666666666682 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5130228887134964, + "acc_ci": 0.0, + "acc_runs": [ + 0.5130228887134964 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014047718393997596, + "centered_acc": 0.026045777426992878, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.026045777426992878 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.52, + "acc_ci": 0.0, + "acc_runs": [ + 0.52 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05021167315686783, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 66809449869.6633, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 66809449869.6633 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.172, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.172 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011939788882495308, + "centered_acc": -0.10400000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10400000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.28444803135580604, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2867279053077699, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.02560145681013737, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.28444803135580604 + ], + "mean_acc_runs": [ + 0.2867279053077699 + ], + "mean_centered_acc_runs": [ + -0.02560145681013737 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/baseline-fulle-lm/zero_shot_eval/summary.jsonl b/results/baseline-fulle-lm/zero_shot_eval/summary.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..90012ff45b41a49d4dc0bb4e58f0a02fadcaec43 --- /dev/null +++ b/results/baseline-fulle-lm/zero_shot_eval/summary.jsonl @@ -0,0 +1,31 @@ +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-72", "checkpoint": "checkpoint-72", "global_step": 72, "tokens_seen": 75497472, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2608046205935073, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2608046205935073], "acc": 0.25692093208524197, "acc_ci": 0.0, "acc_runs": [0.25692093208524197], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00438176194155244, "centered_acc": 0.014406160791343048, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014406160791343048], "ci": 0.0}, "piqa": {"acc_norm": 0.5103373231773667, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5103373231773667], "acc": 0.5190424374319913, "acc_ci": 0.0, "acc_runs": [0.5190424374319913], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01166333067307569, "centered_acc": 0.020674646354733373, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020674646354733373], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2542087542087542, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2542087542087542], "acc": 0.24368686868686867, "acc_ci": 0.0, "acc_runs": [0.24368686868686867], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008934537681141764, "centered_acc": 0.005611672278338957, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005611672278338957], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.257679180887372, "acc_norm_ci": 0.0, "acc_norm_runs": [0.257679180887372], "acc": 0.20648464163822525, "acc_ci": 0.0, "acc_runs": [0.20648464163822525], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012780770562768334, "centered_acc": 0.010238907849829356, "centered_acc_ci": 0.0, "centered_acc_runs": [0.010238907849829356], "ci": 0.0}, "openbookqa": {"acc_norm": 0.278, "acc_norm_ci": 0.0, "acc_norm_runs": [0.278], "acc": 0.148, "acc_ci": 0.0, "acc_runs": [0.148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020055833888070865, "centered_acc": 0.037333333333333364, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037333333333333364], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.49171270718232046, "acc_ci": 0.0, "acc_runs": [0.49171270718232046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405055532282417, "centered_acc": -0.016574585635359074, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.016574585635359074], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 17089.967541159473, "perplexity_ci": 0.0, "perplexity_runs": [17089.967541159473], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.173, "acc_norm_ci": 0.0, "acc_norm_runs": [0.173], "acc": 0.04, "acc_ci": 0.0, "acc_runs": [0.04], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011967214137559927, "centered_acc": -0.10266666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10266666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.28900497981116674, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28241350487983474, "mean_acc_ci": 0.0, "mean_centered_acc": -0.02160017868300522, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.28900497981116674], "mean_acc_runs": [0.28241350487983474], "mean_centered_acc_runs": [-0.02160017868300522], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-288", "checkpoint": "checkpoint-288", "global_step": 288, "tokens_seen": 301989888, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25562636924915355, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25562636924915355], "acc": 0.2548297151961761, "acc_ci": 0.0, "acc_runs": [0.2548297151961761], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004353212146198103, "centered_acc": 0.007501825665538068, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007501825665538068], "ci": 0.0}, "piqa": {"acc_norm": 0.5021762785636561, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5021762785636561], "acc": 0.5239390642002176, "acc_ci": 0.0, "acc_runs": [0.5239390642002176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665713661738719, "centered_acc": 0.0043525571273121955, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0043525571273121955], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2558922558922559, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2558922558922559], "acc": 0.24831649831649832, "acc_ci": 0.0, "acc_runs": [0.24831649831649832], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008953950243013958, "centered_acc": 0.007856341189674524, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007856341189674524], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2627986348122867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2627986348122867], "acc": 0.2030716723549488, "acc_ci": 0.0, "acc_runs": [0.2030716723549488], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012862523175351397, "centered_acc": 0.017064846416382284, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017064846416382284], "ci": 0.0}, "openbookqa": {"acc_norm": 0.284, "acc_norm_ci": 0.0, "acc_norm_runs": [0.284], "acc": 0.162, "acc_ci": 0.0, "acc_runs": [0.162], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020186703693570774, "centered_acc": 0.0453333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0453333333333333], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5201262825572218, "acc_ci": 0.0, "acc_runs": [0.5201262825572218], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014041096664344348, "centered_acc": 0.04025256511444364, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04025256511444364], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.6, "acc_ci": 0.0, "acc_runs": [0.6], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0492365963917331, "centered_acc": 0.19999999999999996, "centered_acc_ci": 0.0, "centered_acc_runs": [0.19999999999999996], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 335678.3049185522, "perplexity_ci": 0.0, "perplexity_runs": [335678.3049185522], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.173, "acc_norm_ci": 0.0, "acc_norm_runs": [0.173], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011967214137559927, "centered_acc": -0.10266666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10266666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.28891558975289205, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29355706943987625, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0026366971413980087, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.28891558975289205], "mean_acc_runs": [0.29355706943987625], "mean_centered_acc_runs": [-0.0026366971413980087], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-576", "checkpoint": "checkpoint-576", "global_step": 576, "tokens_seen": 603979776, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25542720573590916, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25542720573590916], "acc": 0.2539334793865764, "acc_ci": 0.0, "acc_runs": [0.2539334793865764], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004352098082984506, "centered_acc": 0.007236274314545543, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007236274314545543], "ci": 0.0}, "piqa": {"acc_norm": 0.5097932535364527, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5097932535364527], "acc": 0.530467899891186, "acc_ci": 0.0, "acc_runs": [0.530467899891186], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011663586263283307, "centered_acc": 0.019586507072905324, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019586507072905324], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26641414141414144, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26641414141414144], "acc": 0.25252525252525254, "acc_ci": 0.0, "acc_runs": [0.25252525252525254], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009071357971078843, "centered_acc": 0.021885521885521914, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021885521885521914], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25853242320819114, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25853242320819114], "acc": 0.21331058020477817, "acc_ci": 0.0, "acc_runs": [0.21331058020477817], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012794553754288777, "centered_acc": 0.011376564277588189, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011376564277588189], "ci": 0.0}, "openbookqa": {"acc_norm": 0.302, "acc_norm_ci": 0.0, "acc_norm_runs": [0.302], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02055326917420912, "centered_acc": 0.06933333333333332, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06933333333333332], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5082872928176796, "acc_ci": 0.0, "acc_runs": [0.5082872928176796], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405055532282417, "centered_acc": 0.016574585635359185, "centered_acc_ci": 0.0, "centered_acc_runs": [0.016574585635359185], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.52, "acc_ci": 0.0, "acc_runs": [0.52], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05021167315686783, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 1182988.0340544116, "perplexity_ci": 0.0, "perplexity_runs": [1182988.0340544116], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.176, "acc_norm_ci": 0.0, "acc_norm_runs": [0.176], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012048616898597498, "centered_acc": -0.09866666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09866666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29469450398244906, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2857811966599173, "mean_acc_ci": 0.0, "mean_centered_acc": -0.017344328511112506, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29469450398244906], "mean_acc_runs": [0.2857811966599173], "mean_centered_acc_runs": [-0.017344328511112506], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-864", "checkpoint": "checkpoint-864", "global_step": 864, "tokens_seen": 905969664, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25184226249751046, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25184226249751046], "acc": 0.25423222465644296, "acc_ci": 0.0, "acc_runs": [0.25423222465644296], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004331840012787794, "centered_acc": 0.0024563499966806144, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0024563499966806144], "ci": 0.0}, "piqa": {"acc_norm": 0.5048966267682263, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5048966267682263], "acc": 0.5228509249183896, "acc_ci": 0.0, "acc_runs": [0.5228509249183896], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665264730078137, "centered_acc": 0.009793253536452662, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009793253536452662], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2718855218855219, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2718855218855219], "acc": 0.2680976430976431, "acc_ci": 0.0, "acc_runs": [0.2680976430976431], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009129795867310605, "centered_acc": 0.02918069584736253, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02918069584736253], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2636518771331058, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2636518771331058], "acc": 0.19965870307167236, "acc_ci": 0.0, "acc_runs": [0.19965870307167236], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012875929151296968, "centered_acc": 0.018202502844141044, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018202502844141044], "ci": 0.0}, "openbookqa": {"acc_norm": 0.3, "acc_norm_ci": 0.0, "acc_norm_runs": [0.3], "acc": 0.156, "acc_ci": 0.0, "acc_runs": [0.156], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020514426225627987, "centered_acc": 0.06666666666666665, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06666666666666665], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4972375690607735, "acc_ci": 0.0, "acc_runs": [0.4972375690607735], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052271211616401, "centered_acc": -0.0055248618784530246, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0055248618784530246], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.51, "acc_ci": 0.0, "acc_runs": [0.51], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05024183937956913, "centered_acc": 0.020000000000000018, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020000000000000018], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 3174241.2432426736, "perplexity_ci": 0.0, "perplexity_runs": [3174241.2432426736], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.175, "acc_norm_ci": 0.0, "acc_norm_runs": [0.175], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012021627157731998, "centered_acc": -0.10000000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10000000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2945460480473941, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28333645265786217, "mean_acc_ci": 0.0, "mean_centered_acc": -0.022516718826638766, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2945460480473941], "mean_acc_runs": [0.28333645265786217], "mean_centered_acc_runs": [-0.022516718826638766], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-1152", "checkpoint": "checkpoint-1152", "global_step": 1152, "tokens_seen": 1207959552, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25144393547102173, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25144393547102173], "acc": 0.2545309699263095, "acc_ci": 0.0, "acc_runs": [0.2545309699263095], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043295650165269565, "centered_acc": 0.0019252472946956385, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0019252472946956385], "ci": 0.0}, "piqa": {"acc_norm": 0.5168661588683352, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5168661588683352], "acc": 0.5310119695321001, "acc_ci": 0.0, "acc_runs": [0.5310119695321001], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011659185184878962, "centered_acc": 0.0337323177366704, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0337323177366704], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2647306397306397, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2647306397306397], "acc": 0.26641414141414144, "acc_ci": 0.0, "acc_runs": [0.26641414141414144], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009053021086174166, "centered_acc": 0.019640852974186274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019640852974186274], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25], "acc": 0.197098976109215, "acc_ci": 0.0, "acc_runs": [0.197098976109215], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012653835621466646, "centered_acc": 0.0, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0], "ci": 0.0}, "openbookqa": {"acc_norm": 0.276, "acc_norm_ci": 0.0, "acc_norm_runs": [0.276], "acc": 0.166, "acc_ci": 0.0, "acc_runs": [0.166], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02001121929807355, "centered_acc": 0.0346666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0346666666666667], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5059194948697711, "acc_ci": 0.0, "acc_runs": [0.5059194948697711], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051500838485923, "centered_acc": 0.011838989739542116, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011838989739542116], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.46, "acc_ci": 0.0, "acc_runs": [0.46], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620332, "centered_acc": -0.07999999999999996, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07999999999999996], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 5518097.083106795, "perplexity_ci": 0.0, "perplexity_runs": [5518097.083106795], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.18, "acc_norm_ci": 0.0, "acc_norm_runs": [0.18], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012155153135512024, "centered_acc": -0.09333333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09333333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2898401223449994, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28062630136252376, "mean_acc_ci": 0.0, "mean_centered_acc": -0.03499492615268573, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2898401223449994], "mean_acc_runs": [0.28062630136252376], "mean_centered_acc_runs": [-0.03499492615268573], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-1440", "checkpoint": "checkpoint-1440", "global_step": 1440, "tokens_seen": 1509949440, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2533359888468433, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2533359888468433], "acc": 0.2531368253335989, "acc_ci": 0.0, "acc_runs": [0.2531368253335989], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004340328204135388, "centered_acc": 0.0044479851291243655, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0044479851291243655], "ci": 0.0}, "piqa": {"acc_norm": 0.5114254624591947, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5114254624591947], "acc": 0.5342763873775843, "acc_ci": 0.0, "acc_runs": [0.5342763873775843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011662778026451529, "centered_acc": 0.02285092491838947, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02285092491838947], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2702020202020202, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2702020202020202], "acc": 0.26052188552188554, "acc_ci": 0.0, "acc_runs": [0.26052188552188554], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009112002229119622, "centered_acc": 0.026936026936026963, "centered_acc_ci": 0.0, "centered_acc_runs": [0.026936026936026963], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2645051194539249, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2645051194539249], "acc": 0.2150170648464164, "acc_ci": 0.0, "acc_runs": [0.2150170648464164], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012889272949313479, "centered_acc": 0.019340159271899877, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019340159271899877], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.168, "acc_ci": 0.0, "acc_runs": [0.168], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5082872928176796, "acc_ci": 0.0, "acc_runs": [0.5082872928176796], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405055532282417, "centered_acc": 0.016574585635359185, "centered_acc_ci": 0.0, "centered_acc_runs": [0.016574585635359185], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.52, "acc_ci": 0.0, "acc_runs": [0.52], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05021167315686783, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 8775376.900286412, "perplexity_ci": 0.0, "perplexity_runs": [8775376.900286412], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.179, "acc_norm_ci": 0.0, "acc_norm_runs": [0.179], "acc": 0.049, "acc_ci": 0.0, "acc_runs": [0.049], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012128730605719054, "centered_acc": -0.09466666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09466666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2934114318269972, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2886526917670865, "mean_acc_ci": 0.0, "mean_centered_acc": -0.018363932729088834, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2934114318269972], "mean_acc_runs": [0.2886526917670865], "mean_centered_acc_runs": [-0.018363932729088834], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-1728", "checkpoint": "checkpoint-1728", "global_step": 1728, "tokens_seen": 1811939328, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2492531368253336, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2492531368253336], "acc": 0.25253933479386576, "acc_ci": 0.0, "acc_runs": [0.25253933479386576], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00431696567867478, "centered_acc": -0.0009958175662218756, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0009958175662218756], "ci": 0.0}, "piqa": {"acc_norm": 0.515778019586507, "acc_norm_ci": 0.0, "acc_norm_runs": [0.515778019586507], "acc": 0.5331882480957563, "acc_ci": 0.0, "acc_runs": [0.5331882480957563], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011660014400426034, "centered_acc": 0.03155603917301408, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03155603917301408], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26936026936026936, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26936026936026936], "acc": 0.26304713804713803, "acc_ci": 0.0, "acc_runs": [0.26304713804713803], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009103043207756808, "centered_acc": 0.02581369248035914, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02581369248035914], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26023890784982934, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26023890784982934], "acc": 0.20648464163822525, "acc_ci": 0.0, "acc_runs": [0.20648464163822525], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012821930225112672, "centered_acc": 0.013651877133105783, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013651877133105783], "ci": 0.0}, "openbookqa": {"acc_norm": 0.296, "acc_norm_ci": 0.0, "acc_norm_runs": [0.296], "acc": 0.17, "acc_ci": 0.0, "acc_runs": [0.17], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020435342091896222, "centered_acc": 0.061333333333333316, "centered_acc_ci": 0.0, "centered_acc_runs": [0.061333333333333316], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.510655090765588, "acc_ci": 0.0, "acc_runs": [0.510655090765588], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014049294536290372, "centered_acc": 0.02131018153117603, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02131018153117603], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.51, "acc_ci": 0.0, "acc_runs": [0.51], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05024183937956913, "centered_acc": 0.020000000000000018, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020000000000000018], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 15918475.055290213, "perplexity_ci": 0.0, "perplexity_runs": [15918475.055290213], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.182, "acc_norm_ci": 0.0, "acc_norm_runs": [0.182], "acc": 0.05, "acc_ci": 0.0, "acc_runs": [0.05], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0122075806376622, "centered_acc": -0.09066666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09066666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29543838893698987, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28742019151142734, "mean_acc_ci": 0.0, "mean_centered_acc": -0.017935826337166616, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29543838893698987], "mean_acc_runs": [0.28742019151142734], "mean_centered_acc_runs": [-0.017935826337166616], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-2016", "checkpoint": "checkpoint-2016", "global_step": 2016, "tokens_seen": 2113929216, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25492929695279826, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25492929695279826], "acc": 0.25363473411670984, "acc_ci": 0.0, "acc_runs": [0.25363473411670984], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043493077027353935, "centered_acc": 0.006572395937064342, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006572395937064342], "ci": 0.0}, "piqa": {"acc_norm": 0.5228509249183896, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5228509249183896], "acc": 0.5331882480957563, "acc_ci": 0.0, "acc_runs": [0.5331882480957563], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011653634832401168, "centered_acc": 0.04570184983677916, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04570184983677916], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26515151515151514, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26515151515151514], "acc": 0.25673400673400676, "acc_ci": 0.0, "acc_runs": [0.25673400673400676], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009057621139172679, "centered_acc": 0.020202020202020183, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020202020202020183], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2627986348122867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2627986348122867], "acc": 0.2090443686006826, "acc_ci": 0.0, "acc_runs": [0.2090443686006826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012862523175351397, "centered_acc": 0.017064846416382284, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017064846416382284], "ci": 0.0}, "openbookqa": {"acc_norm": 0.284, "acc_norm_ci": 0.0, "acc_norm_runs": [0.284], "acc": 0.172, "acc_ci": 0.0, "acc_runs": [0.172], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020186703693570774, "centered_acc": 0.0453333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0453333333333333], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5098658247829518, "acc_ci": 0.0, "acc_runs": [0.5098658247829518], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014049749833367674, "centered_acc": 0.019731649565903675, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019731649565903675], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5, "acc_ci": 0.0, "acc_runs": [0.5], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.050251890762960605, "centered_acc": 0.0, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 29474969.753621306, "perplexity_ci": 0.0, "perplexity_runs": [29474969.753621306], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.041, "acc_ci": 0.0, "acc_runs": [0.041], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2951217286391649, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28537546441038075, "mean_acc_ci": 0.0, "mean_centered_acc": -0.020090997906790715, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2951217286391649], "mean_acc_runs": [0.28537546441038075], "mean_centered_acc_runs": [-0.020090997906790715], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-2304", "checkpoint": "checkpoint-2304", "global_step": 2304, "tokens_seen": 2415919104, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25423222465644296, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25423222465644296], "acc": 0.25562636924915355, "acc_ci": 0.0, "acc_runs": [0.25562636924915355], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00434538861452022, "centered_acc": 0.005642966208590616, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005642966208590616], "ci": 0.0}, "piqa": {"acc_norm": 0.5223068552774756, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5223068552774756], "acc": 0.5391730141458106, "acc_ci": 0.0, "acc_runs": [0.5391730141458106], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011654208652596384, "centered_acc": 0.044613710554951114, "centered_acc_ci": 0.0, "centered_acc_runs": [0.044613710554951114], "ci": 0.0}, "arc_easy": {"acc_norm": 0.273989898989899, "acc_norm_ci": 0.0, "acc_norm_runs": [0.273989898989899], "acc": 0.2702020202020202, "acc_ci": 0.0, "acc_runs": [0.2702020202020202], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00915180590154416, "centered_acc": 0.03198653198653201, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03198653198653201], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2593856655290102, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2593856655290102], "acc": 0.2167235494880546, "acc_ci": 0.0, "acc_runs": [0.2167235494880546], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012808273573927224, "centered_acc": 0.012514220705346949, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012514220705346949], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.18, "acc_ci": 0.0, "acc_runs": [0.18], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.500394632991318, "acc_ci": 0.0, "acc_runs": [0.500394632991318], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405248130604961, "centered_acc": 0.0007892659826360671, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0007892659826360671], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.53, "acc_ci": 0.0, "acc_runs": [0.53], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05016135580465919, "centered_acc": 0.06000000000000005, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06000000000000005], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 41786572.40411945, "perplexity_ci": 0.0, "perplexity_runs": [41786572.40411945], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2941524407421379, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29174070478500574, "mean_acc_ci": 0.0, "mean_centered_acc": -0.01643093122383806, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2941524407421379], "mean_acc_runs": [0.29174070478500574], "mean_centered_acc_runs": [-0.01643093122383806], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-2592", "checkpoint": "checkpoint-2592", "global_step": 2592, "tokens_seen": 2717908992, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25413264289982074, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25413264289982074], "acc": 0.255327623979287, "acc_ci": 0.0, "acc_runs": [0.255327623979287], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004344827546976306, "centered_acc": 0.005510190533094317, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005510190533094317], "ci": 0.0}, "piqa": {"acc_norm": 0.5038084874863983, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5038084874863983], "acc": 0.5261153427638737, "acc_ci": 0.0, "acc_runs": [0.5261153427638737], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665485744747022, "centered_acc": 0.007616974972796564, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007616974972796564], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26515151515151514, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26515151515151514], "acc": 0.26304713804713803, "acc_ci": 0.0, "acc_runs": [0.26304713804713803], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009057621139172679, "centered_acc": 0.020202020202020183, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020202020202020183], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2645051194539249, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2645051194539249], "acc": 0.2235494880546075, "acc_ci": 0.0, "acc_runs": [0.2235494880546075], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012889272949313479, "centered_acc": 0.019340159271899877, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019340159271899877], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.18, "acc_ci": 0.0, "acc_runs": [0.18], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.510655090765588, "acc_ci": 0.0, "acc_runs": [0.510655090765588], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014049294536290372, "centered_acc": 0.02131018153117603, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02131018153117603], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.51, "acc_ci": 0.0, "acc_runs": [0.51], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05024183937956913, "centered_acc": 0.020000000000000018, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020000000000000018], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 81037210.12118527, "perplexity_ci": 0.0, "perplexity_runs": [81037210.12118527], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.182, "acc_norm_ci": 0.0, "acc_norm_runs": [0.182], "acc": 0.04, "acc_ci": 0.0, "acc_runs": [0.04], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0122075806376622, "centered_acc": -0.09066666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09066666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29059962749860985, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28869821453841943, "mean_acc_ci": 0.0, "mean_centered_acc": -0.023123579623142113, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29059962749860985], "mean_acc_runs": [0.28869821453841943], "mean_centered_acc_runs": [-0.023123579623142113], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-2880", "checkpoint": "checkpoint-2880", "global_step": 2880, "tokens_seen": 3019898880, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2522405895239992, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2522405895239992], "acc": 0.25323640709022105, "acc_ci": 0.0, "acc_runs": [0.25323640709022105], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004334110169012113, "centered_acc": 0.00298745269866559, "centered_acc_ci": 0.0, "centered_acc_runs": [0.00298745269866559], "ci": 0.0}, "piqa": {"acc_norm": 0.5103373231773667, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5103373231773667], "acc": 0.529923830250272, "acc_ci": 0.0, "acc_runs": [0.529923830250272], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01166333067307569, "centered_acc": 0.020674646354733373, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020674646354733373], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2668350168350168, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2668350168350168], "acc": 0.26052188552188554, "acc_ci": 0.0, "acc_runs": [0.26052188552188554], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009075915859266983, "centered_acc": 0.02244668911335575, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02244668911335575], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.27303754266211605, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27303754266211605], "acc": 0.22098976109215018, "acc_ci": 0.0, "acc_runs": [0.22098976109215018], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01301933276263586, "centered_acc": 0.030716723549488067, "centered_acc_ci": 0.0, "centered_acc_runs": [0.030716723549488067], "ci": 0.0}, "openbookqa": {"acc_norm": 0.292, "acc_norm_ci": 0.0, "acc_norm_runs": [0.292], "acc": 0.184, "acc_ci": 0.0, "acc_runs": [0.184], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020354375480530096, "centered_acc": 0.05599999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05599999999999997], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.500394632991318, "acc_ci": 0.0, "acc_runs": [0.500394632991318], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405248130604961, "centered_acc": 0.0007892659826360671, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0007892659826360671], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.52, "acc_ci": 0.0, "acc_runs": [0.52], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05021167315686783, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 132665039.98033264, "perplexity_ci": 0.0, "perplexity_runs": [132665039.98033264], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.178, "acc_norm_ci": 0.0, "acc_norm_runs": [0.178], "acc": 0.042, "acc_ci": 0.0, "acc_runs": [0.042], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012102167676183571, "centered_acc": -0.09600000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09600000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29540841203308316, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2889353978719547, "mean_acc_ci": 0.0, "mean_centered_acc": -0.01842336652819117, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29540841203308316], "mean_acc_runs": [0.2889353978719547], "mean_centered_acc_runs": [-0.01842336652819117], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-3168", "checkpoint": "checkpoint-3168", "global_step": 3168, "tokens_seen": 3321888768, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25423222465644296, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25423222465644296], "acc": 0.25124477195777734, "acc_ci": 0.0, "acc_runs": [0.25124477195777734], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00434538861452022, "centered_acc": 0.005642966208590616, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005642966208590616], "ci": 0.0}, "piqa": {"acc_norm": 0.514689880304679, "acc_norm_ci": 0.0, "acc_norm_runs": [0.514689880304679], "acc": 0.5239390642002176, "acc_ci": 0.0, "acc_runs": [0.5239390642002176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011660788281735552, "centered_acc": 0.029379760609357986, "centered_acc_ci": 0.0, "centered_acc_runs": [0.029379760609357986], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26641414141414144, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26641414141414144], "acc": 0.26136363636363635, "acc_ci": 0.0, "acc_runs": [0.26136363636363635], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009071357971078843, "centered_acc": 0.021885521885521914, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021885521885521914], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2636518771331058, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2636518771331058], "acc": 0.2150170648464164, "acc_ci": 0.0, "acc_runs": [0.2150170648464164], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012875929151296968, "centered_acc": 0.018202502844141044, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018202502844141044], "ci": 0.0}, "openbookqa": {"acc_norm": 0.302, "acc_norm_ci": 0.0, "acc_norm_runs": [0.302], "acc": 0.18, "acc_ci": 0.0, "acc_runs": [0.18], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02055326917420912, "centered_acc": 0.06933333333333332, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06933333333333332], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.494869771112865, "acc_ci": 0.0, "acc_runs": [0.494869771112865], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051745961790525, "centered_acc": -0.010260457774269982, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.010260457774269982], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.49, "acc_ci": 0.0, "acc_runs": [0.49], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05024183937956913, "centered_acc": -0.020000000000000018, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.020000000000000018], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 198677019.72047818, "perplexity_ci": 0.0, "perplexity_runs": [198677019.72047818], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.177, "acc_norm_ci": 0.0, "acc_norm_runs": [0.177], "acc": 0.048, "acc_ci": 0.0, "acc_runs": [0.048], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012075463420375087, "centered_acc": -0.09733333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09733333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29633135391806154, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2842721770254613, "mean_acc_ci": 0.0, "mean_centered_acc": -0.02517497585325087, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29633135391806154], "mean_acc_runs": [0.2842721770254613], "mean_centered_acc_runs": [-0.02517497585325087], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-3456", "checkpoint": "checkpoint-3456", "global_step": 3456, "tokens_seen": 3623878656, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2529376618203545, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2529376618203545], "acc": 0.2533359888468433, "acc_ci": 0.0, "acc_runs": [0.2533359888468433], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043380713189125845, "centered_acc": 0.003916882427139316, "centered_acc_ci": 0.0, "centered_acc_runs": [0.003916882427139316], "ci": 0.0}, "piqa": {"acc_norm": 0.5097932535364527, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5097932535364527], "acc": 0.5206746463547334, "acc_ci": 0.0, "acc_runs": [0.5206746463547334], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011663586263283307, "centered_acc": 0.019586507072905324, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019586507072905324], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26936026936026936, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26936026936026936], "acc": 0.26304713804713803, "acc_ci": 0.0, "acc_runs": [0.26304713804713803], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009103043207756808, "centered_acc": 0.02581369248035914, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02581369248035914], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26535836177474403, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26535836177474403], "acc": 0.22013651877133106, "acc_ci": 0.0, "acc_runs": [0.22013651877133106], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012902554762313832, "centered_acc": 0.020477815699658713, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020477815699658713], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.178, "acc_ci": 0.0, "acc_runs": [0.178], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5059194948697711, "acc_ci": 0.0, "acc_runs": [0.5059194948697711], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051500838485923, "centered_acc": 0.011838989739542116, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011838989739542116], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 284454300.4646191, "perplexity_ci": 0.0, "perplexity_runs": [284454300.4646191], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.18, "acc_norm_ci": 0.0, "acc_norm_runs": [0.18], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012155153135512024, "centered_acc": -0.09333333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09333333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29324159108197007, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29054012486635167, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0147175395221846, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29324159108197007], "mean_acc_runs": [0.29054012486635167], "mean_centered_acc_runs": [-0.0147175395221846], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-3744", "checkpoint": "checkpoint-3744", "global_step": 3744, "tokens_seen": 3925868544, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2517426807408883, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2517426807408883], "acc": 0.24995020912168892, "acc_ci": 0.0, "acc_runs": [0.24995020912168892], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004331271717773648, "centered_acc": 0.0023235743211843887, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0023235743211843887], "ci": 0.0}, "piqa": {"acc_norm": 0.5119695321001088, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5119695321001088], "acc": 0.5217627856365615, "acc_ci": 0.0, "acc_runs": [0.5217627856365615], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011662480968070035, "centered_acc": 0.02393906420021752, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02393906420021752], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2622053872053872, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2622053872053872], "acc": 0.25252525252525254, "acc_ci": 0.0, "acc_runs": [0.25252525252525254], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009025197991724937, "centered_acc": 0.01627384960718296, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01627384960718296], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2636518771331058, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2636518771331058], "acc": 0.22013651877133106, "acc_ci": 0.0, "acc_runs": [0.22013651877133106], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012875929151296968, "centered_acc": 0.018202502844141044, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018202502844141044], "ci": 0.0}, "openbookqa": {"acc_norm": 0.294, "acc_norm_ci": 0.0, "acc_norm_runs": [0.294], "acc": 0.176, "acc_ci": 0.0, "acc_runs": [0.176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02039509548493655, "centered_acc": 0.058666666666666645, "centered_acc_ci": 0.0, "centered_acc_runs": [0.058666666666666645], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5011838989739542, "acc_ci": 0.0, "acc_runs": [0.5011838989739542], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052446290528871, "centered_acc": 0.0023677979479084232, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0023677979479084232], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 371243385.9219187, "perplexity_ci": 0.0, "perplexity_runs": [371243385.9219187], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.175, "acc_norm_ci": 0.0, "acc_norm_runs": [0.175], "acc": 0.042, "acc_ci": 0.0, "acc_runs": [0.042], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012021627157731998, "centered_acc": -0.10000000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10000000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29309491286324835, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2891846126802488, "mean_acc_ci": 0.0, "mean_centered_acc": -0.013516846762810923, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29309491286324835], "mean_acc_runs": [0.2891846126802488], "mean_centered_acc_runs": [-0.013516846762810923], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-4032", "checkpoint": "checkpoint-4032", "global_step": 4032, "tokens_seen": 4227858432, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25363473411670984, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25363473411670984], "acc": 0.24995020912168892, "acc_ci": 0.0, "acc_runs": [0.24995020912168892], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004342017709967513, "centered_acc": 0.0048463121556131155, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0048463121556131155], "ci": 0.0}, "piqa": {"acc_norm": 0.514145810663765, "acc_norm_ci": 0.0, "acc_norm_runs": [0.514145810663765], "acc": 0.5408052230685527, "acc_ci": 0.0, "acc_runs": [0.5408052230685527], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011661154475524689, "centered_acc": 0.028291621327529937, "centered_acc_ci": 0.0, "centered_acc_runs": [0.028291621327529937], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26262626262626265, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26262626262626265], "acc": 0.26136363636363635, "acc_ci": 0.0, "acc_runs": [0.26136363636363635], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009029861776763957, "centered_acc": 0.01683501683501687, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01683501683501687], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2551194539249147, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2551194539249147], "acc": 0.21245733788395904, "acc_ci": 0.0, "acc_runs": [0.21245733788395904], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012739038695202038, "centered_acc": 0.006825938566552929, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006825938566552929], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.18, "acc_ci": 0.0, "acc_runs": [0.18], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48855564325177586, "acc_ci": 0.0, "acc_runs": [0.48855564325177586], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014048804199859391, "centered_acc": -0.022888713496448276, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.022888713496448276], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.53, "acc_ci": 0.0, "acc_runs": [0.53], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05016135580465919, "centered_acc": 0.06000000000000005, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06000000000000005], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 540085106.0808784, "perplexity_ci": 0.0, "perplexity_runs": [540085106.0808784], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.179, "acc_norm_ci": 0.0, "acc_norm_runs": [0.179], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012128730605719054, "centered_acc": -0.09466666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09466666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2917543768886087, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2885419511463313, "mean_acc_ci": 0.0, "mean_centered_acc": -0.021797951970111272, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2917543768886087], "mean_acc_runs": [0.2885419511463313], "mean_centered_acc_runs": [-0.021797951970111272], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-4320", "checkpoint": "checkpoint-4320", "global_step": 4320, "tokens_seen": 4529848320, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.252141007767377, "acc_norm_ci": 0.0, "acc_norm_runs": [0.252141007767377], "acc": 0.2515435172276439, "acc_ci": 0.0, "acc_runs": [0.2515435172276439], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004333543083293202, "centered_acc": 0.0028546770231693643, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0028546770231693643], "ci": 0.0}, "piqa": {"acc_norm": 0.5239390642002176, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5239390642002176], "acc": 0.5369967355821545, "acc_ci": 0.0, "acc_runs": [0.5369967355821545], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165244562107948, "centered_acc": 0.04787812840043526, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04787812840043526], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2668350168350168, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2668350168350168], "acc": 0.2617845117845118, "acc_ci": 0.0, "acc_runs": [0.2617845117845118], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009075915859266983, "centered_acc": 0.02244668911335575, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02244668911335575], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26109215017064846, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26109215017064846], "acc": 0.2090443686006826, "acc_ci": 0.0, "acc_runs": [0.2090443686006826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012835523909473708, "centered_acc": 0.014789533560864617, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014789533560864617], "ci": 0.0}, "openbookqa": {"acc_norm": 0.29, "acc_norm_ci": 0.0, "acc_norm_runs": [0.29], "acc": 0.176, "acc_ci": 0.0, "acc_runs": [0.176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020313179231745138, "centered_acc": 0.05333333333333331, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05333333333333331], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5098658247829518, "acc_ci": 0.0, "acc_runs": [0.5098658247829518], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014049749833367674, "centered_acc": 0.019731649565903675, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019731649565903675], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 879369525.3834081, "perplexity_ci": 0.0, "perplexity_runs": [879369525.3834081], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.175, "acc_norm_ci": 0.0, "acc_norm_runs": [0.175], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012021627157731998, "centered_acc": -0.10000000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10000000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2948345398288767, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2917522419751645, "mean_acc_ci": 0.0, "mean_centered_acc": -0.009154562828393034, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2948345398288767], "mean_acc_runs": [0.2917522419751645], "mean_centered_acc_runs": [-0.009154562828393034], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-4608", "checkpoint": "checkpoint-4608", "global_step": 4608, "tokens_seen": 4831838208, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2513443537143995, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2513443537143995], "acc": 0.2526389165504879, "acc_ci": 0.0, "acc_runs": [0.2526389165504879], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043289955103121775, "centered_acc": 0.001792471619199339, "centered_acc_ci": 0.0, "centered_acc_runs": [0.001792471619199339], "ci": 0.0}, "piqa": {"acc_norm": 0.5174102285092492, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5174102285092492], "acc": 0.5201305767138193, "acc_ci": 0.0, "acc_runs": [0.5201305767138193], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011658749823107507, "centered_acc": 0.03482045701849845, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03482045701849845], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26515151515151514, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26515151515151514], "acc": 0.26430976430976433, "acc_ci": 0.0, "acc_runs": [0.26430976430976433], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009057621139172679, "centered_acc": 0.020202020202020183, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020202020202020183], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2551194539249147, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2551194539249147], "acc": 0.2030716723549488, "acc_ci": 0.0, "acc_runs": [0.2030716723549488], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012739038695202038, "centered_acc": 0.006825938566552929, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006825938566552929], "ci": 0.0}, "openbookqa": {"acc_norm": 0.29, "acc_norm_ci": 0.0, "acc_norm_runs": [0.29], "acc": 0.178, "acc_ci": 0.0, "acc_runs": [0.178], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020313179231745138, "centered_acc": 0.05333333333333331, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05333333333333331], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5043409629044988, "acc_ci": 0.0, "acc_runs": [0.5043409629044988], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051956064076884, "centered_acc": 0.008681925808997626, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008681925808997626], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 1265625603.7055895, "perplexity_ci": 0.0, "perplexity_runs": [1265625603.7055895], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.177, "acc_norm_ci": 0.0, "acc_norm_runs": [0.177], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012075463420375087, "centered_acc": -0.09733333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09733333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29267092521667976, "mean_acc_norm_ci": 0.0, "mean_acc": 0.288377935460722, "mean_acc_ci": 0.0, "mean_centered_acc": -0.01501136258192564, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29267092521667976], "mean_acc_runs": [0.288377935460722], "mean_centered_acc_runs": [-0.01501136258192564], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-4896", "checkpoint": "checkpoint-4896", "global_step": 4896, "tokens_seen": 5133828096, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25363473411670984, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25363473411670984], "acc": 0.2500497908783111, "acc_ci": 0.0, "acc_runs": [0.2500497908783111], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004342017709967513, "centered_acc": 0.0048463121556131155, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0048463121556131155], "ci": 0.0}, "piqa": {"acc_norm": 0.49836779107725787, "acc_norm_ci": 0.0, "acc_norm_runs": [0.49836779107725787], "acc": 0.5255712731229597, "acc_ci": 0.0, "acc_runs": [0.5255712731229597], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665762007194807, "centered_acc": -0.0032644178454842576, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0032644178454842576], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2622053872053872, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2622053872053872], "acc": 0.2558922558922559, "acc_ci": 0.0, "acc_runs": [0.2558922558922559], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009025197991724937, "centered_acc": 0.01627384960718296, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01627384960718296], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2508532423208191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2508532423208191], "acc": 0.20648464163822525, "acc_ci": 0.0, "acc_runs": [0.20648464163822525], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012668198621315347, "centered_acc": 0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.172, "acc_ci": 0.0, "acc_runs": [0.172], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4980268350434096, "acc_ci": 0.0, "acc_runs": [0.4980268350434096], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052376259225643, "centered_acc": -0.003946329913180779, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.003946329913180779], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 1603216289.6871345, "perplexity_ci": 0.0, "perplexity_runs": [1603216289.6871345], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.172, "acc_norm_ci": 0.0, "acc_norm_runs": [0.172], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011939788882495308, "centered_acc": -0.10400000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10400000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.28617685912002905, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2870312258348862, "mean_acc_ci": 0.0, "mean_centered_acc": -0.023597556224523267, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.28617685912002905], "mean_acc_runs": [0.2870312258348862], "mean_centered_acc_runs": [-0.023597556224523267], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-5184", "checkpoint": "checkpoint-5184", "global_step": 5184, "tokens_seen": 5435817984, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25104560844453294, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25104560844453294], "acc": 0.24805815574586737, "acc_ci": 0.0, "acc_runs": [0.24805815574586737], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004327285172595903, "centered_acc": 0.0013941445927105889, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0013941445927105889], "ci": 0.0}, "piqa": {"acc_norm": 0.49782372143634385, "acc_norm_ci": 0.0, "acc_norm_runs": [0.49782372143634385], "acc": 0.5195865070729053, "acc_ci": 0.0, "acc_runs": [0.5195865070729053], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665713661738719, "centered_acc": -0.0043525571273123065, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0043525571273123065], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2609427609427609, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2609427609427609], "acc": 0.2588383838383838, "acc_ci": 0.0, "acc_runs": [0.2588383838383838], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009011142493235922, "centered_acc": 0.014590347923681227, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014590347923681227], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.257679180887372, "acc_norm_ci": 0.0, "acc_norm_runs": [0.257679180887372], "acc": 0.2175767918088737, "acc_ci": 0.0, "acc_runs": [0.2175767918088737], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012780770562768334, "centered_acc": 0.010238907849829356, "centered_acc_ci": 0.0, "centered_acc_runs": [0.010238907849829356], "ci": 0.0}, "openbookqa": {"acc_norm": 0.278, "acc_norm_ci": 0.0, "acc_norm_runs": [0.278], "acc": 0.172, "acc_ci": 0.0, "acc_runs": [0.172], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020055833888070865, "centered_acc": 0.037333333333333364, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037333333333333364], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5272296764009471, "acc_ci": 0.0, "acc_runs": [0.5272296764009471], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01403163162982771, "centered_acc": 0.05445935280189418, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05445935280189418], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 2223016723.1563606, "perplexity_ci": 0.0, "perplexity_runs": [2223016723.1563606], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.173, "acc_norm_ci": 0.0, "acc_norm_runs": [0.173], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011967214137559927, "centered_acc": -0.10266666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10266666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.28641521195183495, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2917576976640678, "mean_acc_ci": 0.0, "mean_centered_acc": -0.014714245971681063, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.28641521195183495], "mean_acc_runs": [0.2917576976640678], "mean_centered_acc_runs": [-0.014714245971681063], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-5472", "checkpoint": "checkpoint-5472", "global_step": 5472, "tokens_seen": 5737807872, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.24965146385182235, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24965146385182235], "acc": 0.2484564827723561, "acc_ci": 0.0, "acc_runs": [0.2484564827723561], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00431926743246105, "centered_acc": -0.000464714864236863, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.000464714864236863], "ci": 0.0}, "piqa": {"acc_norm": 0.5087051142546246, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5087051142546246], "acc": 0.5163220892274211, "acc_ci": 0.0, "acc_runs": [0.5163220892274211], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011664055982032772, "centered_acc": 0.017410228509249226, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017410228509249226], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2647306397306397, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2647306397306397], "acc": 0.26304713804713803, "acc_ci": 0.0, "acc_runs": [0.26304713804713803], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009053021086174166, "centered_acc": 0.019640852974186274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019640852974186274], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25597269624573377, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25597269624573377], "acc": 0.21928327645051193, "acc_ci": 0.0, "acc_runs": [0.21928327645051193], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012753013241244613, "centered_acc": 0.007963594994311688, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007963594994311688], "ci": 0.0}, "openbookqa": {"acc_norm": 0.298, "acc_norm_ci": 0.0, "acc_norm_runs": [0.298], "acc": 0.17, "acc_ci": 0.0, "acc_runs": [0.17], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020475118092988954, "centered_acc": 0.06399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06399999999999999], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5367008681925809, "acc_ci": 0.0, "acc_runs": [0.5367008681925809], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014014578458843366, "centered_acc": 0.07340173638516179, "centered_acc_ci": 0.0, "centered_acc_runs": [0.07340173638516179], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 3581626111.034089, "perplexity_ci": 0.0, "perplexity_runs": [3581626111.034089], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.176, "acc_norm_ci": 0.0, "acc_norm_runs": [0.176], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012048616898597498, "centered_acc": -0.09866666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09866666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29217665234713674, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2928097316463708, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0066822272356215386, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29217665234713674], "mean_acc_runs": [0.2928097316463708], "mean_centered_acc_runs": [-0.0066822272356215386], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-5760", "checkpoint": "checkpoint-5760", "global_step": 5760, "tokens_seen": 6039797760, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2548297151961761, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2548297151961761], "acc": 0.25104560844453294, "acc_ci": 0.0, "acc_runs": [0.25104560844453294], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004348748730529728, "centered_acc": 0.006439620261568117, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006439620261568117], "ci": 0.0}, "piqa": {"acc_norm": 0.5027203482045702, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5027203482045702], "acc": 0.5233949945593036, "acc_ci": 0.0, "acc_runs": [0.5233949945593036], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665651503000776, "centered_acc": 0.005440696409140466, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005440696409140466], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2647306397306397, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2647306397306397], "acc": 0.26304713804713803, "acc_ci": 0.0, "acc_runs": [0.26304713804713803], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009053021086174166, "centered_acc": 0.019640852974186274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019640852974186274], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2508532423208191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2508532423208191], "acc": 0.21245733788395904, "acc_ci": 0.0, "acc_runs": [0.21245733788395904], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012668198621315347, "centered_acc": 0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.29, "acc_norm_ci": 0.0, "acc_norm_runs": [0.29], "acc": 0.17, "acc_ci": 0.0, "acc_runs": [0.17], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020313179231745138, "centered_acc": 0.05333333333333331, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05333333333333331], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4877663772691397, "acc_ci": 0.0, "acc_runs": [0.4877663772691397], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01404827882040573, "centered_acc": -0.024467245461720633, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.024467245461720633], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 4085138069.983313, "perplexity_ci": 0.0, "perplexity_runs": [4085138069.983313], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.167, "acc_norm_ci": 0.0, "acc_norm_runs": [0.167], "acc": 0.05, "acc_ci": 0.0, "acc_runs": [0.05], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011800434324644591, "centered_acc": -0.11066666666666665, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.11066666666666665], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2883556575753676, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2886192979689398, "mean_acc_ci": 0.0, "mean_centered_acc": -0.02139631435277773, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2883556575753676], "mean_acc_runs": [0.2886192979689398], "mean_centered_acc_runs": [-0.02139631435277773], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-6048", "checkpoint": "checkpoint-6048", "global_step": 6048, "tokens_seen": 6341787648, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25473013343955386, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25473013343955386], "acc": 0.24915355506871142, "acc_ci": 0.0, "acc_runs": [0.24915355506871142], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043481894593367845, "centered_acc": 0.006306844586071818, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006306844586071818], "ci": 0.0}, "piqa": {"acc_norm": 0.5070729053318824, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5070729053318824], "acc": 0.5038084874863983, "acc_ci": 0.0, "acc_runs": [0.5038084874863983], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011664656918146136, "centered_acc": 0.014145810663764857, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014145810663764857], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2680976430976431, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2680976430976431], "acc": 0.2617845117845118, "acc_ci": 0.0, "acc_runs": [0.2617845117845118], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009089526578213825, "centered_acc": 0.024130190796857482, "centered_acc_ci": 0.0, "centered_acc_runs": [0.024130190796857482], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25170648464163825, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25170648464163825], "acc": 0.21331058020477817, "acc_ci": 0.0, "acc_runs": [0.21331058020477817], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012682496334042911, "centered_acc": 0.0022753128555176674, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0022753128555176674], "ci": 0.0}, "openbookqa": {"acc_norm": 0.304, "acc_norm_ci": 0.0, "acc_norm_runs": [0.304], "acc": 0.184, "acc_ci": 0.0, "acc_runs": [0.184], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02059164957122494, "centered_acc": 0.072, "centered_acc_ci": 0.0, "centered_acc_runs": [0.072], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5082872928176796, "acc_ci": 0.0, "acc_runs": [0.5082872928176796], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405055532282417, "centered_acc": 0.016574585635359185, "centered_acc_ci": 0.0, "centered_acc_runs": [0.016574585635359185], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 5348042468.434231, "perplexity_ci": 0.0, "perplexity_runs": [5348042468.434231], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.176, "acc_norm_ci": 0.0, "acc_norm_runs": [0.176], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012048616898597498, "centered_acc": -0.09866666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09866666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29360119441845295, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28956318891357796, "mean_acc_ci": 0.0, "mean_centered_acc": -0.01185099984241055, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29360119441845295], "mean_acc_runs": [0.28956318891357796], "mean_centered_acc_runs": [-0.01185099984241055], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-6336", "checkpoint": "checkpoint-6336", "global_step": 6336, "tokens_seen": 6643777536, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25672176857199763, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25672176857199763], "acc": 0.24975104560844452, "acc_ci": 0.0, "acc_runs": [0.24975104560844452], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043593182064291, "centered_acc": 0.008962358095996844, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008962358095996844], "ci": 0.0}, "piqa": {"acc_norm": 0.5048966267682263, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5048966267682263], "acc": 0.5239390642002176, "acc_ci": 0.0, "acc_runs": [0.5239390642002176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665264730078137, "centered_acc": 0.009793253536452662, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009793253536452662], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26430976430976433, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26430976430976433], "acc": 0.26641414141414144, "acc_ci": 0.0, "acc_runs": [0.26641414141414144], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009048410451862887, "centered_acc": 0.019079685746352437, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019079685746352437], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2508532423208191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2508532423208191], "acc": 0.2158703071672355, "acc_ci": 0.0, "acc_runs": [0.2158703071672355], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012668198621315347, "centered_acc": 0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.29, "acc_norm_ci": 0.0, "acc_norm_runs": [0.29], "acc": 0.184, "acc_ci": 0.0, "acc_runs": [0.184], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020313179231745138, "centered_acc": 0.05333333333333331, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05333333333333331], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.489344909234412, "acc_ci": 0.0, "acc_runs": [0.489344909234412], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014049294536290372, "centered_acc": -0.02131018153117603, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.02131018153117603], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48, "acc_ci": 0.0, "acc_runs": [0.48], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05021167315686783, "centered_acc": -0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.040000000000000036], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 6327859426.910668, "perplexity_ci": 0.0, "perplexity_runs": [6327859426.910668], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.174, "acc_norm_ci": 0.0, "acc_norm_runs": [0.174], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011994493230973449, "centered_acc": -0.10133333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10133333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29013023366180124, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2834606929398152, "mean_acc_ci": 0.0, "mean_centered_acc": -0.03486247824191274, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29013023366180124], "mean_acc_runs": [0.2834606929398152], "mean_centered_acc_runs": [-0.03486247824191274], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-6624", "checkpoint": "checkpoint-6624", "global_step": 6624, "tokens_seen": 6945767424, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25273849830711015, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25273849830711015], "acc": 0.25054769966142204, "acc_ci": 0.0, "acc_runs": [0.25054769966142204], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043369410695685015, "centered_acc": 0.003651331076146865, "centered_acc_ci": 0.0, "centered_acc_runs": [0.003651331076146865], "ci": 0.0}, "piqa": {"acc_norm": 0.514145810663765, "acc_norm_ci": 0.0, "acc_norm_runs": [0.514145810663765], "acc": 0.5163220892274211, "acc_ci": 0.0, "acc_runs": [0.5163220892274211], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011661154475524689, "centered_acc": 0.028291621327529937, "centered_acc_ci": 0.0, "centered_acc_runs": [0.028291621327529937], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2706228956228956, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2706228956228956], "acc": 0.2680976430976431, "acc_ci": 0.0, "acc_runs": [0.2680976430976431], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009116466166404035, "centered_acc": 0.0274971941638608, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0274971941638608], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25853242320819114, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25853242320819114], "acc": 0.22013651877133106, "acc_ci": 0.0, "acc_runs": [0.22013651877133106], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012794553754288777, "centered_acc": 0.011376564277588189, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011376564277588189], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.174, "acc_ci": 0.0, "acc_runs": [0.174], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5082872928176796, "acc_ci": 0.0, "acc_runs": [0.5082872928176796], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405055532282417, "centered_acc": 0.016574585635359185, "centered_acc_ci": 0.0, "centered_acc_runs": [0.016574585635359185], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.52, "acc_ci": 0.0, "acc_runs": [0.52], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05021167315686783, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 10927725204.81932, "perplexity_ci": 0.0, "perplexity_runs": [10927725204.81932], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.174, "acc_norm_ci": 0.0, "acc_norm_runs": [0.174], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011994493230973449, "centered_acc": -0.10133333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10133333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.291673271300327, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2882872767060822, "mean_acc_ci": 0.0, "mean_centered_acc": -0.019707457033938627, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.291673271300327], "mean_acc_runs": [0.2882872767060822], "mean_centered_acc_runs": [-0.019707457033938627], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-6912", "checkpoint": "checkpoint-6912", "global_step": 6912, "tokens_seen": 7247757312, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25273849830711015, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25273849830711015], "acc": 0.24945230033857796, "acc_ci": 0.0, "acc_runs": [0.24945230033857796], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043369410695685015, "centered_acc": 0.003651331076146865, "centered_acc_ci": 0.0, "centered_acc_runs": [0.003651331076146865], "ci": 0.0}, "piqa": {"acc_norm": 0.5163220892274211, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5163220892274211], "acc": 0.515233949945593, "acc_ci": 0.0, "acc_runs": [0.515233949945593], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011659606710151597, "centered_acc": 0.03264417845484213, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03264417845484213], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2601010101010101, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2601010101010101], "acc": 0.26304713804713803, "acc_ci": 0.0, "acc_runs": [0.26304713804713803], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009001718541079735, "centered_acc": 0.013468013468013481, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013468013468013481], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25597269624573377, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25597269624573377], "acc": 0.21928327645051193, "acc_ci": 0.0, "acc_runs": [0.21928327645051193], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012753013241244613, "centered_acc": 0.007963594994311688, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007963594994311688], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.176, "acc_ci": 0.0, "acc_runs": [0.176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5011838989739542, "acc_ci": 0.0, "acc_runs": [0.5011838989739542], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052446290528871, "centered_acc": 0.0023677979479084232, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0023677979479084232], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.51, "acc_ci": 0.0, "acc_runs": [0.51], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05024183937956913, "centered_acc": 0.020000000000000018, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020000000000000018], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 12196371440.19413, "perplexity_ci": 0.0, "perplexity_runs": [12196371440.19413], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.174, "acc_norm_ci": 0.0, "acc_norm_runs": [0.174], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011994493230973449, "centered_acc": -0.10133333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10133333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2898557156468792, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28584880255294753, "mean_acc_ci": 0.0, "mean_centered_acc": -0.02496261042719001, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2898557156468792], "mean_acc_runs": [0.28584880255294753], "mean_centered_acc_runs": [-0.02496261042719001], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-7200", "checkpoint": "checkpoint-7200", "global_step": 7200, "tokens_seen": 7549747200, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2531368253335989, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2531368253335989], "acc": 0.24965146385182235, "acc_ci": 0.0, "acc_runs": [0.24965146385182235], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004339200363454809, "centered_acc": 0.00418243377813184, "centered_acc_ci": 0.0, "centered_acc_runs": [0.00418243377813184], "ci": 0.0}, "piqa": {"acc_norm": 0.5136017410228509, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5136017410228509], "acc": 0.5212187159956474, "acc_ci": 0.0, "acc_runs": [0.5212187159956474], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011661506839823792, "centered_acc": 0.027203482045701888, "centered_acc_ci": 0.0, "centered_acc_runs": [0.027203482045701888], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2680976430976431, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2680976430976431], "acc": 0.2727272727272727, "acc_ci": 0.0, "acc_runs": [0.2727272727272727], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009089526578213825, "centered_acc": 0.024130190796857482, "centered_acc_ci": 0.0, "centered_acc_runs": [0.024130190796857482], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2636518771331058, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2636518771331058], "acc": 0.20563139931740615, "acc_ci": 0.0, "acc_runs": [0.20563139931740615], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012875929151296968, "centered_acc": 0.018202502844141044, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018202502844141044], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.174, "acc_ci": 0.0, "acc_runs": [0.174], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4980268350434096, "acc_ci": 0.0, "acc_runs": [0.4980268350434096], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052376259225643, "centered_acc": -0.003946329913180779, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.003946329913180779], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 19251662182.07262, "perplexity_ci": 0.0, "perplexity_runs": [19251662182.07262], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.17, "acc_norm_ci": 0.0, "acc_norm_runs": [0.17], "acc": 0.04, "acc_ci": 0.0, "acc_runs": [0.04], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011884495834541793, "centered_acc": -0.10666666666666665, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10666666666666665], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29241468109786645, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2879543148709258, "mean_acc_ci": 0.0, "mean_centered_acc": -0.016924384840846057, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29241468109786645], "mean_acc_runs": [0.2879543148709258], "mean_centered_acc_runs": [-0.016924384840846057], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-7488", "checkpoint": "checkpoint-7488", "global_step": 7488, "tokens_seen": 7851737088, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.252141007767377, "acc_norm_ci": 0.0, "acc_norm_runs": [0.252141007767377], "acc": 0.2526389165504879, "acc_ci": 0.0, "acc_runs": [0.2526389165504879], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004333543083293202, "centered_acc": 0.0028546770231693643, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0028546770231693643], "ci": 0.0}, "piqa": {"acc_norm": 0.5136017410228509, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5136017410228509], "acc": 0.528835690968444, "acc_ci": 0.0, "acc_runs": [0.528835690968444], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011661506839823792, "centered_acc": 0.027203482045701888, "centered_acc_ci": 0.0, "centered_acc_runs": [0.027203482045701888], "ci": 0.0}, "arc_easy": {"acc_norm": 0.25925925925925924, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25925925925925924], "acc": 0.26262626262626265, "acc_ci": 0.0, "acc_runs": [0.26262626262626265], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008992251535805688, "centered_acc": 0.01234567901234566, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01234567901234566], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2568259385665529, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2568259385665529], "acc": 0.22098976109215018, "acc_ci": 0.0, "acc_runs": [0.22098976109215018], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012766923794116735, "centered_acc": 0.009101251422070522, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009101251422070522], "ci": 0.0}, "openbookqa": {"acc_norm": 0.288, "acc_norm_ci": 0.0, "acc_norm_runs": [0.288], "acc": 0.184, "acc_ci": 0.0, "acc_runs": [0.184], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020271503835075255, "centered_acc": 0.05066666666666664, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05066666666666664], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5185477505919495, "acc_ci": 0.0, "acc_runs": [0.5185477505919495], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014042813708888347, "centered_acc": 0.03709550118389893, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03709550118389893], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.53, "acc_ci": 0.0, "acc_runs": [0.53], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05016135580465919, "centered_acc": 0.06000000000000005, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06000000000000005], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 25041298117.38943, "perplexity_ci": 0.0, "perplexity_runs": [25041298117.38943], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.175, "acc_norm_ci": 0.0, "acc_norm_runs": [0.175], "acc": 0.039, "acc_ci": 0.0, "acc_runs": [0.039], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012021627157731998, "centered_acc": -0.10000000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10000000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29080465776934, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29149258436029946, "mean_acc_ci": 0.0, "mean_centered_acc": -0.01601753545541626, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29080465776934], "mean_acc_runs": [0.29149258436029946], "mean_centered_acc_runs": [-0.01601753545541626], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-7776", "checkpoint": "checkpoint-7776", "global_step": 7776, "tokens_seen": 8153726976, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.253734315873332, "acc_norm_ci": 0.0, "acc_norm_runs": [0.253734315873332], "acc": 0.25253933479386576, "acc_ci": 0.0, "acc_runs": [0.25253933479386576], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004342580277662561, "centered_acc": 0.004979087831109341, "centered_acc_ci": 0.0, "centered_acc_runs": [0.004979087831109341], "ci": 0.0}, "piqa": {"acc_norm": 0.5076169749727966, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5076169749727966], "acc": 0.5228509249183896, "acc_ci": 0.0, "acc_runs": [0.5228509249183896], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011664470424044771, "centered_acc": 0.015233949945593128, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015233949945593128], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2622053872053872, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2622053872053872], "acc": 0.2638888888888889, "acc_ci": 0.0, "acc_runs": [0.2638888888888889], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009025197991724937, "centered_acc": 0.01627384960718296, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01627384960718296], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2508532423208191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2508532423208191], "acc": 0.20136518771331058, "acc_ci": 0.0, "acc_runs": [0.20136518771331058], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012668198621315347, "centered_acc": 0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.176, "acc_ci": 0.0, "acc_runs": [0.176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5067087608524072, "acc_ci": 0.0, "acc_runs": [0.5067087608524072], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051220692330247, "centered_acc": 0.013417521704814472, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013417521704814472], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.52, "acc_ci": 0.0, "acc_runs": [0.52], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05021167315686783, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 26042077270.405506, "perplexity_ci": 0.0, "perplexity_runs": [26042077270.405506], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.172, "acc_norm_ci": 0.0, "acc_norm_runs": [0.172], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011939788882495308, "centered_acc": -0.10400000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10400000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.28806832006205585, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28656405589405626, "mean_acc_ci": 0.0, "mean_centered_acc": -0.023746260474386, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.28806832006205585], "mean_acc_runs": [0.28656405589405626], "mean_centered_acc_runs": [-0.023746260474386], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-8064", "checkpoint": "checkpoint-8064", "global_step": 8064, "tokens_seen": 8455716864, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25124477195777734, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25124477195777734], "acc": 0.24955188209520016, "acc_ci": 0.0, "acc_runs": [0.24955188209520016], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004328425700998576, "centered_acc": 0.0016596959437031138, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0016596959437031138], "ci": 0.0}, "piqa": {"acc_norm": 0.5108813928182807, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5108813928182807], "acc": 0.5212187159956474, "acc_ci": 0.0, "acc_runs": [0.5212187159956474], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011663061261117717, "centered_acc": 0.02176278563656142, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02176278563656142], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2617845117845118, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2617845117845118], "acc": 0.27146464646464646, "acc_ci": 0.0, "acc_runs": [0.27146464646464646], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009020523527210056, "centered_acc": 0.015712682379349047, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015712682379349047], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2568259385665529, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2568259385665529], "acc": 0.22013651877133106, "acc_ci": 0.0, "acc_runs": [0.22013651877133106], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012766923794116735, "centered_acc": 0.009101251422070522, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009101251422070522], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.182, "acc_ci": 0.0, "acc_runs": [0.182], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4972375690607735, "acc_ci": 0.0, "acc_runs": [0.4972375690607735], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052271211616401, "centered_acc": -0.0055248618784530246, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0055248618784530246], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.52, "acc_ci": 0.0, "acc_runs": [0.52], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05021167315686783, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 34040228153.797745, "perplexity_ci": 0.0, "perplexity_runs": [34040228153.797745], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.169, "acc_norm_ci": 0.0, "acc_norm_runs": [0.169], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011856625977890152, "centered_acc": -0.10799999999999998, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10799999999999998], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2886227691878538, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28828967941612993, "mean_acc_ci": 0.0, "mean_centered_acc": -0.025116317364744623, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2886227691878538], "mean_acc_runs": [0.28828967941612993], "mean_centered_acc_runs": [-0.025116317364744623], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-8352", "checkpoint": "checkpoint-8352", "global_step": 8352, "tokens_seen": 8757706752, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2523401712806214, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2523401712806214], "acc": 0.24885480979884486, "acc_ci": 0.0, "acc_runs": [0.24885480979884486], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004334676952703991, "centered_acc": 0.0031202283741618895, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0031202283741618895], "ci": 0.0}, "piqa": {"acc_norm": 0.514689880304679, "acc_norm_ci": 0.0, "acc_norm_runs": [0.514689880304679], "acc": 0.5195865070729053, "acc_ci": 0.0, "acc_runs": [0.5195865070729053], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011660788281735552, "centered_acc": 0.029379760609357986, "centered_acc_ci": 0.0, "centered_acc_runs": [0.029379760609357986], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26725589225589225, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26725589225589225], "acc": 0.2638888888888889, "acc_ci": 0.0, "acc_runs": [0.2638888888888889], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009080463246017426, "centered_acc": 0.023007856341189663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023007856341189663], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.257679180887372, "acc_norm_ci": 0.0, "acc_norm_runs": [0.257679180887372], "acc": 0.21928327645051193, "acc_ci": 0.0, "acc_runs": [0.21928327645051193], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012780770562768334, "centered_acc": 0.010238907849829356, "centered_acc_ci": 0.0, "centered_acc_runs": [0.010238907849829356], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.18, "acc_ci": 0.0, "acc_runs": [0.18], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5035516969218626, "acc_ci": 0.0, "acc_runs": [0.5035516969218626], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052131146915852, "centered_acc": 0.00710339384372527, "centered_acc_ci": 0.0, "centered_acc_runs": [0.00710339384372527], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.53, "acc_ci": 0.0, "acc_runs": [0.53], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05016135580465919, "centered_acc": 0.06000000000000005, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06000000000000005], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 49483206131.061844, "perplexity_ci": 0.0, "perplexity_runs": [49483206131.061844], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.177, "acc_norm_ci": 0.0, "acc_norm_runs": [0.177], "acc": 0.049, "acc_ci": 0.0, "acc_runs": [0.049], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012075463420375087, "centered_acc": -0.09733333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09733333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2914941874547608, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2892452640906714, "mean_acc_ci": 0.0, "mean_centered_acc": -0.018656473640852052, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2914941874547608], "mean_acc_runs": [0.2892452640906714], "mean_centered_acc_runs": [-0.018656473640852052], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/baseline-fulle-lm", "run": "baseline-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/baseline-fulle-lm/checkpoint-8640", "checkpoint": "checkpoint-8640", "global_step": 8640, "tokens_seen": 9059696640, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25184226249751046, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25184226249751046], "acc": 0.2513443537143995, "acc_ci": 0.0, "acc_runs": [0.2513443537143995], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004331840012787794, "centered_acc": 0.0024563499966806144, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0024563499966806144], "ci": 0.0}, "piqa": {"acc_norm": 0.5070729053318824, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5070729053318824], "acc": 0.5184983677910773, "acc_ci": 0.0, "acc_runs": [0.5184983677910773], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011664656918146136, "centered_acc": 0.014145810663764857, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014145810663764857], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26262626262626265, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26262626262626265], "acc": 0.26052188552188554, "acc_ci": 0.0, "acc_runs": [0.26052188552188554], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009029861776763957, "centered_acc": 0.01683501683501687, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01683501683501687], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24914675767918087, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24914675767918087], "acc": 0.21160409556313994, "acc_ci": 0.0, "acc_runs": [0.21160409556313994], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012639407111926505, "centered_acc": -0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.264, "acc_norm_ci": 0.0, "acc_norm_runs": [0.264], "acc": 0.168, "acc_ci": 0.0, "acc_runs": [0.168], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01973288558592215, "centered_acc": 0.018666666666666682, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018666666666666682], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5130228887134964, "acc_ci": 0.0, "acc_runs": [0.5130228887134964], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014047718393997596, "centered_acc": 0.026045777426992878, "centered_acc_ci": 0.0, "centered_acc_runs": [0.026045777426992878], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.52, "acc_ci": 0.0, "acc_runs": [0.52], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05021167315686783, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 66809449869.6633, "perplexity_ci": 0.0, "perplexity_runs": [66809449869.6633], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.172, "acc_norm_ci": 0.0, "acc_norm_runs": [0.172], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011939788882495308, "centered_acc": -0.10400000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10400000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.28444803135580604, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2867279053077699, "mean_acc_ci": 0.0, "mean_centered_acc": -0.02560145681013737, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.28444803135580604], "mean_acc_runs": [0.2867279053077699], "mean_centered_acc_runs": [-0.02560145681013737], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-10080.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-10080.json new file mode 100644 index 0000000000000000000000000000000000000000..94fa3dfd0125f58134b9d2b062961292186785b2 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-10080.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-10080", + "checkpoint": "checkpoint-10080", + "global_step": 10080, + "eval_loss": 4.913107395172119, + "metrics": { + "eval_loss": 4.913107395172119, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 130.1141, + "eval_samples_per_second": 18.768, + "eval_steps_per_second": 2.352, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-10368.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-10368.json new file mode 100644 index 0000000000000000000000000000000000000000..ddf51e468cdc9a69193331b830fec6a466c267b2 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-10368.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-10368", + "checkpoint": "checkpoint-10368", + "global_step": 10368, + "eval_loss": 4.950941562652588, + "metrics": { + "eval_loss": 4.950941562652588, + "eval_model_preparation_time": 0.004, + "eval_runtime": 130.2393, + "eval_samples_per_second": 18.75, + "eval_steps_per_second": 2.35, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-10656.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-10656.json new file mode 100644 index 0000000000000000000000000000000000000000..6e972c3f6acfd1a9e0b837f64c8893b47bc266ff --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-10656.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-10656", + "checkpoint": "checkpoint-10656", + "global_step": 10656, + "eval_loss": 4.9802350997924805, + "metrics": { + "eval_loss": 4.9802350997924805, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 130.2406, + "eval_samples_per_second": 18.75, + "eval_steps_per_second": 2.349, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-1152.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-1152.json new file mode 100644 index 0000000000000000000000000000000000000000..e4527b67b2bdf5bc341ed9066a1b67cd1a9631eb --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-1152.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-1152", + "checkpoint": "checkpoint-1152", + "global_step": 1152, + "eval_loss": 5.748093128204346, + "metrics": { + "eval_loss": 5.748093128204346, + "eval_model_preparation_time": 0.0039, + "eval_runtime": 129.3962, + "eval_samples_per_second": 18.872, + "eval_steps_per_second": 2.365, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-1440.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-1440.json new file mode 100644 index 0000000000000000000000000000000000000000..2bced4cd768c047b1915eb6eddc90eb8101ee606 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-1440.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-1440", + "checkpoint": "checkpoint-1440", + "global_step": 1440, + "eval_loss": 6.151500225067139, + "metrics": { + "eval_loss": 6.151500225067139, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.2395, + "eval_samples_per_second": 18.895, + "eval_steps_per_second": 2.368, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-1728.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-1728.json new file mode 100644 index 0000000000000000000000000000000000000000..0934ae8fb4df920bc7df6028ef75d2856ebce307 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-1728.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-1728", + "checkpoint": "checkpoint-1728", + "global_step": 1728, + "eval_loss": 5.215778827667236, + "metrics": { + "eval_loss": 5.215778827667236, + "eval_model_preparation_time": 0.0038, + "eval_runtime": 129.1815, + "eval_samples_per_second": 18.904, + "eval_steps_per_second": 2.369, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-2016.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-2016.json new file mode 100644 index 0000000000000000000000000000000000000000..d1beeb773fec2ea9f2b43f2d04aa36546e2b3aea --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-2016.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2016", + "checkpoint": "checkpoint-2016", + "global_step": 2016, + "eval_loss": 4.573704242706299, + "metrics": { + "eval_loss": 4.573704242706299, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.4132, + "eval_samples_per_second": 18.87, + "eval_steps_per_second": 2.365, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-2304.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-2304.json new file mode 100644 index 0000000000000000000000000000000000000000..42042c340549b3d83ad656c765576cb28d90a0d0 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-2304.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2304", + "checkpoint": "checkpoint-2304", + "global_step": 2304, + "eval_loss": 4.476961612701416, + "metrics": { + "eval_loss": 4.476961612701416, + "eval_model_preparation_time": 0.0038, + "eval_runtime": 130.0087, + "eval_samples_per_second": 18.783, + "eval_steps_per_second": 2.354, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-2592.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-2592.json new file mode 100644 index 0000000000000000000000000000000000000000..efdb61d9b2e919728b3827777fe5327d704791c4 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-2592.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2592", + "checkpoint": "checkpoint-2592", + "global_step": 2592, + "eval_loss": 4.093716144561768, + "metrics": { + "eval_loss": 4.093716144561768, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 130.1606, + "eval_samples_per_second": 18.761, + "eval_steps_per_second": 2.351, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-288.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-288.json new file mode 100644 index 0000000000000000000000000000000000000000..ee27b41035b7fd5cf4cb7c0abc7235c2ffaf2164 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-288.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-288", + "checkpoint": "checkpoint-288", + "global_step": 288, + "eval_loss": 7.010690212249756, + "metrics": { + "eval_loss": 7.010690212249756, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.6306, + "eval_samples_per_second": 18.838, + "eval_steps_per_second": 2.361, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-2880.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-2880.json new file mode 100644 index 0000000000000000000000000000000000000000..371d3cd89b6343ddc1bec591b3a7b99d2d544f3a --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-2880.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2880", + "checkpoint": "checkpoint-2880", + "global_step": 2880, + "eval_loss": 4.068608283996582, + "metrics": { + "eval_loss": 4.068608283996582, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 130.1948, + "eval_samples_per_second": 18.757, + "eval_steps_per_second": 2.35, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-3168.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-3168.json new file mode 100644 index 0000000000000000000000000000000000000000..8e9c4aaf0877d457700a41bbe2f8ce1e91f76709 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-3168.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-3168", + "checkpoint": "checkpoint-3168", + "global_step": 3168, + "eval_loss": 4.0662760734558105, + "metrics": { + "eval_loss": 4.0662760734558105, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 130.2039, + "eval_samples_per_second": 18.755, + "eval_steps_per_second": 2.35, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-3456.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-3456.json new file mode 100644 index 0000000000000000000000000000000000000000..813350c88204d2abad9d76d6cc45cc2d88b9f10b --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-3456.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-3456", + "checkpoint": "checkpoint-3456", + "global_step": 3456, + "eval_loss": 4.074501991271973, + "metrics": { + "eval_loss": 4.074501991271973, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.8404, + "eval_samples_per_second": 18.808, + "eval_steps_per_second": 2.357, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-3744.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-3744.json new file mode 100644 index 0000000000000000000000000000000000000000..ed5e7c99e2fc43da604afd8559331271490ae6e6 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-3744.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-3744", + "checkpoint": "checkpoint-3744", + "global_step": 3744, + "eval_loss": 4.087250709533691, + "metrics": { + "eval_loss": 4.087250709533691, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.5216, + "eval_samples_per_second": 18.854, + "eval_steps_per_second": 2.363, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-4032.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-4032.json new file mode 100644 index 0000000000000000000000000000000000000000..6a5dd7e08daefeeb22aab8e1dc9cd9ae6fb351cf --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-4032.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4032", + "checkpoint": "checkpoint-4032", + "global_step": 4032, + "eval_loss": 4.020753860473633, + "metrics": { + "eval_loss": 4.020753860473633, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.3458, + "eval_samples_per_second": 18.88, + "eval_steps_per_second": 2.366, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-4320.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-4320.json new file mode 100644 index 0000000000000000000000000000000000000000..4a88bc8b2e2d9d5549250c653a655a89b03ac293 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-4320.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4320", + "checkpoint": "checkpoint-4320", + "global_step": 4320, + "eval_loss": 4.043672561645508, + "metrics": { + "eval_loss": 4.043672561645508, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.4758, + "eval_samples_per_second": 18.861, + "eval_steps_per_second": 2.363, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-4608.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-4608.json new file mode 100644 index 0000000000000000000000000000000000000000..48b6c8078a74faff6093757d6a818fca8e19355c --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-4608.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4608", + "checkpoint": "checkpoint-4608", + "global_step": 4608, + "eval_loss": 4.065646648406982, + "metrics": { + "eval_loss": 4.065646648406982, + "eval_model_preparation_time": 0.004, + "eval_runtime": 129.9088, + "eval_samples_per_second": 18.798, + "eval_steps_per_second": 2.355, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-4896.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-4896.json new file mode 100644 index 0000000000000000000000000000000000000000..56930421e80b08ae3c0ff77a83f785cb3d86b63f --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-4896.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4896", + "checkpoint": "checkpoint-4896", + "global_step": 4896, + "eval_loss": 4.111476421356201, + "metrics": { + "eval_loss": 4.111476421356201, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 130.1422, + "eval_samples_per_second": 18.764, + "eval_steps_per_second": 2.351, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-5184.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-5184.json new file mode 100644 index 0000000000000000000000000000000000000000..4e7f267c2aba97e0eaf8a398508e0efb5495854b --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-5184.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-5184", + "checkpoint": "checkpoint-5184", + "global_step": 5184, + "eval_loss": 4.137632846832275, + "metrics": { + "eval_loss": 4.137632846832275, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 130.0875, + "eval_samples_per_second": 18.772, + "eval_steps_per_second": 2.352, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-5472.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-5472.json new file mode 100644 index 0000000000000000000000000000000000000000..c5bc35a1cd5f9f069def26bdd95c17c2a5b5c9c7 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-5472.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-5472", + "checkpoint": "checkpoint-5472", + "global_step": 5472, + "eval_loss": 4.172383785247803, + "metrics": { + "eval_loss": 4.172383785247803, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 130.2369, + "eval_samples_per_second": 18.75, + "eval_steps_per_second": 2.35, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-576.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-576.json new file mode 100644 index 0000000000000000000000000000000000000000..672b18f56b760a2214d82df500fad14ab5220a0a --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-576.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-576", + "checkpoint": "checkpoint-576", + "global_step": 576, + "eval_loss": 6.8527302742004395, + "metrics": { + "eval_loss": 6.8527302742004395, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 130.0356, + "eval_samples_per_second": 18.779, + "eval_steps_per_second": 2.353, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-5760.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-5760.json new file mode 100644 index 0000000000000000000000000000000000000000..134fd2171869699dcaf506b43233071e6bfab21c --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-5760.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-5760", + "checkpoint": "checkpoint-5760", + "global_step": 5760, + "eval_loss": 4.216954231262207, + "metrics": { + "eval_loss": 4.216954231262207, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 130.2182, + "eval_samples_per_second": 18.753, + "eval_steps_per_second": 2.35, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-6048.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-6048.json new file mode 100644 index 0000000000000000000000000000000000000000..3b4b0ec6b33230a9d3a3693c3f27463f1fbec4d0 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-6048.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-6048", + "checkpoint": "checkpoint-6048", + "global_step": 6048, + "eval_loss": 4.262066841125488, + "metrics": { + "eval_loss": 4.262066841125488, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 130.2397, + "eval_samples_per_second": 18.75, + "eval_steps_per_second": 2.35, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-6336.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-6336.json new file mode 100644 index 0000000000000000000000000000000000000000..30e55fbce6201a885ed0469e413d4bb90e33c8df --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-6336.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-6336", + "checkpoint": "checkpoint-6336", + "global_step": 6336, + "eval_loss": 4.308366775512695, + "metrics": { + "eval_loss": 4.308366775512695, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.7853, + "eval_samples_per_second": 18.816, + "eval_steps_per_second": 2.358, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-6624.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-6624.json new file mode 100644 index 0000000000000000000000000000000000000000..070080a0b42aac02dbcf80f447f2faac36c720f8 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-6624.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-6624", + "checkpoint": "checkpoint-6624", + "global_step": 6624, + "eval_loss": 4.354447841644287, + "metrics": { + "eval_loss": 4.354447841644287, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.4397, + "eval_samples_per_second": 18.866, + "eval_steps_per_second": 2.364, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-6912.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-6912.json new file mode 100644 index 0000000000000000000000000000000000000000..dea38174a23e0ea7d9c86bc2b366dafec3345993 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-6912.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-6912", + "checkpoint": "checkpoint-6912", + "global_step": 6912, + "eval_loss": 4.413780212402344, + "metrics": { + "eval_loss": 4.413780212402344, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.4671, + "eval_samples_per_second": 18.862, + "eval_steps_per_second": 2.364, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-72.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-72.json new file mode 100644 index 0000000000000000000000000000000000000000..cff9a9223896eac61bc5a55a9e60cd7b75647fab --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-72.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-72", + "checkpoint": "checkpoint-72", + "global_step": 72, + "eval_loss": 7.52717399597168, + "metrics": { + "eval_loss": 7.52717399597168, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 127.3709, + "eval_samples_per_second": 19.172, + "eval_steps_per_second": 2.402, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-7200.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-7200.json new file mode 100644 index 0000000000000000000000000000000000000000..69de352d788a355b5ca9a17f5ccc5d63e1075c91 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-7200.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-7200", + "checkpoint": "checkpoint-7200", + "global_step": 7200, + "eval_loss": 4.455929279327393, + "metrics": { + "eval_loss": 4.455929279327393, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.8303, + "eval_samples_per_second": 18.809, + "eval_steps_per_second": 2.357, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-7488.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-7488.json new file mode 100644 index 0000000000000000000000000000000000000000..682b72f5c6fe55a6e1da8dc46dcfdb75db33b9e6 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-7488.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-7488", + "checkpoint": "checkpoint-7488", + "global_step": 7488, + "eval_loss": 4.50410795211792, + "metrics": { + "eval_loss": 4.50410795211792, + "eval_model_preparation_time": 0.0038, + "eval_runtime": 130.1386, + "eval_samples_per_second": 18.765, + "eval_steps_per_second": 2.351, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-7776.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-7776.json new file mode 100644 index 0000000000000000000000000000000000000000..63f69a7bdcb88cd656f39e4a451b0547c1b507bc --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-7776.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-7776", + "checkpoint": "checkpoint-7776", + "global_step": 7776, + "eval_loss": 4.549941062927246, + "metrics": { + "eval_loss": 4.549941062927246, + "eval_model_preparation_time": 0.0039, + "eval_runtime": 130.2535, + "eval_samples_per_second": 18.748, + "eval_steps_per_second": 2.349, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-8064.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-8064.json new file mode 100644 index 0000000000000000000000000000000000000000..2160afb82e824bd3e85a09113468f2f6ce15fe01 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-8064.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-8064", + "checkpoint": "checkpoint-8064", + "global_step": 8064, + "eval_loss": 4.480454921722412, + "metrics": { + "eval_loss": 4.480454921722412, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 130.24, + "eval_samples_per_second": 18.75, + "eval_steps_per_second": 2.35, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-8352.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-8352.json new file mode 100644 index 0000000000000000000000000000000000000000..245b77aa7cecfb35ce939c1ddc4da2755666ea6b --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-8352.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-8352", + "checkpoint": "checkpoint-8352", + "global_step": 8352, + "eval_loss": 4.603067874908447, + "metrics": { + "eval_loss": 4.603067874908447, + "eval_model_preparation_time": 0.0039, + "eval_runtime": 130.3151, + "eval_samples_per_second": 18.739, + "eval_steps_per_second": 2.348, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-864.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-864.json new file mode 100644 index 0000000000000000000000000000000000000000..70cc696223ba6b8c97906be837be9c6aa32b7728 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-864.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-864", + "checkpoint": "checkpoint-864", + "global_step": 864, + "eval_loss": 5.2929229736328125, + "metrics": { + "eval_loss": 5.2929229736328125, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 130.0278, + "eval_samples_per_second": 18.781, + "eval_steps_per_second": 2.353, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-8640.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-8640.json new file mode 100644 index 0000000000000000000000000000000000000000..7cdc265bff93f59d8ac0cbea5118216a069f663e --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-8640.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-8640", + "checkpoint": "checkpoint-8640", + "global_step": 8640, + "eval_loss": 4.671885967254639, + "metrics": { + "eval_loss": 4.671885967254639, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.9916, + "eval_samples_per_second": 18.786, + "eval_steps_per_second": 2.354, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-8928.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-8928.json new file mode 100644 index 0000000000000000000000000000000000000000..20d3448a3b3122ebfb78e176feb2fad06946159f --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-8928.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-8928", + "checkpoint": "checkpoint-8928", + "global_step": 8928, + "eval_loss": 4.700623989105225, + "metrics": { + "eval_loss": 4.700623989105225, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.6463, + "eval_samples_per_second": 18.836, + "eval_steps_per_second": 2.36, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-9216.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-9216.json new file mode 100644 index 0000000000000000000000000000000000000000..04fc0d330c16bb0c55094ad5361c8ccb6fbbc6cd --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-9216.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-9216", + "checkpoint": "checkpoint-9216", + "global_step": 9216, + "eval_loss": 4.762052536010742, + "metrics": { + "eval_loss": 4.762052536010742, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.4329, + "eval_samples_per_second": 18.867, + "eval_steps_per_second": 2.364, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-9504.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-9504.json new file mode 100644 index 0000000000000000000000000000000000000000..fa40efb9617e041db4d5ee85944e7e8101ee8c12 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-9504.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-9504", + "checkpoint": "checkpoint-9504", + "global_step": 9504, + "eval_loss": 4.8133416175842285, + "metrics": { + "eval_loss": 4.8133416175842285, + "eval_model_preparation_time": 0.004, + "eval_runtime": 129.5296, + "eval_samples_per_second": 18.853, + "eval_steps_per_second": 2.362, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-9792.json b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-9792.json new file mode 100644 index 0000000000000000000000000000000000000000..f0fe4400417c3f0c1acb761ea14659b08fd32790 --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/checkpoint-9792.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-9792", + "checkpoint": "checkpoint-9792", + "global_step": 9792, + "eval_loss": 4.863399028778076, + "metrics": { + "eval_loss": 4.863399028778076, + "eval_model_preparation_time": 0.0043, + "eval_runtime": 130.0049, + "eval_samples_per_second": 18.784, + "eval_steps_per_second": 2.354, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-exp-fulle-lm/l2r_n1_eval/summary.jsonl b/results/i5-exp-fulle-lm/l2r_n1_eval/summary.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..98ffd99a6252a8bb31d97853051dd630ed579b7f --- /dev/null +++ b/results/i5-exp-fulle-lm/l2r_n1_eval/summary.jsonl @@ -0,0 +1,38 @@ +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-72", "checkpoint": "checkpoint-72", "global_step": 72, "eval_loss": 7.52717399597168, "metrics": {"eval_loss": 7.52717399597168, "eval_model_preparation_time": 0.0034, "eval_runtime": 127.3709, "eval_samples_per_second": 19.172, "eval_steps_per_second": 2.402, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-288", "checkpoint": "checkpoint-288", "global_step": 288, "eval_loss": 7.010690212249756, "metrics": {"eval_loss": 7.010690212249756, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.6306, "eval_samples_per_second": 18.838, "eval_steps_per_second": 2.361, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-576", "checkpoint": "checkpoint-576", "global_step": 576, "eval_loss": 6.8527302742004395, "metrics": {"eval_loss": 6.8527302742004395, "eval_model_preparation_time": 0.0037, "eval_runtime": 130.0356, "eval_samples_per_second": 18.779, "eval_steps_per_second": 2.353, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-864", "checkpoint": "checkpoint-864", "global_step": 864, "eval_loss": 5.2929229736328125, "metrics": {"eval_loss": 5.2929229736328125, "eval_model_preparation_time": 0.0035, "eval_runtime": 130.0278, "eval_samples_per_second": 18.781, "eval_steps_per_second": 2.353, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-1152", "checkpoint": "checkpoint-1152", "global_step": 1152, "eval_loss": 5.748093128204346, "metrics": {"eval_loss": 5.748093128204346, "eval_model_preparation_time": 0.0039, "eval_runtime": 129.3962, "eval_samples_per_second": 18.872, "eval_steps_per_second": 2.365, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-1440", "checkpoint": "checkpoint-1440", "global_step": 1440, "eval_loss": 6.151500225067139, "metrics": {"eval_loss": 6.151500225067139, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.2395, "eval_samples_per_second": 18.895, "eval_steps_per_second": 2.368, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-1728", "checkpoint": "checkpoint-1728", "global_step": 1728, "eval_loss": 5.215778827667236, "metrics": {"eval_loss": 5.215778827667236, "eval_model_preparation_time": 0.0038, "eval_runtime": 129.1815, "eval_samples_per_second": 18.904, "eval_steps_per_second": 2.369, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2016", "checkpoint": "checkpoint-2016", "global_step": 2016, "eval_loss": 4.573704242706299, "metrics": {"eval_loss": 4.573704242706299, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.4132, "eval_samples_per_second": 18.87, "eval_steps_per_second": 2.365, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2304", "checkpoint": "checkpoint-2304", "global_step": 2304, "eval_loss": 4.476961612701416, "metrics": {"eval_loss": 4.476961612701416, "eval_model_preparation_time": 0.0038, "eval_runtime": 130.0087, "eval_samples_per_second": 18.783, "eval_steps_per_second": 2.354, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2592", "checkpoint": "checkpoint-2592", "global_step": 2592, "eval_loss": 4.093716144561768, "metrics": {"eval_loss": 4.093716144561768, "eval_model_preparation_time": 0.0036, "eval_runtime": 130.1606, "eval_samples_per_second": 18.761, "eval_steps_per_second": 2.351, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2880", "checkpoint": "checkpoint-2880", "global_step": 2880, "eval_loss": 4.068608283996582, "metrics": {"eval_loss": 4.068608283996582, "eval_model_preparation_time": 0.0034, "eval_runtime": 130.1948, "eval_samples_per_second": 18.757, "eval_steps_per_second": 2.35, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-3168", "checkpoint": "checkpoint-3168", "global_step": 3168, "eval_loss": 4.0662760734558105, "metrics": {"eval_loss": 4.0662760734558105, "eval_model_preparation_time": 0.0034, "eval_runtime": 130.2039, "eval_samples_per_second": 18.755, "eval_steps_per_second": 2.35, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-3456", "checkpoint": "checkpoint-3456", "global_step": 3456, "eval_loss": 4.074501991271973, "metrics": {"eval_loss": 4.074501991271973, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.8404, "eval_samples_per_second": 18.808, "eval_steps_per_second": 2.357, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-3744", "checkpoint": "checkpoint-3744", "global_step": 3744, "eval_loss": 4.087250709533691, "metrics": {"eval_loss": 4.087250709533691, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.5216, "eval_samples_per_second": 18.854, "eval_steps_per_second": 2.363, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4032", "checkpoint": "checkpoint-4032", "global_step": 4032, "eval_loss": 4.020753860473633, "metrics": {"eval_loss": 4.020753860473633, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.3458, "eval_samples_per_second": 18.88, "eval_steps_per_second": 2.366, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4320", "checkpoint": "checkpoint-4320", "global_step": 4320, "eval_loss": 4.043672561645508, "metrics": {"eval_loss": 4.043672561645508, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.4758, "eval_samples_per_second": 18.861, "eval_steps_per_second": 2.363, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4608", "checkpoint": "checkpoint-4608", "global_step": 4608, "eval_loss": 4.065646648406982, "metrics": {"eval_loss": 4.065646648406982, "eval_model_preparation_time": 0.004, "eval_runtime": 129.9088, "eval_samples_per_second": 18.798, "eval_steps_per_second": 2.355, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4896", "checkpoint": "checkpoint-4896", "global_step": 4896, "eval_loss": 4.111476421356201, "metrics": {"eval_loss": 4.111476421356201, "eval_model_preparation_time": 0.0036, "eval_runtime": 130.1422, "eval_samples_per_second": 18.764, "eval_steps_per_second": 2.351, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-5184", "checkpoint": "checkpoint-5184", "global_step": 5184, "eval_loss": 4.137632846832275, "metrics": {"eval_loss": 4.137632846832275, "eval_model_preparation_time": 0.0034, "eval_runtime": 130.0875, "eval_samples_per_second": 18.772, "eval_steps_per_second": 2.352, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-5472", "checkpoint": "checkpoint-5472", "global_step": 5472, "eval_loss": 4.172383785247803, "metrics": {"eval_loss": 4.172383785247803, "eval_model_preparation_time": 0.0037, "eval_runtime": 130.2369, "eval_samples_per_second": 18.75, "eval_steps_per_second": 2.35, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-5760", "checkpoint": "checkpoint-5760", "global_step": 5760, "eval_loss": 4.216954231262207, "metrics": {"eval_loss": 4.216954231262207, "eval_model_preparation_time": 0.0037, "eval_runtime": 130.2182, "eval_samples_per_second": 18.753, "eval_steps_per_second": 2.35, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-6048", "checkpoint": "checkpoint-6048", "global_step": 6048, "eval_loss": 4.262066841125488, "metrics": {"eval_loss": 4.262066841125488, "eval_model_preparation_time": 0.0034, "eval_runtime": 130.2397, "eval_samples_per_second": 18.75, "eval_steps_per_second": 2.35, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-6336", "checkpoint": "checkpoint-6336", "global_step": 6336, "eval_loss": 4.308366775512695, "metrics": {"eval_loss": 4.308366775512695, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.7853, "eval_samples_per_second": 18.816, "eval_steps_per_second": 2.358, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-6624", "checkpoint": "checkpoint-6624", "global_step": 6624, "eval_loss": 4.354447841644287, "metrics": {"eval_loss": 4.354447841644287, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.4397, "eval_samples_per_second": 18.866, "eval_steps_per_second": 2.364, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-6912", "checkpoint": "checkpoint-6912", "global_step": 6912, "eval_loss": 4.413780212402344, "metrics": {"eval_loss": 4.413780212402344, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.4671, "eval_samples_per_second": 18.862, "eval_steps_per_second": 2.364, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-7200", "checkpoint": "checkpoint-7200", "global_step": 7200, "eval_loss": 4.455929279327393, "metrics": {"eval_loss": 4.455929279327393, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.8303, "eval_samples_per_second": 18.809, "eval_steps_per_second": 2.357, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-7488", "checkpoint": "checkpoint-7488", "global_step": 7488, "eval_loss": 4.50410795211792, "metrics": {"eval_loss": 4.50410795211792, "eval_model_preparation_time": 0.0038, "eval_runtime": 130.1386, "eval_samples_per_second": 18.765, "eval_steps_per_second": 2.351, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-7776", "checkpoint": "checkpoint-7776", "global_step": 7776, "eval_loss": 4.549941062927246, "metrics": {"eval_loss": 4.549941062927246, "eval_model_preparation_time": 0.0039, "eval_runtime": 130.2535, "eval_samples_per_second": 18.748, "eval_steps_per_second": 2.349, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-8064", "checkpoint": "checkpoint-8064", "global_step": 8064, "eval_loss": 4.480454921722412, "metrics": {"eval_loss": 4.480454921722412, "eval_model_preparation_time": 0.0036, "eval_runtime": 130.24, "eval_samples_per_second": 18.75, "eval_steps_per_second": 2.35, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-8352", "checkpoint": "checkpoint-8352", "global_step": 8352, "eval_loss": 4.603067874908447, "metrics": {"eval_loss": 4.603067874908447, "eval_model_preparation_time": 0.0039, "eval_runtime": 130.3151, "eval_samples_per_second": 18.739, "eval_steps_per_second": 2.348, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-8640", "checkpoint": "checkpoint-8640", "global_step": 8640, "eval_loss": 4.671885967254639, "metrics": {"eval_loss": 4.671885967254639, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.9916, "eval_samples_per_second": 18.786, "eval_steps_per_second": 2.354, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-8928", "checkpoint": "checkpoint-8928", "global_step": 8928, "eval_loss": 4.700623989105225, "metrics": {"eval_loss": 4.700623989105225, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.6463, "eval_samples_per_second": 18.836, "eval_steps_per_second": 2.36, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-9216", "checkpoint": "checkpoint-9216", "global_step": 9216, "eval_loss": 4.762052536010742, "metrics": {"eval_loss": 4.762052536010742, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.4329, "eval_samples_per_second": 18.867, "eval_steps_per_second": 2.364, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-9504", "checkpoint": "checkpoint-9504", "global_step": 9504, "eval_loss": 4.8133416175842285, "metrics": {"eval_loss": 4.8133416175842285, "eval_model_preparation_time": 0.004, "eval_runtime": 129.5296, "eval_samples_per_second": 18.853, "eval_steps_per_second": 2.362, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-9792", "checkpoint": "checkpoint-9792", "global_step": 9792, "eval_loss": 4.863399028778076, "metrics": {"eval_loss": 4.863399028778076, "eval_model_preparation_time": 0.0043, "eval_runtime": 130.0049, "eval_samples_per_second": 18.784, "eval_steps_per_second": 2.354, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-10080", "checkpoint": "checkpoint-10080", "global_step": 10080, "eval_loss": 4.913107395172119, "metrics": {"eval_loss": 4.913107395172119, "eval_model_preparation_time": 0.0035, "eval_runtime": 130.1141, "eval_samples_per_second": 18.768, "eval_steps_per_second": 2.352, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-10368", "checkpoint": "checkpoint-10368", "global_step": 10368, "eval_loss": 4.950941562652588, "metrics": {"eval_loss": 4.950941562652588, "eval_model_preparation_time": 0.004, "eval_runtime": 130.2393, "eval_samples_per_second": 18.75, "eval_steps_per_second": 2.35, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-10656", "checkpoint": "checkpoint-10656", "global_step": 10656, "eval_loss": 4.9802350997924805, "metrics": {"eval_loss": 4.9802350997924805, "eval_model_preparation_time": 0.0034, "eval_runtime": 130.2406, "eval_samples_per_second": 18.75, "eval_steps_per_second": 2.349, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-1152.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-1152.json new file mode 100644 index 0000000000000000000000000000000000000000..b7d12fdb213143844034e156c3427a231771d6c0 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-1152.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-1152", + "checkpoint": "checkpoint-1152", + "global_step": 1152, + "tokens_seen": 1207959552, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25164309898426607, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25164309898426607 + ], + "acc": 0.2550288787094204, + "acc_ci": 0.0, + "acc_runs": [ + 0.2550288787094204 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00433070312013829, + "centered_acc": 0.0021907986456880892, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0021907986456880892 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.515778019586507, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.515778019586507 + ], + "acc": 0.5424374319912949, + "acc_ci": 0.0, + "acc_runs": [ + 0.5424374319912949 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011660014400426034, + "centered_acc": 0.03155603917301408, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03155603917301408 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2638888888888889, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2638888888888889 + ], + "acc": 0.2542087542087542, + "acc_ci": 0.0, + "acc_runs": [ + 0.2542087542087542 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009043789220055392, + "centered_acc": 0.018518518518518528, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018518518518518528 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26621160409556316, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26621160409556316 + ], + "acc": 0.20563139931740615, + "acc_ci": 0.0, + "acc_runs": [ + 0.20563139931740615 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012915774781523113, + "centered_acc": 0.021615472127417545, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021615472127417545 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4940805051302289, + "acc_ci": 0.0, + "acc_runs": [ + 0.4940805051302289 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051500838485925, + "centered_acc": -0.011838989739542227, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.011838989739542227 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.51, + "acc_ci": 0.0, + "acc_runs": [ + 0.51 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05024183937956913, + "centered_acc": 0.020000000000000018, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020000000000000018 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 4012181.9624468433, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 4012181.9624468433 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.186, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.186 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012310790208412926, + "centered_acc": -0.08533333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08533333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2949202685925375, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2842674431130805, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.02207961900675963, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2949202685925375 + ], + "mean_acc_runs": [ + 0.2842674431130805 + ], + "mean_centered_acc_runs": [ + -0.02207961900675963 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-1440.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-1440.json new file mode 100644 index 0000000000000000000000000000000000000000..3bd7af48cc1cb4f39aa9b504f67199e3e1894ab2 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-1440.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-1440", + "checkpoint": "checkpoint-1440", + "global_step": 1440, + "tokens_seen": 1509949440, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2533359888468433, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2533359888468433 + ], + "acc": 0.2550288787094204, + "acc_ci": 0.0, + "acc_runs": [ + 0.2550288787094204 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004340328204135388, + "centered_acc": 0.0044479851291243655, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0044479851291243655 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5190424374319913, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5190424374319913 + ], + "acc": 0.530467899891186, + "acc_ci": 0.0, + "acc_runs": [ + 0.530467899891186 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011657360703051524, + "centered_acc": 0.0380848748639826, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0380848748639826 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2601010101010101, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2601010101010101 + ], + "acc": 0.25462962962962965, + "acc_ci": 0.0, + "acc_runs": [ + 0.25462962962962965 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009001718541079735, + "centered_acc": 0.013468013468013481, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013468013468013481 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2696245733788396, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2696245733788396 + ], + "acc": 0.20733788395904437, + "acc_ci": 0.0, + "acc_runs": [ + 0.20733788395904437 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012968040686869096, + "centered_acc": 0.026166097838452806, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.026166097838452806 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.288, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.288 + ], + "acc": 0.162, + "acc_ci": 0.0, + "acc_runs": [ + 0.162 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020271503835075255, + "centered_acc": 0.05066666666666664, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05066666666666664 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.510655090765588, + "acc_ci": 0.0, + "acc_runs": [ + 0.510655090765588 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014049294536290372, + "centered_acc": 0.02131018153117603, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02131018153117603 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 8641940.766986432, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 8641940.766986432 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.187, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.187 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012336254828074166, + "centered_acc": -0.084, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.084 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29618400162644737, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2893406844728569, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.00814236188390926, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29618400162644737 + ], + "mean_acc_runs": [ + 0.2893406844728569 + ], + "mean_centered_acc_runs": [ + -0.00814236188390926 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-1728.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-1728.json new file mode 100644 index 0000000000000000000000000000000000000000..0a66558bc24d2e0cb1bacd402373a8231a61cda2 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-1728.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-1728", + "checkpoint": "checkpoint-1728", + "global_step": 1728, + "tokens_seen": 1811939328, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2557259510057757, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2557259510057757 + ], + "acc": 0.25672176857199763, + "acc_ci": 0.0, + "acc_runs": [ + 0.25672176857199763 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004353768730644862, + "centered_acc": 0.007634601341034293, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007634601341034293 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5179542981501633, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5179542981501633 + ], + "acc": 0.5495103373231773, + "acc_ci": 0.0, + "acc_runs": [ + 0.5495103373231773 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011658300623287121, + "centered_acc": 0.0359085963003265, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0359085963003265 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26346801346801346, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26346801346801346 + ], + "acc": 0.25126262626262624, + "acc_ci": 0.0, + "acc_runs": [ + 0.25126262626262624 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009039157374497906, + "centered_acc": 0.017957351290684615, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017957351290684615 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26706484641638223, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26706484641638223 + ], + "acc": 0.2090443686006826, + "acc_ci": 0.0, + "acc_runs": [ + 0.2090443686006826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01292893319649643, + "centered_acc": 0.022753128555176305, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.022753128555176305 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.272, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.272 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019920483209566107, + "centered_acc": 0.02933333333333336, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02933333333333336 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4956590370955012, + "acc_ci": 0.0, + "acc_runs": [ + 0.4956590370955012 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051956064076884, + "centered_acc": -0.008681925808997626, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.008681925808997626 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 3304271.085373901, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 3304271.085373901 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.195, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.195 + ], + "acc": 0.04, + "acc_ci": 0.0, + "acc_runs": [ + 0.04 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012535235623319256, + "centered_acc": -0.07333333333333332, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07333333333333332 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2952021848400558, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29024855996276855, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.00798370275270837, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2952021848400558 + ], + "mean_acc_runs": [ + 0.29024855996276855 + ], + "mean_centered_acc_runs": [ + -0.00798370275270837 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-2016.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-2016.json new file mode 100644 index 0000000000000000000000000000000000000000..f92389f4133cc7eaf875ac06a7b96c8c5c7fe769 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-2016.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2016", + "checkpoint": "checkpoint-2016", + "global_step": 2016, + "tokens_seen": 2113929216, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2566221868153754, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2566221868153754 + ], + "acc": 0.25692093208524197, + "acc_ci": 0.0, + "acc_runs": [ + 0.25692093208524197 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004358764596401022, + "centered_acc": 0.008829582420500545, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008829582420500545 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.515233949945593, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.515233949945593 + ], + "acc": 0.5418933623503809, + "acc_ci": 0.0, + "acc_runs": [ + 0.5418933623503809 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011660408257153443, + "centered_acc": 0.030467899891186034, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.030467899891186034 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2668350168350168, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2668350168350168 + ], + "acc": 0.25673400673400676, + "acc_ci": 0.0, + "acc_runs": [ + 0.25673400673400676 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009075915859266983, + "centered_acc": 0.02244668911335575, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02244668911335575 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.27047781569965873, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27047781569965873 + ], + "acc": 0.2167235494880546, + "acc_ci": 0.0, + "acc_runs": [ + 0.2167235494880546 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012980954547659553, + "centered_acc": 0.02730375426621164, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02730375426621164 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.16, + "acc_ci": 0.0, + "acc_runs": [ + 0.16 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.49013417521704816, + "acc_ci": 0.0, + "acc_runs": [ + 0.49013417521704816 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014049749833367674, + "centered_acc": -0.019731649565903675, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.019731649565903675 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.53, + "acc_ci": 0.0, + "acc_runs": [ + 0.53 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05016135580465919, + "centered_acc": 0.06000000000000005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06000000000000005 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 6513855.836385778, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 6513855.836385778 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.185, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.185 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012285191326386655, + "centered_acc": -0.08666666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08666666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2956948282159407, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2874693487648433, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.01786394077710174, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2956948282159407 + ], + "mean_acc_runs": [ + 0.2874693487648433 + ], + "mean_centered_acc_runs": [ + -0.01786394077710174 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-2304.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-2304.json new file mode 100644 index 0000000000000000000000000000000000000000..40cd388761da2932629ab0581f59fbcb252d5aba --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-2304.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2304", + "checkpoint": "checkpoint-2304", + "global_step": 2304, + "tokens_seen": 2415919104, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2560246962756423, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2560246962756423 + ], + "acc": 0.2574188408683529, + "acc_ci": 0.0, + "acc_runs": [ + 0.2574188408683529 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004355436696716235, + "centered_acc": 0.008032928367523043, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008032928367523043 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5136017410228509, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5136017410228509 + ], + "acc": 0.5375408052230686, + "acc_ci": 0.0, + "acc_runs": [ + 0.5375408052230686 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011661506839823792, + "centered_acc": 0.027203482045701888, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.027203482045701888 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26262626262626265, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26262626262626265 + ], + "acc": 0.2542087542087542, + "acc_ci": 0.0, + "acc_runs": [ + 0.2542087542087542 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009029861776763957, + "centered_acc": 0.01683501683501687, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01683501683501687 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26621160409556316, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26621160409556316 + ], + "acc": 0.22098976109215018, + "acc_ci": 0.0, + "acc_runs": [ + 0.22098976109215018 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012915774781523113, + "centered_acc": 0.021615472127417545, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021615472127417545 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.156, + "acc_ci": 0.0, + "acc_runs": [ + 0.156 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4964483030781373, + "acc_ci": 0.0, + "acc_runs": [ + 0.4964483030781373 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052131146915853, + "centered_acc": -0.007103393843725381, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.007103393843725381 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 9766817.521419045, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 9766817.521419045 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.19, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.19 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012411851354816256, + "centered_acc": -0.08, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29474405067005316, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2883893926244164, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.015204618991185033, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29474405067005316 + ], + "mean_acc_runs": [ + 0.2883893926244164 + ], + "mean_centered_acc_runs": [ + -0.015204618991185033 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-2592.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-2592.json new file mode 100644 index 0000000000000000000000000000000000000000..7f2506952edb4624397f799c19d0144abbe005ad --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-2592.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2592", + "checkpoint": "checkpoint-2592", + "global_step": 2592, + "tokens_seen": 2717908992, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25413264289982074, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25413264289982074 + ], + "acc": 0.25652260505875324, + "acc_ci": 0.0, + "acc_runs": [ + 0.25652260505875324 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004344827546976306, + "centered_acc": 0.005510190533094317, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005510190533094317 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5125136017410229, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5125136017410229 + ], + "acc": 0.543525571273123, + "acc_ci": 0.0, + "acc_runs": [ + 0.543525571273123 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011662170084916882, + "centered_acc": 0.02502720348204579, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02502720348204579 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26430976430976433, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26430976430976433 + ], + "acc": 0.25042087542087543, + "acc_ci": 0.0, + "acc_runs": [ + 0.25042087542087543 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009048410451862887, + "centered_acc": 0.019079685746352437, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019079685746352437 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2627986348122867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2627986348122867 + ], + "acc": 0.21928327645051193, + "acc_ci": 0.0, + "acc_runs": [ + 0.21928327645051193 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012862523175351397, + "centered_acc": 0.017064846416382284, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017064846416382284 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.146, + "acc_ci": 0.0, + "acc_runs": [ + 0.146 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48145224940805054, + "acc_ci": 0.0, + "acc_runs": [ + 0.48145224940805054 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014042813708888347, + "centered_acc": -0.03709550118389893, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.03709550118389893 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.56, + "acc_ci": 0.0, + "acc_runs": [ + 0.56 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049888765156985884, + "centered_acc": 0.1200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1200000000000001 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 14432407.391368594, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 14432407.391368594 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.185, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.185 + ], + "acc": 0.04, + "acc_ci": 0.0, + "acc_runs": [ + 0.04 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012285191326386655, + "centered_acc": -0.08666666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08666666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2921257739604825, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2875492039385015, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.01650059090281, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2921257739604825 + ], + "mean_acc_runs": [ + 0.2875492039385015 + ], + "mean_centered_acc_runs": [ + -0.01650059090281 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-288.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-288.json new file mode 100644 index 0000000000000000000000000000000000000000..cfa210b0b9780dcbde14f231ed6afdce921fffc6 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-288.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-288", + "checkpoint": "checkpoint-288", + "global_step": 288, + "tokens_seen": 301989888, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25801633140808605, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25801633140808605 + ], + "acc": 0.2550288787094204, + "acc_ci": 0.0, + "acc_runs": [ + 0.2550288787094204 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004366488167386302, + "centered_acc": 0.01068844187744807, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01068844187744807 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5119695321001088, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5119695321001088 + ], + "acc": 0.5266594124047879, + "acc_ci": 0.0, + "acc_runs": [ + 0.5266594124047879 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011662480968070035, + "centered_acc": 0.02393906420021752, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02393906420021752 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.25, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25 + ], + "acc": 0.24831649831649832, + "acc_ci": 0.0, + "acc_runs": [ + 0.24831649831649832 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008885233166386385, + "centered_acc": 0.0, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2696245733788396, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2696245733788396 + ], + "acc": 0.20819112627986347, + "acc_ci": 0.0, + "acc_runs": [ + 0.20819112627986347 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012968040686869096, + "centered_acc": 0.026166097838452806, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.026166097838452806 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.3, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.3 + ], + "acc": 0.154, + "acc_ci": 0.0, + "acc_runs": [ + 0.154 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020514426225627987, + "centered_acc": 0.06666666666666665, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06666666666666665 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4996053670086819, + "acc_ci": 0.0, + "acc_runs": [ + 0.4996053670086819 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405248130604961, + "centered_acc": -0.0007892659826361781, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0007892659826361781 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 397391.13165549963, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 397391.13165549963 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.176, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.176 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012048616898597498, + "centered_acc": -0.09866666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09866666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29426840614783906, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2855088744492952, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.015046748724346344, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29426840614783906 + ], + "mean_acc_runs": [ + 0.2855088744492952 + ], + "mean_centered_acc_runs": [ + -0.015046748724346344 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-2880.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-2880.json new file mode 100644 index 0000000000000000000000000000000000000000..5fc8b03ec533993684b53a28709e07e1b9f09ab5 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-2880.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2880", + "checkpoint": "checkpoint-2880", + "global_step": 2880, + "tokens_seen": 3019898880, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2570205138418642, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2570205138418642 + ], + "acc": 0.2559251145190201, + "acc_ci": 0.0, + "acc_runs": [ + 0.2559251145190201 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004360977256058395, + "centered_acc": 0.009360685122485593, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009360685122485593 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5190424374319913, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5190424374319913 + ], + "acc": 0.5467899891186072, + "acc_ci": 0.0, + "acc_runs": [ + 0.5467899891186072 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011657360703051524, + "centered_acc": 0.0380848748639826, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0380848748639826 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2697811447811448, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2697811447811448 + ], + "acc": 0.2563131313131313, + "acc_ci": 0.0, + "acc_runs": [ + 0.2563131313131313 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009107527914671116, + "centered_acc": 0.02637485970819305, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02637485970819305 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26023890784982934, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26023890784982934 + ], + "acc": 0.2098976109215017, + "acc_ci": 0.0, + "acc_runs": [ + 0.2098976109215017 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012821930225112672, + "centered_acc": 0.013651877133105783, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013651877133105783 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.148, + "acc_ci": 0.0, + "acc_runs": [ + 0.148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.47987371744277824, + "acc_ci": 0.0, + "acc_runs": [ + 0.47987371744277824 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014041096664344346, + "centered_acc": -0.04025256511444353, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.04025256511444353 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 15686717.349860167, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 15686717.349860167 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.191 + ], + "acc": 0.04, + "acc_ci": 0.0, + "acc_runs": [ + 0.04 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012436787112179482, + "centered_acc": -0.07866666666666666, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07866666666666666 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2951805006508049, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2895087025088739, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.009208001267326953, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2951805006508049 + ], + "mean_acc_runs": [ + 0.2895087025088739 + ], + "mean_centered_acc_runs": [ + -0.009208001267326953 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-3168.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-3168.json new file mode 100644 index 0000000000000000000000000000000000000000..189df5b6c02d4a3e840d4fa83e7b0f563f34bb69 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-3168.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-3168", + "checkpoint": "checkpoint-3168", + "global_step": 3168, + "tokens_seen": 3321888768, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2570205138418642, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2570205138418642 + ], + "acc": 0.25632344154550885, + "acc_ci": 0.0, + "acc_runs": [ + 0.25632344154550885 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004360977256058395, + "centered_acc": 0.009360685122485593, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009360685122485593 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.529923830250272, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.529923830250272 + ], + "acc": 0.545157780195865, + "acc_ci": 0.0, + "acc_runs": [ + 0.545157780195865 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011644913435420134, + "centered_acc": 0.05984766050054402, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05984766050054402 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2638888888888889, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2638888888888889 + ], + "acc": 0.25462962962962965, + "acc_ci": 0.0, + "acc_runs": [ + 0.25462962962962965 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009043789220055392, + "centered_acc": 0.018518518518518528, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018518518518518528 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2627986348122867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2627986348122867 + ], + "acc": 0.21245733788395904, + "acc_ci": 0.0, + "acc_runs": [ + 0.21245733788395904 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012862523175351397, + "centered_acc": 0.017064846416382284, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017064846416382284 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.284, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.284 + ], + "acc": 0.148, + "acc_ci": 0.0, + "acc_runs": [ + 0.148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020186703693570774, + "centered_acc": 0.0453333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0453333333333333 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48382004735595896, + "acc_ci": 0.0, + "acc_runs": [ + 0.48382004735595896 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014045126130978676, + "centered_acc": -0.03235990528808208, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.03235990528808208 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 18623383.013392877, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 18623383.013392877 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.189, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.189 + ], + "acc": 0.041, + "acc_ci": 0.0, + "acc_runs": [ + 0.041 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01238678458811779, + "centered_acc": -0.08133333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08133333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.297771977965552, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28996756983846217, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.005221474575861234, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.297771977965552 + ], + "mean_acc_runs": [ + 0.28996756983846217 + ], + "mean_centered_acc_runs": [ + -0.005221474575861234 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-3456.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-3456.json new file mode 100644 index 0000000000000000000000000000000000000000..f024d6401a8087ca264d977d83aacfa460df4ed2 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-3456.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-3456", + "checkpoint": "checkpoint-3456", + "global_step": 3456, + "tokens_seen": 3623878656, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25851424019119695, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25851424019119695 + ], + "acc": 0.2539334793865764, + "acc_ci": 0.0, + "acc_runs": [ + 0.2539334793865764 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004369232540126326, + "centered_acc": 0.011352320254929271, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011352320254929271 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5315560391730142, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5315560391730142 + ], + "acc": 0.5500544069640914, + "acc_ci": 0.0, + "acc_runs": [ + 0.5500544069640914 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011642567710878958, + "centered_acc": 0.06311207834602839, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06311207834602839 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26725589225589225, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26725589225589225 + ], + "acc": 0.25715488215488214, + "acc_ci": 0.0, + "acc_runs": [ + 0.25715488215488214 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009080463246017426, + "centered_acc": 0.023007856341189663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023007856341189663 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2636518771331058, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2636518771331058 + ], + "acc": 0.2175767918088737, + "acc_ci": 0.0, + "acc_runs": [ + 0.2175767918088737 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012875929151296968, + "centered_acc": 0.018202502844141044, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018202502844141044 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.29, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.29 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020313179231745138, + "centered_acc": 0.05333333333333331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05333333333333331 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4964483030781373, + "acc_ci": 0.0, + "acc_runs": [ + 0.4964483030781373 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052131146915853, + "centered_acc": -0.007103393843725381, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.007103393843725381 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 22909114.683314577, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 22909114.683314577 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.189, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.189 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01238678458811779, + "centered_acc": -0.08133333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08133333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2999963414588682, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29314553251662606, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0003170791677818311, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2999963414588682 + ], + "mean_acc_runs": [ + 0.29314553251662606 + ], + "mean_centered_acc_runs": [ + -0.0003170791677818311 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-3744.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-3744.json new file mode 100644 index 0000000000000000000000000000000000000000..178aa859871dc71bb41a2703b5a451705c7342e0 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-3744.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-3744", + "checkpoint": "checkpoint-3744", + "global_step": 3744, + "tokens_seen": 3925868544, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26090420235012945, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26090420235012945 + ], + "acc": 0.25642302330213107, + "acc_ci": 0.0, + "acc_runs": [ + 0.25642302330213107 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004382303181183701, + "centered_acc": 0.014538936466839273, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014538936466839273 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5239390642002176, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5239390642002176 + ], + "acc": 0.5429815016322089, + "acc_ci": 0.0, + "acc_runs": [ + 0.5429815016322089 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165244562107948, + "centered_acc": 0.04787812840043526, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04787812840043526 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2676767676767677, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2676767676767677 + ], + "acc": 0.25462962962962965, + "acc_ci": 0.0, + "acc_runs": [ + 0.25462962962962965 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009085000147099275, + "centered_acc": 0.023569023569023573, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023569023569023573 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26621160409556316, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26621160409556316 + ], + "acc": 0.21416382252559726, + "acc_ci": 0.0, + "acc_runs": [ + 0.21416382252559726 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012915774781523113, + "centered_acc": 0.021615472127417545, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021615472127417545 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.276, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.276 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02001121929807355, + "centered_acc": 0.0346666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0346666666666667 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.47592738752959746, + "acc_ci": 0.0, + "acc_runs": [ + 0.47592738752959746 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014036189665395169, + "centered_acc": -0.04814522494080509, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.04814522494080509 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 16668273.563277863, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 16668273.563277863 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.186, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.186 + ], + "acc": 0.041, + "acc_ci": 0.0, + "acc_runs": [ + 0.041 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012310790208412926, + "centered_acc": -0.08533333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08533333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29678860638711296, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28914128263928646, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.010515045277372837, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29678860638711296 + ], + "mean_acc_runs": [ + 0.28914128263928646 + ], + "mean_centered_acc_runs": [ + -0.010515045277372837 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-4032.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-4032.json new file mode 100644 index 0000000000000000000000000000000000000000..f14a9cabf168ec445cf28e37435a931a87c41459 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-4032.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4032", + "checkpoint": "checkpoint-4032", + "global_step": 4032, + "tokens_seen": 4227858432, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25951005775741887, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25951005775741887 + ], + "acc": 0.2562238597888867, + "acc_ci": 0.0, + "acc_runs": [ + 0.2562238597888867 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004374699189284837, + "centered_acc": 0.012680077009891821, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012680077009891821 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5348204570184983, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5348204570184983 + ], + "acc": 0.5511425462459195, + "acc_ci": 0.0, + "acc_runs": [ + 0.5511425462459195 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011637500993816005, + "centered_acc": 0.06964091403699668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06964091403699668 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2668350168350168, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2668350168350168 + ], + "acc": 0.2563131313131313, + "acc_ci": 0.0, + "acc_runs": [ + 0.2563131313131313 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009075915859266983, + "centered_acc": 0.02244668911335575, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02244668911335575 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26621160409556316, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26621160409556316 + ], + "acc": 0.22098976109215018, + "acc_ci": 0.0, + "acc_runs": [ + 0.22098976109215018 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012915774781523113, + "centered_acc": 0.021615472127417545, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021615472127417545 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.47592738752959746, + "acc_ci": 0.0, + "acc_runs": [ + 0.47592738752959746 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014036189665395169, + "centered_acc": -0.04814522494080509, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.04814522494080509 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 26286782.22605941, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 26286782.22605941 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.186, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.186 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012310790208412926, + "centered_acc": -0.08533333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08533333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2988961892844162, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29308841477433856, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0033911647154528917, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2988961892844162 + ], + "mean_acc_runs": [ + 0.29308841477433856 + ], + "mean_centered_acc_runs": [ + -0.0033911647154528917 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-4320.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-4320.json new file mode 100644 index 0000000000000000000000000000000000000000..9739fed60656a89dc8f6a64ae86adb2523532f87 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-4320.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4320", + "checkpoint": "checkpoint-4320", + "global_step": 4320, + "tokens_seen": 4529848320, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2599083847839076, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2599083847839076 + ], + "acc": 0.2575184226249751, + "acc_ci": 0.0, + "acc_runs": [ + 0.2575184226249751 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004376877619233649, + "centered_acc": 0.013211179711876797, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013211179711876797 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5255712731229597, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5255712731229597 + ], + "acc": 0.544613710554951, + "acc_ci": 0.0, + "acc_runs": [ + 0.544613710554951 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165055784457337, + "centered_acc": 0.05114254624591941, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05114254624591941 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26346801346801346, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26346801346801346 + ], + "acc": 0.25462962962962965, + "acc_ci": 0.0, + "acc_runs": [ + 0.25462962962962965 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009039157374497906, + "centered_acc": 0.017957351290684615, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017957351290684615 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26621160409556316, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26621160409556316 + ], + "acc": 0.2158703071672355, + "acc_ci": 0.0, + "acc_runs": [ + 0.2158703071672355 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012915774781523113, + "centered_acc": 0.021615472127417545, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021615472127417545 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.288, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.288 + ], + "acc": 0.162, + "acc_ci": 0.0, + "acc_runs": [ + 0.162 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020271503835075255, + "centered_acc": 0.05066666666666664, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05066666666666664 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4940805051302289, + "acc_ci": 0.0, + "acc_runs": [ + 0.4940805051302289 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051500838485925, + "centered_acc": -0.011838989739542227, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.011838989739542227 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 32800322.35247887, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 32800322.35247887 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.19, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.19 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012411851354816256, + "centered_acc": -0.08, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29885987924507396, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29220000368807203, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.004518983349952966, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29885987924507396 + ], + "mean_acc_runs": [ + 0.29220000368807203 + ], + "mean_centered_acc_runs": [ + -0.004518983349952966 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-4608.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-4608.json new file mode 100644 index 0000000000000000000000000000000000000000..31e7c5730eb32fcf357660fa1a5aef21b4188ae0 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-4608.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4608", + "checkpoint": "checkpoint-4608", + "global_step": 4608, + "tokens_seen": 4831838208, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26289583748257317, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26289583748257317 + ], + "acc": 0.25642302330213107, + "acc_ci": 0.0, + "acc_runs": [ + 0.25642302330213107 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004393066760916671, + "centered_acc": 0.017194449976764226, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017194449976764226 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5282916213275299, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5282916213275299 + ], + "acc": 0.544069640914037, + "acc_ci": 0.0, + "acc_runs": [ + 0.544069640914037 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0116471341727495, + "centered_acc": 0.05658324265505987, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05658324265505987 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26304713804713803, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26304713804713803 + ], + "acc": 0.25841750841750843, + "acc_ci": 0.0, + "acc_runs": [ + 0.25841750841750843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009034514898865944, + "centered_acc": 0.017396184062850706, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017396184062850706 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2636518771331058, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2636518771331058 + ], + "acc": 0.21928327645051193, + "acc_ci": 0.0, + "acc_runs": [ + 0.21928327645051193 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012875929151296968, + "centered_acc": 0.018202502844141044, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018202502844141044 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.288, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.288 + ], + "acc": 0.156, + "acc_ci": 0.0, + "acc_runs": [ + 0.156 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020271503835075255, + "centered_acc": 0.05066666666666664, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05066666666666664 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4988161010260458, + "acc_ci": 0.0, + "acc_runs": [ + 0.4988161010260458 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052446290528871, + "centered_acc": -0.0023677979479084232, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0023677979479084232 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 45741965.08677404, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 45741965.08677404 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.19, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.19 + ], + "acc": 0.049, + "acc_ci": 0.0, + "acc_runs": [ + 0.049 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012411851354816256, + "centered_acc": -0.08, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29931441233172446, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29402970118839344, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0006388697994472648, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29931441233172446 + ], + "mean_acc_runs": [ + 0.29402970118839344 + ], + "mean_centered_acc_runs": [ + -0.0006388697994472648 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-4896.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-4896.json new file mode 100644 index 0000000000000000000000000000000000000000..51dd5f35e7481b5034162d8d989a22805274b72f --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-4896.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4896", + "checkpoint": "checkpoint-4896", + "global_step": 4896, + "tokens_seen": 5133828096, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.261700856403107, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.261700856403107 + ], + "acc": 0.25731925911173076, + "acc_ci": 0.0, + "acc_runs": [ + 0.25731925911173076 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004386622589118801, + "centered_acc": 0.015601141870809299, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015601141870809299 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.530467899891186, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.530467899891186 + ], + "acc": 0.5364526659412405, + "acc_ci": 0.0, + "acc_runs": [ + 0.5364526659412405 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011644145418355097, + "centered_acc": 0.06093579978237207, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06093579978237207 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2638888888888889, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2638888888888889 + ], + "acc": 0.257996632996633, + "acc_ci": 0.0, + "acc_runs": [ + 0.257996632996633 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009043789220055392, + "centered_acc": 0.018518518518518528, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018518518518518528 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25597269624573377, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25597269624573377 + ], + "acc": 0.22098976109215018, + "acc_ci": 0.0, + "acc_runs": [ + 0.22098976109215018 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012753013241244613, + "centered_acc": 0.007963594994311688, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007963594994311688 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.16, + "acc_ci": 0.0, + "acc_runs": [ + 0.16 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4909234411996843, + "acc_ci": 0.0, + "acc_runs": [ + 0.4909234411996843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050170094497657, + "centered_acc": -0.01815311760063143, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.01815311760063143 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 33790798.58420492, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 33790798.58420492 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.193, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.193 + ], + "acc": 0.051, + "acc_ci": 0.0, + "acc_runs": [ + 0.051 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012486268734370044, + "centered_acc": -0.076, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.076 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29750505690481927, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2922969222115139, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.006062126543024363, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29750505690481927 + ], + "mean_acc_runs": [ + 0.2922969222115139 + ], + "mean_centered_acc_runs": [ + -0.006062126543024363 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-5184.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-5184.json new file mode 100644 index 0000000000000000000000000000000000000000..c5267d58fc81ed556519beb6d6e411ecf4cd8d67 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-5184.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-5184", + "checkpoint": "checkpoint-5184", + "global_step": 5184, + "tokens_seen": 5435817984, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2621987651862179, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2621987651862179 + ], + "acc": 0.2562238597888867, + "acc_ci": 0.0, + "acc_runs": [ + 0.2562238597888867 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004389312748012449, + "centered_acc": 0.0162650202482905, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0162650202482905 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5250272034820457, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5250272034820457 + ], + "acc": 0.5402611534276387, + "acc_ci": 0.0, + "acc_runs": [ + 0.5402611534276387 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011651200967623562, + "centered_acc": 0.05005440696409136, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05005440696409136 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26262626262626265, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26262626262626265 + ], + "acc": 0.2558922558922559, + "acc_ci": 0.0, + "acc_runs": [ + 0.2558922558922559 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009029861776763957, + "centered_acc": 0.01683501683501687, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01683501683501687 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2627986348122867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2627986348122867 + ], + "acc": 0.21928327645051193, + "acc_ci": 0.0, + "acc_runs": [ + 0.21928327645051193 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012862523175351397, + "centered_acc": 0.017064846416382284, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017064846416382284 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.284, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.284 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020186703693570774, + "centered_acc": 0.0453333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0453333333333333 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5035516969218626, + "acc_ci": 0.0, + "acc_runs": [ + 0.5035516969218626 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052131146915852, + "centered_acc": 0.00710339384372527, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.00710339384372527 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 52279569.52421754, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 52279569.52421754 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.192, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.192 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01246159264666003, + "centered_acc": -0.07733333333333332, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07733333333333332 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29810847768446885, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2928499704254856, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0031224880161214667, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29810847768446885 + ], + "mean_acc_runs": [ + 0.2928499704254856 + ], + "mean_centered_acc_runs": [ + -0.0031224880161214667 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-5472.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-5472.json new file mode 100644 index 0000000000000000000000000000000000000000..4c71b9974c63faa54c8acf92d82aea63e84775d5 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-5472.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-5472", + "checkpoint": "checkpoint-5472", + "global_step": 5472, + "tokens_seen": 5737807872, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.261700856403107, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.261700856403107 + ], + "acc": 0.25473013343955386, + "acc_ci": 0.0, + "acc_runs": [ + 0.25473013343955386 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004386622589118801, + "centered_acc": 0.015601141870809299, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015601141870809299 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5277475516866159, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5277475516866159 + ], + "acc": 0.5397170837867247, + "acc_ci": 0.0, + "acc_runs": [ + 0.5397170837867247 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011647846656062001, + "centered_acc": 0.055495103373231824, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.055495103373231824 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26136363636363635, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26136363636363635 + ], + "acc": 0.25336700336700335, + "acc_ci": 0.0, + "acc_runs": [ + 0.25336700336700335 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009015838366608263, + "centered_acc": 0.015151515151515138, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015151515151515138 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25597269624573377, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25597269624573377 + ], + "acc": 0.22013651877133106, + "acc_ci": 0.0, + "acc_runs": [ + 0.22013651877133106 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012753013241244613, + "centered_acc": 0.007963594994311688, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007963594994311688 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5114443567482242, + "acc_ci": 0.0, + "acc_runs": [ + 0.5114443567482242 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014048804199859391, + "centered_acc": 0.022888713496448387, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.022888713496448387 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 67436357.29264966, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 67436357.29264966 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.195, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.195 + ], + "acc": 0.048, + "acc_ci": 0.0, + "acc_runs": [ + 0.048 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012535235623319256, + "centered_acc": -0.07333333333333332, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07333333333333332 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2972974567831822, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2953682557886538, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.002556480640783357, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2972974567831822 + ], + "mean_acc_runs": [ + 0.2953682557886538 + ], + "mean_centered_acc_runs": [ + 0.002556480640783357 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-576.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-576.json new file mode 100644 index 0000000000000000000000000000000000000000..019c2fba51ab8711a28e33c6e8a3e05f4e9a6026 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-576.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-576", + "checkpoint": "checkpoint-576", + "global_step": 576, + "tokens_seen": 603979776, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2557259510057757, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2557259510057757 + ], + "acc": 0.25582553276239794, + "acc_ci": 0.0, + "acc_runs": [ + 0.25582553276239794 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004353768730644862, + "centered_acc": 0.007634601341034293, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007634601341034293 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5190424374319913, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5190424374319913 + ], + "acc": 0.5353645266594124, + "acc_ci": 0.0, + "acc_runs": [ + 0.5353645266594124 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011657360703051524, + "centered_acc": 0.0380848748639826, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0380848748639826 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2609427609427609, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2609427609427609 + ], + "acc": 0.24663299663299662, + "acc_ci": 0.0, + "acc_runs": [ + 0.24663299663299662 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009011142493235922, + "centered_acc": 0.014590347923681227, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014590347923681227 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.27303754266211605, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27303754266211605 + ], + "acc": 0.20563139931740615, + "acc_ci": 0.0, + "acc_runs": [ + 0.20563139931740615 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01301933276263586, + "centered_acc": 0.030716723549488067, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.030716723549488067 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.292, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.292 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020354375480530096, + "centered_acc": 0.05599999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05599999999999997 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5035516969218626, + "acc_ci": 0.0, + "acc_runs": [ + 0.5035516969218626 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052131146915852, + "centered_acc": 0.00710339384372527, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.00710339384372527 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 2620075.134941783, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 2620075.134941783 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29729144867377405, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28712936140677764, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.008736496473780131, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29729144867377405 + ], + "mean_acc_runs": [ + 0.28712936140677764 + ], + "mean_centered_acc_runs": [ + -0.008736496473780131 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-5760.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-5760.json new file mode 100644 index 0000000000000000000000000000000000000000..b7e6f6da6e3a8e25ef531669b282a4af1b40ce33 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-5760.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-5760", + "checkpoint": "checkpoint-5760", + "global_step": 5760, + "tokens_seen": 6039797760, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.261700856403107, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.261700856403107 + ], + "acc": 0.2559251145190201, + "acc_ci": 0.0, + "acc_runs": [ + 0.2559251145190201 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004386622589118801, + "centered_acc": 0.015601141870809299, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015601141870809299 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5201305767138193, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5201305767138193 + ], + "acc": 0.5402611534276387, + "acc_ci": 0.0, + "acc_runs": [ + 0.5402611534276387 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011656365410780507, + "centered_acc": 0.040261153427638696, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040261153427638696 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2647306397306397, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2647306397306397 + ], + "acc": 0.2563131313131313, + "acc_ci": 0.0, + "acc_runs": [ + 0.2563131313131313 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009053021086174166, + "centered_acc": 0.019640852974186274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019640852974186274 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25853242320819114, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25853242320819114 + ], + "acc": 0.22525597269624573, + "acc_ci": 0.0, + "acc_runs": [ + 0.22525597269624573 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012794553754288777, + "centered_acc": 0.011376564277588189, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011376564277588189 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.288, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.288 + ], + "acc": 0.154, + "acc_ci": 0.0, + "acc_runs": [ + 0.154 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020271503835075255, + "centered_acc": 0.05066666666666664, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05066666666666664 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5027624309392266, + "acc_ci": 0.0, + "acc_runs": [ + 0.5027624309392266 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052271211616401, + "centered_acc": 0.005524861878453136, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005524861878453136 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 97347632.52327858, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 97347632.52327858 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.195, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.195 + ], + "acc": 0.05, + "acc_ci": 0.0, + "acc_runs": [ + 0.05 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012535235623319256, + "centered_acc": -0.07333333333333332, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07333333333333332 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2980157493426262, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2902805264668963, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0104096854100656, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2980157493426262 + ], + "mean_acc_runs": [ + 0.2902805264668963 + ], + "mean_centered_acc_runs": [ + -0.0104096854100656 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-6048.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-6048.json new file mode 100644 index 0000000000000000000000000000000000000000..c049357ce07150f299b38a03fada49cddec3d22f --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-6048.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-6048", + "checkpoint": "checkpoint-6048", + "global_step": 6048, + "tokens_seen": 6341787648, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25951005775741887, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25951005775741887 + ], + "acc": 0.2550288787094204, + "acc_ci": 0.0, + "acc_runs": [ + 0.2550288787094204 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004374699189284837, + "centered_acc": 0.012680077009891821, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012680077009891821 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5239390642002176, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5239390642002176 + ], + "acc": 0.5413492927094669, + "acc_ci": 0.0, + "acc_runs": [ + 0.5413492927094669 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165244562107948, + "centered_acc": 0.04787812840043526, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04787812840043526 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26641414141414144, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26641414141414144 + ], + "acc": 0.2537878787878788, + "acc_ci": 0.0, + "acc_runs": [ + 0.2537878787878788 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009071357971078843, + "centered_acc": 0.021885521885521914, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021885521885521914 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2627986348122867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2627986348122867 + ], + "acc": 0.22013651877133106, + "acc_ci": 0.0, + "acc_runs": [ + 0.22013651877133106 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012862523175351397, + "centered_acc": 0.017064846416382284, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017064846416382284 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.148, + "acc_ci": 0.0, + "acc_runs": [ + 0.148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5059194948697711, + "acc_ci": 0.0, + "acc_runs": [ + 0.5059194948697711 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051500838485923, + "centered_acc": 0.011838989739542116, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011838989739542116 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 100225959.61885129, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 100225959.61885129 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.192, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.192 + ], + "acc": 0.05, + "acc_ci": 0.0, + "acc_runs": [ + 0.05 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01246159264666003, + "centered_acc": -0.07733333333333332, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07733333333333332 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2984436496973441, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29225095256215683, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.002378982926017714, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2984436496973441 + ], + "mean_acc_runs": [ + 0.29225095256215683 + ], + "mean_centered_acc_runs": [ + -0.002378982926017714 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-6336.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-6336.json new file mode 100644 index 0000000000000000000000000000000000000000..91a17e2fb23b298969f483ba264c2a70e1c1e94c --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-6336.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-6336", + "checkpoint": "checkpoint-6336", + "global_step": 6336, + "tokens_seen": 6643777536, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26090420235012945, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26090420235012945 + ], + "acc": 0.2570205138418642, + "acc_ci": 0.0, + "acc_runs": [ + 0.2570205138418642 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004382303181183701, + "centered_acc": 0.014538936466839273, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014538936466839273 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5266594124047879, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5266594124047879 + ], + "acc": 0.530467899891186, + "acc_ci": 0.0, + "acc_runs": [ + 0.530467899891186 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01164922999434743, + "centered_acc": 0.05331882480957573, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05331882480957573 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2638888888888889, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2638888888888889 + ], + "acc": 0.2521043771043771, + "acc_ci": 0.0, + "acc_runs": [ + 0.2521043771043771 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009043789220055392, + "centered_acc": 0.018518518518518528, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018518518518518528 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2551194539249147, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2551194539249147 + ], + "acc": 0.2090443686006826, + "acc_ci": 0.0, + "acc_runs": [ + 0.2090443686006826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012739038695202038, + "centered_acc": 0.006825938566552929, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006825938566552929 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.148, + "acc_ci": 0.0, + "acc_runs": [ + 0.148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5146014206787688, + "acc_ci": 0.0, + "acc_runs": [ + 0.5146014206787688 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014046492383275955, + "centered_acc": 0.02920284135753759, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02920284135753759 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.61, + "acc_ci": 0.0, + "acc_runs": [ + 0.61 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04902071300001973, + "centered_acc": 0.21999999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.21999999999999997 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 122580291.84873, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 122580291.84873 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.193, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.193 + ], + "acc": 0.048, + "acc_ci": 0.0, + "acc_runs": [ + 0.048 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012486268734370044, + "centered_acc": -0.076, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.076 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2969286595947868, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2947526041890579, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.0072940722147879205, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2969286595947868 + ], + "mean_acc_runs": [ + 0.2947526041890579 + ], + "mean_centered_acc_runs": [ + 0.0072940722147879205 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-72.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-72.json new file mode 100644 index 0000000000000000000000000000000000000000..53a5f6512780013af48cb271306cf5a640f04c28 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-72.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-72", + "checkpoint": "checkpoint-72", + "global_step": 72, + "tokens_seen": 75497472, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26309500099581756, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26309500099581756 + ], + "acc": 0.25761800438159727, + "acc_ci": 0.0, + "acc_runs": [ + 0.25761800438159727 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004394136724172891, + "centered_acc": 0.01746000132775675, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01746000132775675 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5114254624591947, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5114254624591947 + ], + "acc": 0.5223068552774756, + "acc_ci": 0.0, + "acc_runs": [ + 0.5223068552774756 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011662778026451529, + "centered_acc": 0.02285092491838947, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02285092491838947 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2529461279461279, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2529461279461279 + ], + "acc": 0.24368686868686867, + "acc_ci": 0.0, + "acc_runs": [ + 0.24368686868686867 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00891986273916551, + "centered_acc": 0.003928170594837225, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.003928170594837225 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25597269624573377, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25597269624573377 + ], + "acc": 0.2030716723549488, + "acc_ci": 0.0, + "acc_runs": [ + 0.2030716723549488 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012753013241244613, + "centered_acc": 0.007963594994311688, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007963594994311688 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.27, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019874354831287473, + "centered_acc": 0.02666666666666669, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02666666666666669 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4972375690607735, + "acc_ci": 0.0, + "acc_runs": [ + 0.4972375690607735 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052271211616401, + "centered_acc": -0.0055248618784530246, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0055248618784530246 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 12810.200034143752, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 12810.200034143752 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.169, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.169 + ], + "acc": 0.035, + "acc_ci": 0.0, + "acc_runs": [ + 0.035 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011856625977890152, + "centered_acc": -0.10799999999999998, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10799999999999998 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2870732146078123, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28292084315353644, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.022008953314343386, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2870732146078123 + ], + "mean_acc_runs": [ + 0.28292084315353644 + ], + "mean_centered_acc_runs": [ + -0.022008953314343386 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-864.json b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-864.json new file mode 100644 index 0000000000000000000000000000000000000000..e06db64c025bd162295b2be46f80c251dc770350 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/checkpoint-864.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", + "run": "i5-exp-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-864", + "checkpoint": "checkpoint-864", + "global_step": 864, + "tokens_seen": 905969664, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.252141007767377, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.252141007767377 + ], + "acc": 0.2539334793865764, + "acc_ci": 0.0, + "acc_runs": [ + 0.2539334793865764 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004333543083293202, + "centered_acc": 0.0028546770231693643, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0028546770231693643 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5087051142546246, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5087051142546246 + ], + "acc": 0.5408052230685527, + "acc_ci": 0.0, + "acc_runs": [ + 0.5408052230685527 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011664055982032772, + "centered_acc": 0.017410228509249226, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017410228509249226 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26430976430976433, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26430976430976433 + ], + "acc": 0.2521043771043771, + "acc_ci": 0.0, + "acc_runs": [ + 0.2521043771043771 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009048410451862887, + "centered_acc": 0.019079685746352437, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019079685746352437 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2832764505119454, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2832764505119454 + ], + "acc": 0.21245733788395904, + "acc_ci": 0.0, + "acc_runs": [ + 0.21245733788395904 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.013167478735134564, + "centered_acc": 0.04436860068259385, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04436860068259385 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.29, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.29 + ], + "acc": 0.15, + "acc_ci": 0.0, + "acc_runs": [ + 0.15 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020313179231745138, + "centered_acc": 0.05333333333333331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05333333333333331 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4925019731649566, + "acc_ci": 0.0, + "acc_runs": [ + 0.4925019731649566 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050905521228646, + "centered_acc": -0.014996053670086829, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.014996053670086829 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 3664910.5153558697, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 3664910.5153558697 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.182, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.182 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0122075806376622, + "centered_acc": -0.09066666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09066666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29673872280728525, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28740898523821223, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.012449030166072732, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29673872280728525 + ], + "mean_acc_runs": [ + 0.28740898523821223 + ], + "mean_centered_acc_runs": [ + -0.012449030166072732 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-exp-fulle-lm/zero_shot_eval/summary.jsonl b/results/i5-exp-fulle-lm/zero_shot_eval/summary.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..72feb96243712624990434c08107f34617ce80e9 --- /dev/null +++ b/results/i5-exp-fulle-lm/zero_shot_eval/summary.jsonl @@ -0,0 +1,23 @@ +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-72", "checkpoint": "checkpoint-72", "global_step": 72, "tokens_seen": 75497472, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26309500099581756, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26309500099581756], "acc": 0.25761800438159727, "acc_ci": 0.0, "acc_runs": [0.25761800438159727], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004394136724172891, "centered_acc": 0.01746000132775675, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01746000132775675], "ci": 0.0}, "piqa": {"acc_norm": 0.5114254624591947, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5114254624591947], "acc": 0.5223068552774756, "acc_ci": 0.0, "acc_runs": [0.5223068552774756], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011662778026451529, "centered_acc": 0.02285092491838947, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02285092491838947], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2529461279461279, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2529461279461279], "acc": 0.24368686868686867, "acc_ci": 0.0, "acc_runs": [0.24368686868686867], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00891986273916551, "centered_acc": 0.003928170594837225, "centered_acc_ci": 0.0, "centered_acc_runs": [0.003928170594837225], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25597269624573377, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25597269624573377], "acc": 0.2030716723549488, "acc_ci": 0.0, "acc_runs": [0.2030716723549488], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012753013241244613, "centered_acc": 0.007963594994311688, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007963594994311688], "ci": 0.0}, "openbookqa": {"acc_norm": 0.27, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019874354831287473, "centered_acc": 0.02666666666666669, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02666666666666669], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4972375690607735, "acc_ci": 0.0, "acc_runs": [0.4972375690607735], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052271211616401, "centered_acc": -0.0055248618784530246, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0055248618784530246], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 12810.200034143752, "perplexity_ci": 0.0, "perplexity_runs": [12810.200034143752], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.169, "acc_norm_ci": 0.0, "acc_norm_runs": [0.169], "acc": 0.035, "acc_ci": 0.0, "acc_runs": [0.035], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011856625977890152, "centered_acc": -0.10799999999999998, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10799999999999998], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2870732146078123, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28292084315353644, "mean_acc_ci": 0.0, "mean_centered_acc": -0.022008953314343386, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2870732146078123], "mean_acc_runs": [0.28292084315353644], "mean_centered_acc_runs": [-0.022008953314343386], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-288", "checkpoint": "checkpoint-288", "global_step": 288, "tokens_seen": 301989888, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25801633140808605, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25801633140808605], "acc": 0.2550288787094204, "acc_ci": 0.0, "acc_runs": [0.2550288787094204], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004366488167386302, "centered_acc": 0.01068844187744807, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01068844187744807], "ci": 0.0}, "piqa": {"acc_norm": 0.5119695321001088, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5119695321001088], "acc": 0.5266594124047879, "acc_ci": 0.0, "acc_runs": [0.5266594124047879], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011662480968070035, "centered_acc": 0.02393906420021752, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02393906420021752], "ci": 0.0}, "arc_easy": {"acc_norm": 0.25, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25], "acc": 0.24831649831649832, "acc_ci": 0.0, "acc_runs": [0.24831649831649832], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008885233166386385, "centered_acc": 0.0, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2696245733788396, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2696245733788396], "acc": 0.20819112627986347, "acc_ci": 0.0, "acc_runs": [0.20819112627986347], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012968040686869096, "centered_acc": 0.026166097838452806, "centered_acc_ci": 0.0, "centered_acc_runs": [0.026166097838452806], "ci": 0.0}, "openbookqa": {"acc_norm": 0.3, "acc_norm_ci": 0.0, "acc_norm_runs": [0.3], "acc": 0.154, "acc_ci": 0.0, "acc_runs": [0.154], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020514426225627987, "centered_acc": 0.06666666666666665, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06666666666666665], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4996053670086819, "acc_ci": 0.0, "acc_runs": [0.4996053670086819], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405248130604961, "centered_acc": -0.0007892659826361781, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0007892659826361781], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 397391.13165549963, "perplexity_ci": 0.0, "perplexity_runs": [397391.13165549963], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.176, "acc_norm_ci": 0.0, "acc_norm_runs": [0.176], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012048616898597498, "centered_acc": -0.09866666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09866666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29426840614783906, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2855088744492952, "mean_acc_ci": 0.0, "mean_centered_acc": -0.015046748724346344, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29426840614783906], "mean_acc_runs": [0.2855088744492952], "mean_centered_acc_runs": [-0.015046748724346344], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-576", "checkpoint": "checkpoint-576", "global_step": 576, "tokens_seen": 603979776, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2557259510057757, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2557259510057757], "acc": 0.25582553276239794, "acc_ci": 0.0, "acc_runs": [0.25582553276239794], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004353768730644862, "centered_acc": 0.007634601341034293, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007634601341034293], "ci": 0.0}, "piqa": {"acc_norm": 0.5190424374319913, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5190424374319913], "acc": 0.5353645266594124, "acc_ci": 0.0, "acc_runs": [0.5353645266594124], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011657360703051524, "centered_acc": 0.0380848748639826, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0380848748639826], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2609427609427609, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2609427609427609], "acc": 0.24663299663299662, "acc_ci": 0.0, "acc_runs": [0.24663299663299662], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009011142493235922, "centered_acc": 0.014590347923681227, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014590347923681227], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.27303754266211605, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27303754266211605], "acc": 0.20563139931740615, "acc_ci": 0.0, "acc_runs": [0.20563139931740615], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01301933276263586, "centered_acc": 0.030716723549488067, "centered_acc_ci": 0.0, "centered_acc_runs": [0.030716723549488067], "ci": 0.0}, "openbookqa": {"acc_norm": 0.292, "acc_norm_ci": 0.0, "acc_norm_runs": [0.292], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020354375480530096, "centered_acc": 0.05599999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05599999999999997], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5035516969218626, "acc_ci": 0.0, "acc_runs": [0.5035516969218626], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052131146915852, "centered_acc": 0.00710339384372527, "centered_acc_ci": 0.0, "centered_acc_runs": [0.00710339384372527], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 2620075.134941783, "perplexity_ci": 0.0, "perplexity_runs": [2620075.134941783], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29729144867377405, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28712936140677764, "mean_acc_ci": 0.0, "mean_centered_acc": -0.008736496473780131, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29729144867377405], "mean_acc_runs": [0.28712936140677764], "mean_centered_acc_runs": [-0.008736496473780131], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-864", "checkpoint": "checkpoint-864", "global_step": 864, "tokens_seen": 905969664, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.252141007767377, "acc_norm_ci": 0.0, "acc_norm_runs": [0.252141007767377], "acc": 0.2539334793865764, "acc_ci": 0.0, "acc_runs": [0.2539334793865764], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004333543083293202, "centered_acc": 0.0028546770231693643, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0028546770231693643], "ci": 0.0}, "piqa": {"acc_norm": 0.5087051142546246, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5087051142546246], "acc": 0.5408052230685527, "acc_ci": 0.0, "acc_runs": [0.5408052230685527], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011664055982032772, "centered_acc": 0.017410228509249226, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017410228509249226], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26430976430976433, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26430976430976433], "acc": 0.2521043771043771, "acc_ci": 0.0, "acc_runs": [0.2521043771043771], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009048410451862887, "centered_acc": 0.019079685746352437, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019079685746352437], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2832764505119454, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2832764505119454], "acc": 0.21245733788395904, "acc_ci": 0.0, "acc_runs": [0.21245733788395904], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.013167478735134564, "centered_acc": 0.04436860068259385, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04436860068259385], "ci": 0.0}, "openbookqa": {"acc_norm": 0.29, "acc_norm_ci": 0.0, "acc_norm_runs": [0.29], "acc": 0.15, "acc_ci": 0.0, "acc_runs": [0.15], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020313179231745138, "centered_acc": 0.05333333333333331, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05333333333333331], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4925019731649566, "acc_ci": 0.0, "acc_runs": [0.4925019731649566], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050905521228646, "centered_acc": -0.014996053670086829, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.014996053670086829], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 3664910.5153558697, "perplexity_ci": 0.0, "perplexity_runs": [3664910.5153558697], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.182, "acc_norm_ci": 0.0, "acc_norm_runs": [0.182], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0122075806376622, "centered_acc": -0.09066666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09066666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29673872280728525, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28740898523821223, "mean_acc_ci": 0.0, "mean_centered_acc": -0.012449030166072732, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29673872280728525], "mean_acc_runs": [0.28740898523821223], "mean_centered_acc_runs": [-0.012449030166072732], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-1152", "checkpoint": "checkpoint-1152", "global_step": 1152, "tokens_seen": 1207959552, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25164309898426607, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25164309898426607], "acc": 0.2550288787094204, "acc_ci": 0.0, "acc_runs": [0.2550288787094204], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00433070312013829, "centered_acc": 0.0021907986456880892, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0021907986456880892], "ci": 0.0}, "piqa": {"acc_norm": 0.515778019586507, "acc_norm_ci": 0.0, "acc_norm_runs": [0.515778019586507], "acc": 0.5424374319912949, "acc_ci": 0.0, "acc_runs": [0.5424374319912949], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011660014400426034, "centered_acc": 0.03155603917301408, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03155603917301408], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2638888888888889, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2638888888888889], "acc": 0.2542087542087542, "acc_ci": 0.0, "acc_runs": [0.2542087542087542], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009043789220055392, "centered_acc": 0.018518518518518528, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018518518518518528], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26621160409556316, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26621160409556316], "acc": 0.20563139931740615, "acc_ci": 0.0, "acc_runs": [0.20563139931740615], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012915774781523113, "centered_acc": 0.021615472127417545, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021615472127417545], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4940805051302289, "acc_ci": 0.0, "acc_runs": [0.4940805051302289], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051500838485925, "centered_acc": -0.011838989739542227, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.011838989739542227], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.51, "acc_ci": 0.0, "acc_runs": [0.51], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05024183937956913, "centered_acc": 0.020000000000000018, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020000000000000018], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 4012181.9624468433, "perplexity_ci": 0.0, "perplexity_runs": [4012181.9624468433], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.186, "acc_norm_ci": 0.0, "acc_norm_runs": [0.186], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012310790208412926, "centered_acc": -0.08533333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08533333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2949202685925375, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2842674431130805, "mean_acc_ci": 0.0, "mean_centered_acc": -0.02207961900675963, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2949202685925375], "mean_acc_runs": [0.2842674431130805], "mean_centered_acc_runs": [-0.02207961900675963], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-1440", "checkpoint": "checkpoint-1440", "global_step": 1440, "tokens_seen": 1509949440, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2533359888468433, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2533359888468433], "acc": 0.2550288787094204, "acc_ci": 0.0, "acc_runs": [0.2550288787094204], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004340328204135388, "centered_acc": 0.0044479851291243655, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0044479851291243655], "ci": 0.0}, "piqa": {"acc_norm": 0.5190424374319913, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5190424374319913], "acc": 0.530467899891186, "acc_ci": 0.0, "acc_runs": [0.530467899891186], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011657360703051524, "centered_acc": 0.0380848748639826, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0380848748639826], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2601010101010101, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2601010101010101], "acc": 0.25462962962962965, "acc_ci": 0.0, "acc_runs": [0.25462962962962965], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009001718541079735, "centered_acc": 0.013468013468013481, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013468013468013481], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2696245733788396, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2696245733788396], "acc": 0.20733788395904437, "acc_ci": 0.0, "acc_runs": [0.20733788395904437], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012968040686869096, "centered_acc": 0.026166097838452806, "centered_acc_ci": 0.0, "centered_acc_runs": [0.026166097838452806], "ci": 0.0}, "openbookqa": {"acc_norm": 0.288, "acc_norm_ci": 0.0, "acc_norm_runs": [0.288], "acc": 0.162, "acc_ci": 0.0, "acc_runs": [0.162], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020271503835075255, "centered_acc": 0.05066666666666664, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05066666666666664], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.510655090765588, "acc_ci": 0.0, "acc_runs": [0.510655090765588], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014049294536290372, "centered_acc": 0.02131018153117603, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02131018153117603], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 8641940.766986432, "perplexity_ci": 0.0, "perplexity_runs": [8641940.766986432], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.187, "acc_norm_ci": 0.0, "acc_norm_runs": [0.187], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012336254828074166, "centered_acc": -0.084, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.084], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29618400162644737, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2893406844728569, "mean_acc_ci": 0.0, "mean_centered_acc": -0.00814236188390926, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29618400162644737], "mean_acc_runs": [0.2893406844728569], "mean_centered_acc_runs": [-0.00814236188390926], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-1728", "checkpoint": "checkpoint-1728", "global_step": 1728, "tokens_seen": 1811939328, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2557259510057757, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2557259510057757], "acc": 0.25672176857199763, "acc_ci": 0.0, "acc_runs": [0.25672176857199763], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004353768730644862, "centered_acc": 0.007634601341034293, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007634601341034293], "ci": 0.0}, "piqa": {"acc_norm": 0.5179542981501633, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5179542981501633], "acc": 0.5495103373231773, "acc_ci": 0.0, "acc_runs": [0.5495103373231773], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011658300623287121, "centered_acc": 0.0359085963003265, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0359085963003265], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26346801346801346, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26346801346801346], "acc": 0.25126262626262624, "acc_ci": 0.0, "acc_runs": [0.25126262626262624], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009039157374497906, "centered_acc": 0.017957351290684615, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017957351290684615], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26706484641638223, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26706484641638223], "acc": 0.2090443686006826, "acc_ci": 0.0, "acc_runs": [0.2090443686006826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01292893319649643, "centered_acc": 0.022753128555176305, "centered_acc_ci": 0.0, "centered_acc_runs": [0.022753128555176305], "ci": 0.0}, "openbookqa": {"acc_norm": 0.272, "acc_norm_ci": 0.0, "acc_norm_runs": [0.272], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019920483209566107, "centered_acc": 0.02933333333333336, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02933333333333336], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4956590370955012, "acc_ci": 0.0, "acc_runs": [0.4956590370955012], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051956064076884, "centered_acc": -0.008681925808997626, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.008681925808997626], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 3304271.085373901, "perplexity_ci": 0.0, "perplexity_runs": [3304271.085373901], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.195, "acc_norm_ci": 0.0, "acc_norm_runs": [0.195], "acc": 0.04, "acc_ci": 0.0, "acc_runs": [0.04], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012535235623319256, "centered_acc": -0.07333333333333332, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07333333333333332], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2952021848400558, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29024855996276855, "mean_acc_ci": 0.0, "mean_centered_acc": -0.00798370275270837, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2952021848400558], "mean_acc_runs": [0.29024855996276855], "mean_centered_acc_runs": [-0.00798370275270837], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2016", "checkpoint": "checkpoint-2016", "global_step": 2016, "tokens_seen": 2113929216, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2566221868153754, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2566221868153754], "acc": 0.25692093208524197, "acc_ci": 0.0, "acc_runs": [0.25692093208524197], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004358764596401022, "centered_acc": 0.008829582420500545, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008829582420500545], "ci": 0.0}, "piqa": {"acc_norm": 0.515233949945593, "acc_norm_ci": 0.0, "acc_norm_runs": [0.515233949945593], "acc": 0.5418933623503809, "acc_ci": 0.0, "acc_runs": [0.5418933623503809], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011660408257153443, "centered_acc": 0.030467899891186034, "centered_acc_ci": 0.0, "centered_acc_runs": [0.030467899891186034], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2668350168350168, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2668350168350168], "acc": 0.25673400673400676, "acc_ci": 0.0, "acc_runs": [0.25673400673400676], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009075915859266983, "centered_acc": 0.02244668911335575, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02244668911335575], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.27047781569965873, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27047781569965873], "acc": 0.2167235494880546, "acc_ci": 0.0, "acc_runs": [0.2167235494880546], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012980954547659553, "centered_acc": 0.02730375426621164, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02730375426621164], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.16, "acc_ci": 0.0, "acc_runs": [0.16], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.49013417521704816, "acc_ci": 0.0, "acc_runs": [0.49013417521704816], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014049749833367674, "centered_acc": -0.019731649565903675, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.019731649565903675], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.53, "acc_ci": 0.0, "acc_runs": [0.53], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05016135580465919, "centered_acc": 0.06000000000000005, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06000000000000005], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 6513855.836385778, "perplexity_ci": 0.0, "perplexity_runs": [6513855.836385778], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.185, "acc_norm_ci": 0.0, "acc_norm_runs": [0.185], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012285191326386655, "centered_acc": -0.08666666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08666666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2956948282159407, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2874693487648433, "mean_acc_ci": 0.0, "mean_centered_acc": -0.01786394077710174, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2956948282159407], "mean_acc_runs": [0.2874693487648433], "mean_centered_acc_runs": [-0.01786394077710174], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2304", "checkpoint": "checkpoint-2304", "global_step": 2304, "tokens_seen": 2415919104, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2560246962756423, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2560246962756423], "acc": 0.2574188408683529, "acc_ci": 0.0, "acc_runs": [0.2574188408683529], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004355436696716235, "centered_acc": 0.008032928367523043, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008032928367523043], "ci": 0.0}, "piqa": {"acc_norm": 0.5136017410228509, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5136017410228509], "acc": 0.5375408052230686, "acc_ci": 0.0, "acc_runs": [0.5375408052230686], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011661506839823792, "centered_acc": 0.027203482045701888, "centered_acc_ci": 0.0, "centered_acc_runs": [0.027203482045701888], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26262626262626265, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26262626262626265], "acc": 0.2542087542087542, "acc_ci": 0.0, "acc_runs": [0.2542087542087542], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009029861776763957, "centered_acc": 0.01683501683501687, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01683501683501687], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26621160409556316, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26621160409556316], "acc": 0.22098976109215018, "acc_ci": 0.0, "acc_runs": [0.22098976109215018], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012915774781523113, "centered_acc": 0.021615472127417545, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021615472127417545], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.156, "acc_ci": 0.0, "acc_runs": [0.156], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4964483030781373, "acc_ci": 0.0, "acc_runs": [0.4964483030781373], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052131146915853, "centered_acc": -0.007103393843725381, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.007103393843725381], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 9766817.521419045, "perplexity_ci": 0.0, "perplexity_runs": [9766817.521419045], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.19, "acc_norm_ci": 0.0, "acc_norm_runs": [0.19], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012411851354816256, "centered_acc": -0.08, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29474405067005316, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2883893926244164, "mean_acc_ci": 0.0, "mean_centered_acc": -0.015204618991185033, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29474405067005316], "mean_acc_runs": [0.2883893926244164], "mean_centered_acc_runs": [-0.015204618991185033], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2592", "checkpoint": "checkpoint-2592", "global_step": 2592, "tokens_seen": 2717908992, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25413264289982074, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25413264289982074], "acc": 0.25652260505875324, "acc_ci": 0.0, "acc_runs": [0.25652260505875324], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004344827546976306, "centered_acc": 0.005510190533094317, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005510190533094317], "ci": 0.0}, "piqa": {"acc_norm": 0.5125136017410229, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5125136017410229], "acc": 0.543525571273123, "acc_ci": 0.0, "acc_runs": [0.543525571273123], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011662170084916882, "centered_acc": 0.02502720348204579, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02502720348204579], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26430976430976433, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26430976430976433], "acc": 0.25042087542087543, "acc_ci": 0.0, "acc_runs": [0.25042087542087543], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009048410451862887, "centered_acc": 0.019079685746352437, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019079685746352437], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2627986348122867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2627986348122867], "acc": 0.21928327645051193, "acc_ci": 0.0, "acc_runs": [0.21928327645051193], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012862523175351397, "centered_acc": 0.017064846416382284, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017064846416382284], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.146, "acc_ci": 0.0, "acc_runs": [0.146], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48145224940805054, "acc_ci": 0.0, "acc_runs": [0.48145224940805054], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014042813708888347, "centered_acc": -0.03709550118389893, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.03709550118389893], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.56, "acc_ci": 0.0, "acc_runs": [0.56], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049888765156985884, "centered_acc": 0.1200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1200000000000001], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 14432407.391368594, "perplexity_ci": 0.0, "perplexity_runs": [14432407.391368594], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.185, "acc_norm_ci": 0.0, "acc_norm_runs": [0.185], "acc": 0.04, "acc_ci": 0.0, "acc_runs": [0.04], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012285191326386655, "centered_acc": -0.08666666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08666666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2921257739604825, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2875492039385015, "mean_acc_ci": 0.0, "mean_centered_acc": -0.01650059090281, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2921257739604825], "mean_acc_runs": [0.2875492039385015], "mean_centered_acc_runs": [-0.01650059090281], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-2880", "checkpoint": "checkpoint-2880", "global_step": 2880, "tokens_seen": 3019898880, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2570205138418642, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2570205138418642], "acc": 0.2559251145190201, "acc_ci": 0.0, "acc_runs": [0.2559251145190201], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004360977256058395, "centered_acc": 0.009360685122485593, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009360685122485593], "ci": 0.0}, "piqa": {"acc_norm": 0.5190424374319913, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5190424374319913], "acc": 0.5467899891186072, "acc_ci": 0.0, "acc_runs": [0.5467899891186072], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011657360703051524, "centered_acc": 0.0380848748639826, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0380848748639826], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2697811447811448, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2697811447811448], "acc": 0.2563131313131313, "acc_ci": 0.0, "acc_runs": [0.2563131313131313], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009107527914671116, "centered_acc": 0.02637485970819305, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02637485970819305], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26023890784982934, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26023890784982934], "acc": 0.2098976109215017, "acc_ci": 0.0, "acc_runs": [0.2098976109215017], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012821930225112672, "centered_acc": 0.013651877133105783, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013651877133105783], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.148, "acc_ci": 0.0, "acc_runs": [0.148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.47987371744277824, "acc_ci": 0.0, "acc_runs": [0.47987371744277824], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014041096664344346, "centered_acc": -0.04025256511444353, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.04025256511444353], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 15686717.349860167, "perplexity_ci": 0.0, "perplexity_runs": [15686717.349860167], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.191], "acc": 0.04, "acc_ci": 0.0, "acc_runs": [0.04], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012436787112179482, "centered_acc": -0.07866666666666666, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07866666666666666], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2951805006508049, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2895087025088739, "mean_acc_ci": 0.0, "mean_centered_acc": -0.009208001267326953, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2951805006508049], "mean_acc_runs": [0.2895087025088739], "mean_centered_acc_runs": [-0.009208001267326953], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-3168", "checkpoint": "checkpoint-3168", "global_step": 3168, "tokens_seen": 3321888768, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2570205138418642, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2570205138418642], "acc": 0.25632344154550885, "acc_ci": 0.0, "acc_runs": [0.25632344154550885], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004360977256058395, "centered_acc": 0.009360685122485593, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009360685122485593], "ci": 0.0}, "piqa": {"acc_norm": 0.529923830250272, "acc_norm_ci": 0.0, "acc_norm_runs": [0.529923830250272], "acc": 0.545157780195865, "acc_ci": 0.0, "acc_runs": [0.545157780195865], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011644913435420134, "centered_acc": 0.05984766050054402, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05984766050054402], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2638888888888889, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2638888888888889], "acc": 0.25462962962962965, "acc_ci": 0.0, "acc_runs": [0.25462962962962965], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009043789220055392, "centered_acc": 0.018518518518518528, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018518518518518528], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2627986348122867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2627986348122867], "acc": 0.21245733788395904, "acc_ci": 0.0, "acc_runs": [0.21245733788395904], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012862523175351397, "centered_acc": 0.017064846416382284, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017064846416382284], "ci": 0.0}, "openbookqa": {"acc_norm": 0.284, "acc_norm_ci": 0.0, "acc_norm_runs": [0.284], "acc": 0.148, "acc_ci": 0.0, "acc_runs": [0.148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020186703693570774, "centered_acc": 0.0453333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0453333333333333], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48382004735595896, "acc_ci": 0.0, "acc_runs": [0.48382004735595896], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014045126130978676, "centered_acc": -0.03235990528808208, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.03235990528808208], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 18623383.013392877, "perplexity_ci": 0.0, "perplexity_runs": [18623383.013392877], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.189, "acc_norm_ci": 0.0, "acc_norm_runs": [0.189], "acc": 0.041, "acc_ci": 0.0, "acc_runs": [0.041], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01238678458811779, "centered_acc": -0.08133333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08133333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.297771977965552, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28996756983846217, "mean_acc_ci": 0.0, "mean_centered_acc": -0.005221474575861234, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.297771977965552], "mean_acc_runs": [0.28996756983846217], "mean_centered_acc_runs": [-0.005221474575861234], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-3456", "checkpoint": "checkpoint-3456", "global_step": 3456, "tokens_seen": 3623878656, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25851424019119695, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25851424019119695], "acc": 0.2539334793865764, "acc_ci": 0.0, "acc_runs": [0.2539334793865764], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004369232540126326, "centered_acc": 0.011352320254929271, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011352320254929271], "ci": 0.0}, "piqa": {"acc_norm": 0.5315560391730142, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5315560391730142], "acc": 0.5500544069640914, "acc_ci": 0.0, "acc_runs": [0.5500544069640914], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011642567710878958, "centered_acc": 0.06311207834602839, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06311207834602839], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26725589225589225, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26725589225589225], "acc": 0.25715488215488214, "acc_ci": 0.0, "acc_runs": [0.25715488215488214], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009080463246017426, "centered_acc": 0.023007856341189663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023007856341189663], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2636518771331058, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2636518771331058], "acc": 0.2175767918088737, "acc_ci": 0.0, "acc_runs": [0.2175767918088737], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012875929151296968, "centered_acc": 0.018202502844141044, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018202502844141044], "ci": 0.0}, "openbookqa": {"acc_norm": 0.29, "acc_norm_ci": 0.0, "acc_norm_runs": [0.29], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020313179231745138, "centered_acc": 0.05333333333333331, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05333333333333331], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4964483030781373, "acc_ci": 0.0, "acc_runs": [0.4964483030781373], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052131146915853, "centered_acc": -0.007103393843725381, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.007103393843725381], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 22909114.683314577, "perplexity_ci": 0.0, "perplexity_runs": [22909114.683314577], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.189, "acc_norm_ci": 0.0, "acc_norm_runs": [0.189], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01238678458811779, "centered_acc": -0.08133333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08133333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2999963414588682, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29314553251662606, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0003170791677818311, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2999963414588682], "mean_acc_runs": [0.29314553251662606], "mean_centered_acc_runs": [-0.0003170791677818311], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-3744", "checkpoint": "checkpoint-3744", "global_step": 3744, "tokens_seen": 3925868544, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26090420235012945, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26090420235012945], "acc": 0.25642302330213107, "acc_ci": 0.0, "acc_runs": [0.25642302330213107], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004382303181183701, "centered_acc": 0.014538936466839273, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014538936466839273], "ci": 0.0}, "piqa": {"acc_norm": 0.5239390642002176, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5239390642002176], "acc": 0.5429815016322089, "acc_ci": 0.0, "acc_runs": [0.5429815016322089], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165244562107948, "centered_acc": 0.04787812840043526, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04787812840043526], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2676767676767677, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2676767676767677], "acc": 0.25462962962962965, "acc_ci": 0.0, "acc_runs": [0.25462962962962965], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009085000147099275, "centered_acc": 0.023569023569023573, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023569023569023573], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26621160409556316, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26621160409556316], "acc": 0.21416382252559726, "acc_ci": 0.0, "acc_runs": [0.21416382252559726], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012915774781523113, "centered_acc": 0.021615472127417545, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021615472127417545], "ci": 0.0}, "openbookqa": {"acc_norm": 0.276, "acc_norm_ci": 0.0, "acc_norm_runs": [0.276], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02001121929807355, "centered_acc": 0.0346666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0346666666666667], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.47592738752959746, "acc_ci": 0.0, "acc_runs": [0.47592738752959746], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014036189665395169, "centered_acc": -0.04814522494080509, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.04814522494080509], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 16668273.563277863, "perplexity_ci": 0.0, "perplexity_runs": [16668273.563277863], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.186, "acc_norm_ci": 0.0, "acc_norm_runs": [0.186], "acc": 0.041, "acc_ci": 0.0, "acc_runs": [0.041], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012310790208412926, "centered_acc": -0.08533333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08533333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29678860638711296, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28914128263928646, "mean_acc_ci": 0.0, "mean_centered_acc": -0.010515045277372837, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29678860638711296], "mean_acc_runs": [0.28914128263928646], "mean_centered_acc_runs": [-0.010515045277372837], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4032", "checkpoint": "checkpoint-4032", "global_step": 4032, "tokens_seen": 4227858432, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25951005775741887, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25951005775741887], "acc": 0.2562238597888867, "acc_ci": 0.0, "acc_runs": [0.2562238597888867], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004374699189284837, "centered_acc": 0.012680077009891821, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012680077009891821], "ci": 0.0}, "piqa": {"acc_norm": 0.5348204570184983, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5348204570184983], "acc": 0.5511425462459195, "acc_ci": 0.0, "acc_runs": [0.5511425462459195], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011637500993816005, "centered_acc": 0.06964091403699668, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06964091403699668], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2668350168350168, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2668350168350168], "acc": 0.2563131313131313, "acc_ci": 0.0, "acc_runs": [0.2563131313131313], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009075915859266983, "centered_acc": 0.02244668911335575, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02244668911335575], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26621160409556316, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26621160409556316], "acc": 0.22098976109215018, "acc_ci": 0.0, "acc_runs": [0.22098976109215018], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012915774781523113, "centered_acc": 0.021615472127417545, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021615472127417545], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.47592738752959746, "acc_ci": 0.0, "acc_runs": [0.47592738752959746], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014036189665395169, "centered_acc": -0.04814522494080509, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.04814522494080509], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 26286782.22605941, "perplexity_ci": 0.0, "perplexity_runs": [26286782.22605941], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.186, "acc_norm_ci": 0.0, "acc_norm_runs": [0.186], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012310790208412926, "centered_acc": -0.08533333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08533333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2988961892844162, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29308841477433856, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0033911647154528917, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2988961892844162], "mean_acc_runs": [0.29308841477433856], "mean_centered_acc_runs": [-0.0033911647154528917], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4320", "checkpoint": "checkpoint-4320", "global_step": 4320, "tokens_seen": 4529848320, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2599083847839076, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2599083847839076], "acc": 0.2575184226249751, "acc_ci": 0.0, "acc_runs": [0.2575184226249751], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004376877619233649, "centered_acc": 0.013211179711876797, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013211179711876797], "ci": 0.0}, "piqa": {"acc_norm": 0.5255712731229597, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5255712731229597], "acc": 0.544613710554951, "acc_ci": 0.0, "acc_runs": [0.544613710554951], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165055784457337, "centered_acc": 0.05114254624591941, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05114254624591941], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26346801346801346, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26346801346801346], "acc": 0.25462962962962965, "acc_ci": 0.0, "acc_runs": [0.25462962962962965], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009039157374497906, "centered_acc": 0.017957351290684615, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017957351290684615], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26621160409556316, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26621160409556316], "acc": 0.2158703071672355, "acc_ci": 0.0, "acc_runs": [0.2158703071672355], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012915774781523113, "centered_acc": 0.021615472127417545, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021615472127417545], "ci": 0.0}, "openbookqa": {"acc_norm": 0.288, "acc_norm_ci": 0.0, "acc_norm_runs": [0.288], "acc": 0.162, "acc_ci": 0.0, "acc_runs": [0.162], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020271503835075255, "centered_acc": 0.05066666666666664, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05066666666666664], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4940805051302289, "acc_ci": 0.0, "acc_runs": [0.4940805051302289], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051500838485925, "centered_acc": -0.011838989739542227, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.011838989739542227], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 32800322.35247887, "perplexity_ci": 0.0, "perplexity_runs": [32800322.35247887], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.19, "acc_norm_ci": 0.0, "acc_norm_runs": [0.19], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012411851354816256, "centered_acc": -0.08, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29885987924507396, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29220000368807203, "mean_acc_ci": 0.0, "mean_centered_acc": -0.004518983349952966, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29885987924507396], "mean_acc_runs": [0.29220000368807203], "mean_centered_acc_runs": [-0.004518983349952966], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4608", "checkpoint": "checkpoint-4608", "global_step": 4608, "tokens_seen": 4831838208, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26289583748257317, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26289583748257317], "acc": 0.25642302330213107, "acc_ci": 0.0, "acc_runs": [0.25642302330213107], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004393066760916671, "centered_acc": 0.017194449976764226, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017194449976764226], "ci": 0.0}, "piqa": {"acc_norm": 0.5282916213275299, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5282916213275299], "acc": 0.544069640914037, "acc_ci": 0.0, "acc_runs": [0.544069640914037], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0116471341727495, "centered_acc": 0.05658324265505987, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05658324265505987], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26304713804713803, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26304713804713803], "acc": 0.25841750841750843, "acc_ci": 0.0, "acc_runs": [0.25841750841750843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009034514898865944, "centered_acc": 0.017396184062850706, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017396184062850706], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2636518771331058, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2636518771331058], "acc": 0.21928327645051193, "acc_ci": 0.0, "acc_runs": [0.21928327645051193], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012875929151296968, "centered_acc": 0.018202502844141044, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018202502844141044], "ci": 0.0}, "openbookqa": {"acc_norm": 0.288, "acc_norm_ci": 0.0, "acc_norm_runs": [0.288], "acc": 0.156, "acc_ci": 0.0, "acc_runs": [0.156], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020271503835075255, "centered_acc": 0.05066666666666664, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05066666666666664], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4988161010260458, "acc_ci": 0.0, "acc_runs": [0.4988161010260458], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052446290528871, "centered_acc": -0.0023677979479084232, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0023677979479084232], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 45741965.08677404, "perplexity_ci": 0.0, "perplexity_runs": [45741965.08677404], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.19, "acc_norm_ci": 0.0, "acc_norm_runs": [0.19], "acc": 0.049, "acc_ci": 0.0, "acc_runs": [0.049], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012411851354816256, "centered_acc": -0.08, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29931441233172446, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29402970118839344, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0006388697994472648, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29931441233172446], "mean_acc_runs": [0.29402970118839344], "mean_centered_acc_runs": [-0.0006388697994472648], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-4896", "checkpoint": "checkpoint-4896", "global_step": 4896, "tokens_seen": 5133828096, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.261700856403107, "acc_norm_ci": 0.0, "acc_norm_runs": [0.261700856403107], "acc": 0.25731925911173076, "acc_ci": 0.0, "acc_runs": [0.25731925911173076], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004386622589118801, "centered_acc": 0.015601141870809299, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015601141870809299], "ci": 0.0}, "piqa": {"acc_norm": 0.530467899891186, "acc_norm_ci": 0.0, "acc_norm_runs": [0.530467899891186], "acc": 0.5364526659412405, "acc_ci": 0.0, "acc_runs": [0.5364526659412405], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011644145418355097, "centered_acc": 0.06093579978237207, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06093579978237207], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2638888888888889, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2638888888888889], "acc": 0.257996632996633, "acc_ci": 0.0, "acc_runs": [0.257996632996633], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009043789220055392, "centered_acc": 0.018518518518518528, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018518518518518528], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25597269624573377, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25597269624573377], "acc": 0.22098976109215018, "acc_ci": 0.0, "acc_runs": [0.22098976109215018], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012753013241244613, "centered_acc": 0.007963594994311688, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007963594994311688], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.16, "acc_ci": 0.0, "acc_runs": [0.16], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4909234411996843, "acc_ci": 0.0, "acc_runs": [0.4909234411996843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050170094497657, "centered_acc": -0.01815311760063143, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.01815311760063143], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 33790798.58420492, "perplexity_ci": 0.0, "perplexity_runs": [33790798.58420492], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.193, "acc_norm_ci": 0.0, "acc_norm_runs": [0.193], "acc": 0.051, "acc_ci": 0.0, "acc_runs": [0.051], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012486268734370044, "centered_acc": -0.076, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.076], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29750505690481927, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2922969222115139, "mean_acc_ci": 0.0, "mean_centered_acc": -0.006062126543024363, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29750505690481927], "mean_acc_runs": [0.2922969222115139], "mean_centered_acc_runs": [-0.006062126543024363], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-5184", "checkpoint": "checkpoint-5184", "global_step": 5184, "tokens_seen": 5435817984, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2621987651862179, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2621987651862179], "acc": 0.2562238597888867, "acc_ci": 0.0, "acc_runs": [0.2562238597888867], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004389312748012449, "centered_acc": 0.0162650202482905, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0162650202482905], "ci": 0.0}, "piqa": {"acc_norm": 0.5250272034820457, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5250272034820457], "acc": 0.5402611534276387, "acc_ci": 0.0, "acc_runs": [0.5402611534276387], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011651200967623562, "centered_acc": 0.05005440696409136, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05005440696409136], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26262626262626265, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26262626262626265], "acc": 0.2558922558922559, "acc_ci": 0.0, "acc_runs": [0.2558922558922559], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009029861776763957, "centered_acc": 0.01683501683501687, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01683501683501687], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2627986348122867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2627986348122867], "acc": 0.21928327645051193, "acc_ci": 0.0, "acc_runs": [0.21928327645051193], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012862523175351397, "centered_acc": 0.017064846416382284, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017064846416382284], "ci": 0.0}, "openbookqa": {"acc_norm": 0.284, "acc_norm_ci": 0.0, "acc_norm_runs": [0.284], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020186703693570774, "centered_acc": 0.0453333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0453333333333333], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5035516969218626, "acc_ci": 0.0, "acc_runs": [0.5035516969218626], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052131146915852, "centered_acc": 0.00710339384372527, "centered_acc_ci": 0.0, "centered_acc_runs": [0.00710339384372527], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 52279569.52421754, "perplexity_ci": 0.0, "perplexity_runs": [52279569.52421754], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.192, "acc_norm_ci": 0.0, "acc_norm_runs": [0.192], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01246159264666003, "centered_acc": -0.07733333333333332, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07733333333333332], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29810847768446885, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2928499704254856, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0031224880161214667, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29810847768446885], "mean_acc_runs": [0.2928499704254856], "mean_centered_acc_runs": [-0.0031224880161214667], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-5472", "checkpoint": "checkpoint-5472", "global_step": 5472, "tokens_seen": 5737807872, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.261700856403107, "acc_norm_ci": 0.0, "acc_norm_runs": [0.261700856403107], "acc": 0.25473013343955386, "acc_ci": 0.0, "acc_runs": [0.25473013343955386], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004386622589118801, "centered_acc": 0.015601141870809299, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015601141870809299], "ci": 0.0}, "piqa": {"acc_norm": 0.5277475516866159, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5277475516866159], "acc": 0.5397170837867247, "acc_ci": 0.0, "acc_runs": [0.5397170837867247], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011647846656062001, "centered_acc": 0.055495103373231824, "centered_acc_ci": 0.0, "centered_acc_runs": [0.055495103373231824], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26136363636363635, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26136363636363635], "acc": 0.25336700336700335, "acc_ci": 0.0, "acc_runs": [0.25336700336700335], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009015838366608263, "centered_acc": 0.015151515151515138, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015151515151515138], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25597269624573377, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25597269624573377], "acc": 0.22013651877133106, "acc_ci": 0.0, "acc_runs": [0.22013651877133106], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012753013241244613, "centered_acc": 0.007963594994311688, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007963594994311688], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5114443567482242, "acc_ci": 0.0, "acc_runs": [0.5114443567482242], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014048804199859391, "centered_acc": 0.022888713496448387, "centered_acc_ci": 0.0, "centered_acc_runs": [0.022888713496448387], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 67436357.29264966, "perplexity_ci": 0.0, "perplexity_runs": [67436357.29264966], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.195, "acc_norm_ci": 0.0, "acc_norm_runs": [0.195], "acc": 0.048, "acc_ci": 0.0, "acc_runs": [0.048], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012535235623319256, "centered_acc": -0.07333333333333332, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07333333333333332], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2972974567831822, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2953682557886538, "mean_acc_ci": 0.0, "mean_centered_acc": 0.002556480640783357, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2972974567831822], "mean_acc_runs": [0.2953682557886538], "mean_centered_acc_runs": [0.002556480640783357], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-5760", "checkpoint": "checkpoint-5760", "global_step": 5760, "tokens_seen": 6039797760, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.261700856403107, "acc_norm_ci": 0.0, "acc_norm_runs": [0.261700856403107], "acc": 0.2559251145190201, "acc_ci": 0.0, "acc_runs": [0.2559251145190201], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004386622589118801, "centered_acc": 0.015601141870809299, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015601141870809299], "ci": 0.0}, "piqa": {"acc_norm": 0.5201305767138193, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5201305767138193], "acc": 0.5402611534276387, "acc_ci": 0.0, "acc_runs": [0.5402611534276387], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011656365410780507, "centered_acc": 0.040261153427638696, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040261153427638696], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2647306397306397, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2647306397306397], "acc": 0.2563131313131313, "acc_ci": 0.0, "acc_runs": [0.2563131313131313], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009053021086174166, "centered_acc": 0.019640852974186274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019640852974186274], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25853242320819114, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25853242320819114], "acc": 0.22525597269624573, "acc_ci": 0.0, "acc_runs": [0.22525597269624573], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012794553754288777, "centered_acc": 0.011376564277588189, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011376564277588189], "ci": 0.0}, "openbookqa": {"acc_norm": 0.288, "acc_norm_ci": 0.0, "acc_norm_runs": [0.288], "acc": 0.154, "acc_ci": 0.0, "acc_runs": [0.154], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020271503835075255, "centered_acc": 0.05066666666666664, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05066666666666664], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5027624309392266, "acc_ci": 0.0, "acc_runs": [0.5027624309392266], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052271211616401, "centered_acc": 0.005524861878453136, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005524861878453136], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 97347632.52327858, "perplexity_ci": 0.0, "perplexity_runs": [97347632.52327858], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.195, "acc_norm_ci": 0.0, "acc_norm_runs": [0.195], "acc": 0.05, "acc_ci": 0.0, "acc_runs": [0.05], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012535235623319256, "centered_acc": -0.07333333333333332, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07333333333333332], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2980157493426262, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2902805264668963, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0104096854100656, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2980157493426262], "mean_acc_runs": [0.2902805264668963], "mean_centered_acc_runs": [-0.0104096854100656], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-6048", "checkpoint": "checkpoint-6048", "global_step": 6048, "tokens_seen": 6341787648, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25951005775741887, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25951005775741887], "acc": 0.2550288787094204, "acc_ci": 0.0, "acc_runs": [0.2550288787094204], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004374699189284837, "centered_acc": 0.012680077009891821, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012680077009891821], "ci": 0.0}, "piqa": {"acc_norm": 0.5239390642002176, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5239390642002176], "acc": 0.5413492927094669, "acc_ci": 0.0, "acc_runs": [0.5413492927094669], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165244562107948, "centered_acc": 0.04787812840043526, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04787812840043526], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26641414141414144, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26641414141414144], "acc": 0.2537878787878788, "acc_ci": 0.0, "acc_runs": [0.2537878787878788], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009071357971078843, "centered_acc": 0.021885521885521914, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021885521885521914], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2627986348122867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2627986348122867], "acc": 0.22013651877133106, "acc_ci": 0.0, "acc_runs": [0.22013651877133106], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012862523175351397, "centered_acc": 0.017064846416382284, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017064846416382284], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.148, "acc_ci": 0.0, "acc_runs": [0.148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5059194948697711, "acc_ci": 0.0, "acc_runs": [0.5059194948697711], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051500838485923, "centered_acc": 0.011838989739542116, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011838989739542116], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 100225959.61885129, "perplexity_ci": 0.0, "perplexity_runs": [100225959.61885129], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.192, "acc_norm_ci": 0.0, "acc_norm_runs": [0.192], "acc": 0.05, "acc_ci": 0.0, "acc_runs": [0.05], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01246159264666003, "centered_acc": -0.07733333333333332, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07733333333333332], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2984436496973441, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29225095256215683, "mean_acc_ci": 0.0, "mean_centered_acc": -0.002378982926017714, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2984436496973441], "mean_acc_runs": [0.29225095256215683], "mean_centered_acc_runs": [-0.002378982926017714], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-exp-fulle-lm", "run": "i5-exp-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-exp-fulle-lm/checkpoint-6336", "checkpoint": "checkpoint-6336", "global_step": 6336, "tokens_seen": 6643777536, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26090420235012945, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26090420235012945], "acc": 0.2570205138418642, "acc_ci": 0.0, "acc_runs": [0.2570205138418642], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004382303181183701, "centered_acc": 0.014538936466839273, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014538936466839273], "ci": 0.0}, "piqa": {"acc_norm": 0.5266594124047879, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5266594124047879], "acc": 0.530467899891186, "acc_ci": 0.0, "acc_runs": [0.530467899891186], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01164922999434743, "centered_acc": 0.05331882480957573, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05331882480957573], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2638888888888889, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2638888888888889], "acc": 0.2521043771043771, "acc_ci": 0.0, "acc_runs": [0.2521043771043771], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009043789220055392, "centered_acc": 0.018518518518518528, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018518518518518528], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2551194539249147, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2551194539249147], "acc": 0.2090443686006826, "acc_ci": 0.0, "acc_runs": [0.2090443686006826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012739038695202038, "centered_acc": 0.006825938566552929, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006825938566552929], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.148, "acc_ci": 0.0, "acc_runs": [0.148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5146014206787688, "acc_ci": 0.0, "acc_runs": [0.5146014206787688], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014046492383275955, "centered_acc": 0.02920284135753759, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02920284135753759], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.61, "acc_ci": 0.0, "acc_runs": [0.61], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04902071300001973, "centered_acc": 0.21999999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.21999999999999997], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 122580291.84873, "perplexity_ci": 0.0, "perplexity_runs": [122580291.84873], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.193, "acc_norm_ci": 0.0, "acc_norm_runs": [0.193], "acc": 0.048, "acc_ci": 0.0, "acc_runs": [0.048], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012486268734370044, "centered_acc": -0.076, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.076], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2969286595947868, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2947526041890579, "mean_acc_ci": 0.0, "mean_centered_acc": 0.0072940722147879205, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2969286595947868], "mean_acc_runs": [0.2947526041890579], "mean_centered_acc_runs": [0.0072940722147879205], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-1152.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-1152.json new file mode 100644 index 0000000000000000000000000000000000000000..641eea67e71040ab9f27c0114b0d5d0a584e760f --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-1152.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-1152", + "checkpoint": "checkpoint-1152", + "global_step": 1152, + "eval_loss": 4.2302422523498535, + "metrics": { + "eval_loss": 4.2302422523498535, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 132.8432, + "eval_samples_per_second": 18.383, + "eval_steps_per_second": 4.599, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-1440.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-1440.json new file mode 100644 index 0000000000000000000000000000000000000000..57ff9d9cfed6540b856cb4ef8ee4c4a23cdc99a2 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-1440.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-1440", + "checkpoint": "checkpoint-1440", + "global_step": 1440, + "eval_loss": 4.151072978973389, + "metrics": { + "eval_loss": 4.151072978973389, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 132.8903, + "eval_samples_per_second": 18.376, + "eval_steps_per_second": 4.598, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-1728.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-1728.json new file mode 100644 index 0000000000000000000000000000000000000000..6d49cddf62cdd0e1c209f9c20047a399bd4957b6 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-1728.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-1728", + "checkpoint": "checkpoint-1728", + "global_step": 1728, + "eval_loss": 4.045864105224609, + "metrics": { + "eval_loss": 4.045864105224609, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 132.8467, + "eval_samples_per_second": 18.382, + "eval_steps_per_second": 4.599, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-2016.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-2016.json new file mode 100644 index 0000000000000000000000000000000000000000..8153f46a3cff33b52f3cff5d8cb166e6d8fad96d --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-2016.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2016", + "checkpoint": "checkpoint-2016", + "global_step": 2016, + "eval_loss": 4.0259318351745605, + "metrics": { + "eval_loss": 4.0259318351745605, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 132.8754, + "eval_samples_per_second": 18.378, + "eval_steps_per_second": 4.598, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-2304.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-2304.json new file mode 100644 index 0000000000000000000000000000000000000000..1a00e228d885b010cbaad571b543a747205539bd --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-2304.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2304", + "checkpoint": "checkpoint-2304", + "global_step": 2304, + "eval_loss": 4.0159807205200195, + "metrics": { + "eval_loss": 4.0159807205200195, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 132.9599, + "eval_samples_per_second": 18.366, + "eval_steps_per_second": 4.595, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-2592.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-2592.json new file mode 100644 index 0000000000000000000000000000000000000000..5e7be6f46884a2531cb221af7ae32b39863d0854 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-2592.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2592", + "checkpoint": "checkpoint-2592", + "global_step": 2592, + "eval_loss": 4.0263776779174805, + "metrics": { + "eval_loss": 4.0263776779174805, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 132.9365, + "eval_samples_per_second": 18.37, + "eval_steps_per_second": 4.596, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-288.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-288.json new file mode 100644 index 0000000000000000000000000000000000000000..f0a98c2e95f50add70cc25bf2cc08de1c8952143 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-288.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-288", + "checkpoint": "checkpoint-288", + "global_step": 288, + "eval_loss": 6.202169418334961, + "metrics": { + "eval_loss": 6.202169418334961, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 132.5473, + "eval_samples_per_second": 18.424, + "eval_steps_per_second": 4.61, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-2880.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-2880.json new file mode 100644 index 0000000000000000000000000000000000000000..c932b4a8d158a6b511f500bdd1ec76b594155bbc --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-2880.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2880", + "checkpoint": "checkpoint-2880", + "global_step": 2880, + "eval_loss": 4.058934688568115, + "metrics": { + "eval_loss": 4.058934688568115, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 133.025, + "eval_samples_per_second": 18.357, + "eval_steps_per_second": 4.593, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-3168.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-3168.json new file mode 100644 index 0000000000000000000000000000000000000000..8caf159c9a033ad30f3a4546ba2039f06d47b47e --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-3168.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-3168", + "checkpoint": "checkpoint-3168", + "global_step": 3168, + "eval_loss": 4.093045234680176, + "metrics": { + "eval_loss": 4.093045234680176, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 132.8826, + "eval_samples_per_second": 18.377, + "eval_steps_per_second": 4.598, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-3456.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-3456.json new file mode 100644 index 0000000000000000000000000000000000000000..fcd3617446bb2b562b4c51a1ba8610bda88e6793 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-3456.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-3456", + "checkpoint": "checkpoint-3456", + "global_step": 3456, + "eval_loss": 4.126306056976318, + "metrics": { + "eval_loss": 4.126306056976318, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 132.9645, + "eval_samples_per_second": 18.366, + "eval_steps_per_second": 4.595, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-3744.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-3744.json new file mode 100644 index 0000000000000000000000000000000000000000..aa3cbdd9a7efa5b54c005d10e04b778ea016f700 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-3744.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-3744", + "checkpoint": "checkpoint-3744", + "global_step": 3744, + "eval_loss": 4.180170059204102, + "metrics": { + "eval_loss": 4.180170059204102, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 132.8656, + "eval_samples_per_second": 18.379, + "eval_steps_per_second": 4.599, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-4032.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-4032.json new file mode 100644 index 0000000000000000000000000000000000000000..74e86beab3dea93c138a6b7e8e4f40cdacb18e1d --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-4032.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4032", + "checkpoint": "checkpoint-4032", + "global_step": 4032, + "eval_loss": 4.2177910804748535, + "metrics": { + "eval_loss": 4.2177910804748535, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 132.9614, + "eval_samples_per_second": 18.366, + "eval_steps_per_second": 4.595, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-4320.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-4320.json new file mode 100644 index 0000000000000000000000000000000000000000..1dfd9e34139c6f53338ac5d95cbec931cbd023b7 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-4320.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4320", + "checkpoint": "checkpoint-4320", + "global_step": 4320, + "eval_loss": 4.260051727294922, + "metrics": { + "eval_loss": 4.260051727294922, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 132.9191, + "eval_samples_per_second": 18.372, + "eval_steps_per_second": 4.597, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-4608.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-4608.json new file mode 100644 index 0000000000000000000000000000000000000000..5648975e83f57ac1757ac1f899f6f53ecb4661ec --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-4608.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4608", + "checkpoint": "checkpoint-4608", + "global_step": 4608, + "eval_loss": 4.314445972442627, + "metrics": { + "eval_loss": 4.314445972442627, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 132.913, + "eval_samples_per_second": 18.373, + "eval_steps_per_second": 4.597, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-4896.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-4896.json new file mode 100644 index 0000000000000000000000000000000000000000..e06a942cbe8fe591b5af04ed9bdfc6c0ad7fc532 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-4896.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4896", + "checkpoint": "checkpoint-4896", + "global_step": 4896, + "eval_loss": 4.3636956214904785, + "metrics": { + "eval_loss": 4.3636956214904785, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 133.0275, + "eval_samples_per_second": 18.357, + "eval_steps_per_second": 4.593, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-5184.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-5184.json new file mode 100644 index 0000000000000000000000000000000000000000..71f4ab249f08204301366c94a5c839a9a1024b12 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-5184.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-5184", + "checkpoint": "checkpoint-5184", + "global_step": 5184, + "eval_loss": 4.42031192779541, + "metrics": { + "eval_loss": 4.42031192779541, + "eval_model_preparation_time": 0.0038, + "eval_runtime": 132.9918, + "eval_samples_per_second": 18.362, + "eval_steps_per_second": 4.594, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-5472.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-5472.json new file mode 100644 index 0000000000000000000000000000000000000000..04d4e001a3dc3d7651f6c1cce249c0ca5c607ce4 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-5472.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-5472", + "checkpoint": "checkpoint-5472", + "global_step": 5472, + "eval_loss": 4.457895755767822, + "metrics": { + "eval_loss": 4.457895755767822, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 133.0391, + "eval_samples_per_second": 18.356, + "eval_steps_per_second": 4.593, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-576.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-576.json new file mode 100644 index 0000000000000000000000000000000000000000..c48e18839a0d9d4eba749e74d921fdb7a49b0735 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-576.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-576", + "checkpoint": "checkpoint-576", + "global_step": 576, + "eval_loss": 5.120080947875977, + "metrics": { + "eval_loss": 5.120080947875977, + "eval_model_preparation_time": 0.004, + "eval_runtime": 132.7851, + "eval_samples_per_second": 18.391, + "eval_steps_per_second": 4.601, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-5760.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-5760.json new file mode 100644 index 0000000000000000000000000000000000000000..dc936d60827b45637b67b6245014dac0b59d89c7 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-5760.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-5760", + "checkpoint": "checkpoint-5760", + "global_step": 5760, + "eval_loss": 4.510598182678223, + "metrics": { + "eval_loss": 4.510598182678223, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 133.0146, + "eval_samples_per_second": 18.359, + "eval_steps_per_second": 4.593, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-6048.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-6048.json new file mode 100644 index 0000000000000000000000000000000000000000..3a4a8281f97245dea5d37a13ba7b73a8e6c60e7f --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-6048.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6048", + "checkpoint": "checkpoint-6048", + "global_step": 6048, + "eval_loss": 4.541780471801758, + "metrics": { + "eval_loss": 4.541780471801758, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 133.0471, + "eval_samples_per_second": 18.354, + "eval_steps_per_second": 4.592, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-6336.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-6336.json new file mode 100644 index 0000000000000000000000000000000000000000..85333c466471efccee9d976e72fecf45053cd410 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-6336.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6336", + "checkpoint": "checkpoint-6336", + "global_step": 6336, + "eval_loss": 4.589366436004639, + "metrics": { + "eval_loss": 4.589366436004639, + "eval_model_preparation_time": 0.0032, + "eval_runtime": 133.1188, + "eval_samples_per_second": 18.345, + "eval_steps_per_second": 4.59, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-6624.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-6624.json new file mode 100644 index 0000000000000000000000000000000000000000..8a505835b7ed67fa5eca72be304329500c29aaf0 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-6624.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6624", + "checkpoint": "checkpoint-6624", + "global_step": 6624, + "eval_loss": 4.640231132507324, + "metrics": { + "eval_loss": 4.640231132507324, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 133.0943, + "eval_samples_per_second": 18.348, + "eval_steps_per_second": 4.591, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-6912.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-6912.json new file mode 100644 index 0000000000000000000000000000000000000000..d6e25cc54e50b0c28bb6408c85cd9818ac6ecc7b --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-6912.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6912", + "checkpoint": "checkpoint-6912", + "global_step": 6912, + "eval_loss": 4.68345308303833, + "metrics": { + "eval_loss": 4.68345308303833, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 131.7405, + "eval_samples_per_second": 18.536, + "eval_steps_per_second": 4.638, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-72.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-72.json new file mode 100644 index 0000000000000000000000000000000000000000..e74b14db5e41f32269d1b548b9c7b8109733bcab --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-72.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-72", + "checkpoint": "checkpoint-72", + "global_step": 72, + "eval_loss": 7.4857707023620605, + "metrics": { + "eval_loss": 7.4857707023620605, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 130.5707, + "eval_samples_per_second": 18.703, + "eval_steps_per_second": 4.679, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-7200.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-7200.json new file mode 100644 index 0000000000000000000000000000000000000000..90c5f961b6e3ad79d669fc36d2faccfe28033a5e --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-7200.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-7200", + "checkpoint": "checkpoint-7200", + "global_step": 7200, + "eval_loss": 4.7398762702941895, + "metrics": { + "eval_loss": 4.7398762702941895, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 132.8065, + "eval_samples_per_second": 18.388, + "eval_steps_per_second": 4.601, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/checkpoint-864.json b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-864.json new file mode 100644 index 0000000000000000000000000000000000000000..b6ebfd4eeb6865efeb748fca55b9c1c5e37548b2 --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/checkpoint-864.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-864", + "checkpoint": "checkpoint-864", + "global_step": 864, + "eval_loss": 4.504836559295654, + "metrics": { + "eval_loss": 4.504836559295654, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 132.8965, + "eval_samples_per_second": 18.375, + "eval_steps_per_second": 4.598, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": true, + "max_next_i_train": 5 +} diff --git a/results/i5-fulle-lm/l2r_n1_eval/summary.jsonl b/results/i5-fulle-lm/l2r_n1_eval/summary.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..ed0cd0716758be4ee9def9bdf1ab96ced3e874dd --- /dev/null +++ b/results/i5-fulle-lm/l2r_n1_eval/summary.jsonl @@ -0,0 +1,26 @@ +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-72", "checkpoint": "checkpoint-72", "global_step": 72, "eval_loss": 7.4857707023620605, "metrics": {"eval_loss": 7.4857707023620605, "eval_model_preparation_time": 0.0035, "eval_runtime": 130.5707, "eval_samples_per_second": 18.703, "eval_steps_per_second": 4.679, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-288", "checkpoint": "checkpoint-288", "global_step": 288, "eval_loss": 6.202169418334961, "metrics": {"eval_loss": 6.202169418334961, "eval_model_preparation_time": 0.0034, "eval_runtime": 132.5473, "eval_samples_per_second": 18.424, "eval_steps_per_second": 4.61, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-576", "checkpoint": "checkpoint-576", "global_step": 576, "eval_loss": 5.120080947875977, "metrics": {"eval_loss": 5.120080947875977, "eval_model_preparation_time": 0.004, "eval_runtime": 132.7851, "eval_samples_per_second": 18.391, "eval_steps_per_second": 4.601, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-864", "checkpoint": "checkpoint-864", "global_step": 864, "eval_loss": 4.504836559295654, "metrics": {"eval_loss": 4.504836559295654, "eval_model_preparation_time": 0.0033, "eval_runtime": 132.8965, "eval_samples_per_second": 18.375, "eval_steps_per_second": 4.598, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-1152", "checkpoint": "checkpoint-1152", "global_step": 1152, "eval_loss": 4.2302422523498535, "metrics": {"eval_loss": 4.2302422523498535, "eval_model_preparation_time": 0.0035, "eval_runtime": 132.8432, "eval_samples_per_second": 18.383, "eval_steps_per_second": 4.599, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-1440", "checkpoint": "checkpoint-1440", "global_step": 1440, "eval_loss": 4.151072978973389, "metrics": {"eval_loss": 4.151072978973389, "eval_model_preparation_time": 0.0034, "eval_runtime": 132.8903, "eval_samples_per_second": 18.376, "eval_steps_per_second": 4.598, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-1728", "checkpoint": "checkpoint-1728", "global_step": 1728, "eval_loss": 4.045864105224609, "metrics": {"eval_loss": 4.045864105224609, "eval_model_preparation_time": 0.0036, "eval_runtime": 132.8467, "eval_samples_per_second": 18.382, "eval_steps_per_second": 4.599, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2016", "checkpoint": "checkpoint-2016", "global_step": 2016, "eval_loss": 4.0259318351745605, "metrics": {"eval_loss": 4.0259318351745605, "eval_model_preparation_time": 0.0034, "eval_runtime": 132.8754, "eval_samples_per_second": 18.378, "eval_steps_per_second": 4.598, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2304", "checkpoint": "checkpoint-2304", "global_step": 2304, "eval_loss": 4.0159807205200195, "metrics": {"eval_loss": 4.0159807205200195, "eval_model_preparation_time": 0.0034, "eval_runtime": 132.9599, "eval_samples_per_second": 18.366, "eval_steps_per_second": 4.595, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2592", "checkpoint": "checkpoint-2592", "global_step": 2592, "eval_loss": 4.0263776779174805, "metrics": {"eval_loss": 4.0263776779174805, "eval_model_preparation_time": 0.0034, "eval_runtime": 132.9365, "eval_samples_per_second": 18.37, "eval_steps_per_second": 4.596, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2880", "checkpoint": "checkpoint-2880", "global_step": 2880, "eval_loss": 4.058934688568115, "metrics": {"eval_loss": 4.058934688568115, "eval_model_preparation_time": 0.0035, "eval_runtime": 133.025, "eval_samples_per_second": 18.357, "eval_steps_per_second": 4.593, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-3168", "checkpoint": "checkpoint-3168", "global_step": 3168, "eval_loss": 4.093045234680176, "metrics": {"eval_loss": 4.093045234680176, "eval_model_preparation_time": 0.0035, "eval_runtime": 132.8826, "eval_samples_per_second": 18.377, "eval_steps_per_second": 4.598, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-3456", "checkpoint": "checkpoint-3456", "global_step": 3456, "eval_loss": 4.126306056976318, "metrics": {"eval_loss": 4.126306056976318, "eval_model_preparation_time": 0.0033, "eval_runtime": 132.9645, "eval_samples_per_second": 18.366, "eval_steps_per_second": 4.595, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-3744", "checkpoint": "checkpoint-3744", "global_step": 3744, "eval_loss": 4.180170059204102, "metrics": {"eval_loss": 4.180170059204102, "eval_model_preparation_time": 0.0037, "eval_runtime": 132.8656, "eval_samples_per_second": 18.379, "eval_steps_per_second": 4.599, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4032", "checkpoint": "checkpoint-4032", "global_step": 4032, "eval_loss": 4.2177910804748535, "metrics": {"eval_loss": 4.2177910804748535, "eval_model_preparation_time": 0.0034, "eval_runtime": 132.9614, "eval_samples_per_second": 18.366, "eval_steps_per_second": 4.595, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4320", "checkpoint": "checkpoint-4320", "global_step": 4320, "eval_loss": 4.260051727294922, "metrics": {"eval_loss": 4.260051727294922, "eval_model_preparation_time": 0.0034, "eval_runtime": 132.9191, "eval_samples_per_second": 18.372, "eval_steps_per_second": 4.597, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4608", "checkpoint": "checkpoint-4608", "global_step": 4608, "eval_loss": 4.314445972442627, "metrics": {"eval_loss": 4.314445972442627, "eval_model_preparation_time": 0.0034, "eval_runtime": 132.913, "eval_samples_per_second": 18.373, "eval_steps_per_second": 4.597, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4896", "checkpoint": "checkpoint-4896", "global_step": 4896, "eval_loss": 4.3636956214904785, "metrics": {"eval_loss": 4.3636956214904785, "eval_model_preparation_time": 0.0033, "eval_runtime": 133.0275, "eval_samples_per_second": 18.357, "eval_steps_per_second": 4.593, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-5184", "checkpoint": "checkpoint-5184", "global_step": 5184, "eval_loss": 4.42031192779541, "metrics": {"eval_loss": 4.42031192779541, "eval_model_preparation_time": 0.0038, "eval_runtime": 132.9918, "eval_samples_per_second": 18.362, "eval_steps_per_second": 4.594, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-5472", "checkpoint": "checkpoint-5472", "global_step": 5472, "eval_loss": 4.457895755767822, "metrics": {"eval_loss": 4.457895755767822, "eval_model_preparation_time": 0.0037, "eval_runtime": 133.0391, "eval_samples_per_second": 18.356, "eval_steps_per_second": 4.593, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-5760", "checkpoint": "checkpoint-5760", "global_step": 5760, "eval_loss": 4.510598182678223, "metrics": {"eval_loss": 4.510598182678223, "eval_model_preparation_time": 0.0036, "eval_runtime": 133.0146, "eval_samples_per_second": 18.359, "eval_steps_per_second": 4.593, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6048", "checkpoint": "checkpoint-6048", "global_step": 6048, "eval_loss": 4.541780471801758, "metrics": {"eval_loss": 4.541780471801758, "eval_model_preparation_time": 0.0037, "eval_runtime": 133.0471, "eval_samples_per_second": 18.354, "eval_steps_per_second": 4.592, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6336", "checkpoint": "checkpoint-6336", "global_step": 6336, "eval_loss": 4.589366436004639, "metrics": {"eval_loss": 4.589366436004639, "eval_model_preparation_time": 0.0032, "eval_runtime": 133.1188, "eval_samples_per_second": 18.345, "eval_steps_per_second": 4.59, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6624", "checkpoint": "checkpoint-6624", "global_step": 6624, "eval_loss": 4.640231132507324, "metrics": {"eval_loss": 4.640231132507324, "eval_model_preparation_time": 0.0034, "eval_runtime": 133.0943, "eval_samples_per_second": 18.348, "eval_steps_per_second": 4.591, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6912", "checkpoint": "checkpoint-6912", "global_step": 6912, "eval_loss": 4.68345308303833, "metrics": {"eval_loss": 4.68345308303833, "eval_model_preparation_time": 0.0034, "eval_runtime": 131.7405, "eval_samples_per_second": 18.536, "eval_steps_per_second": 4.638, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-7200", "checkpoint": "checkpoint-7200", "global_step": 7200, "eval_loss": 4.7398762702941895, "metrics": {"eval_loss": 4.7398762702941895, "eval_model_preparation_time": 0.0033, "eval_runtime": 132.8065, "eval_samples_per_second": 18.388, "eval_steps_per_second": 4.601, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": true, "max_next_i_train": 5} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-1152.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-1152.json new file mode 100644 index 0000000000000000000000000000000000000000..5d241094d31e63da635a042baba27c7a04fbef38 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-1152.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-1152", + "checkpoint": "checkpoint-1152", + "global_step": 1152, + "tokens_seen": 1207959552, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2560246962756423, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2560246962756423 + ], + "acc": 0.2568213503286198, + "acc_ci": 0.0, + "acc_runs": [ + 0.2568213503286198 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004355436696716235, + "centered_acc": 0.008032928367523043, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008032928367523043 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5223068552774756, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5223068552774756 + ], + "acc": 0.5424374319912949, + "acc_ci": 0.0, + "acc_runs": [ + 0.5424374319912949 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011654208652596384, + "centered_acc": 0.044613710554951114, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.044613710554951114 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2689393939393939, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2689393939393939 + ], + "acc": 0.26515151515151514, + "acc_ci": 0.0, + "acc_runs": [ + 0.26515151515151514 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009098548093009251, + "centered_acc": 0.02525252525252523, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02525252525252523 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.27474402730375425, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27474402730375425 + ], + "acc": 0.22184300341296928, + "acc_ci": 0.0, + "acc_runs": [ + 0.22184300341296928 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.013044617212771207, + "centered_acc": 0.03299203640500566, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03299203640500566 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.284, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.284 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020186703693570774, + "centered_acc": 0.0453333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0453333333333333 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.47592738752959746, + "acc_ci": 0.0, + "acc_runs": [ + 0.47592738752959746 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014036189665395169, + "centered_acc": -0.04814522494080509, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.04814522494080509 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 2547853.616903067, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 2547853.616903067 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.188, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.188 + ], + "acc": 0.04, + "acc_ci": 0.0, + "acc_runs": [ + 0.04 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01236158601510376, + "centered_acc": -0.08266666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08266666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29900249546604435, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2882468150187697, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.013112492682970298, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29900249546604435 + ], + "mean_acc_runs": [ + 0.2882468150187697 + ], + "mean_centered_acc_runs": [ + -0.013112492682970298 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-1440.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-1440.json new file mode 100644 index 0000000000000000000000000000000000000000..b4e979c41eed46beb0fd08b6ab6133f041e21c11 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-1440.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-1440", + "checkpoint": "checkpoint-1440", + "global_step": 1440, + "tokens_seen": 1509949440, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25473013343955386, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25473013343955386 + ], + "acc": 0.2559251145190201, + "acc_ci": 0.0, + "acc_runs": [ + 0.2559251145190201 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043481894593367845, + "centered_acc": 0.006306844586071818, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006306844586071818 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5250272034820457, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5250272034820457 + ], + "acc": 0.5505984766050055, + "acc_ci": 0.0, + "acc_runs": [ + 0.5505984766050055 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011651200967623562, + "centered_acc": 0.05005440696409136, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05005440696409136 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26936026936026936, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26936026936026936 + ], + "acc": 0.2596801346801347, + "acc_ci": 0.0, + "acc_runs": [ + 0.2596801346801347 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009103043207756808, + "centered_acc": 0.02581369248035914, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02581369248035914 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2781569965870307, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2781569965870307 + ], + "acc": 0.22098976109215018, + "acc_ci": 0.0, + "acc_runs": [ + 0.22098976109215018 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.013094469919538899, + "centered_acc": 0.037542662116040924, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037542662116040924 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4956590370955012, + "acc_ci": 0.0, + "acc_runs": [ + 0.4956590370955012 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051956064076884, + "centered_acc": -0.008681925808997626, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.008681925808997626 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 2931799.134003098, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 2931799.134003098 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29837910047815, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2908139985765512, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.009080303272040784, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29837910047815 + ], + "mean_acc_runs": [ + 0.2908139985765512 + ], + "mean_centered_acc_runs": [ + -0.009080303272040784 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-1728.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-1728.json new file mode 100644 index 0000000000000000000000000000000000000000..48f88a6e9042fdd122662fb483eafa9c104d0f62 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-1728.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-1728", + "checkpoint": "checkpoint-1728", + "global_step": 1728, + "tokens_seen": 1811939328, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26010754829715194, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26010754829715194 + ], + "acc": 0.2584146584345748, + "acc_ci": 0.0, + "acc_runs": [ + 0.2584146584345748 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004377965074211591, + "centered_acc": 0.013476731062869248, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013476731062869248 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5239390642002176, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5239390642002176 + ], + "acc": 0.5505984766050055, + "acc_ci": 0.0, + "acc_runs": [ + 0.5505984766050055 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165244562107948, + "centered_acc": 0.04787812840043526, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04787812840043526 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2689393939393939, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2689393939393939 + ], + "acc": 0.2601010101010101, + "acc_ci": 0.0, + "acc_runs": [ + 0.2601010101010101 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009098548093009251, + "centered_acc": 0.02525252525252523, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02525252525252523 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2696245733788396, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2696245733788396 + ], + "acc": 0.21843003412969283, + "acc_ci": 0.0, + "acc_runs": [ + 0.21843003412969283 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012968040686869096, + "centered_acc": 0.026166097838452806, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.026166097838452806 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.294, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.294 + ], + "acc": 0.154, + "acc_ci": 0.0, + "acc_runs": [ + 0.154 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02039509548493655, + "centered_acc": 0.058666666666666645, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.058666666666666645 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.494869771112865, + "acc_ci": 0.0, + "acc_runs": [ + 0.494869771112865 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051745961790525, + "centered_acc": -0.010260457774269982, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.010260457774269982 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.56, + "acc_ci": 0.0, + "acc_runs": [ + 0.56 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049888765156985884, + "centered_acc": 0.1200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1200000000000001 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 6849435.695362101, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 6849435.695362101 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.182, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.182 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0122075806376622, + "centered_acc": -0.09066666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09066666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29976842996926717, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29207014121568486, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.005879116852509636, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29976842996926717 + ], + "mean_acc_runs": [ + 0.29207014121568486 + ], + "mean_centered_acc_runs": [ + -0.005879116852509636 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-2016.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-2016.json new file mode 100644 index 0000000000000000000000000000000000000000..766b70caec3cd4069d94179cc642d8fb53262320 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-2016.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2016", + "checkpoint": "checkpoint-2016", + "global_step": 2016, + "tokens_seen": 2113929216, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25941047600079664, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25941047600079664 + ], + "acc": 0.2584146584345748, + "acc_ci": 0.0, + "acc_runs": [ + 0.2584146584345748 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004374153847826732, + "centered_acc": 0.012547301334395522, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012547301334395522 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5326441784548422, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5326441784548422 + ], + "acc": 0.5549510337323177, + "acc_ci": 0.0, + "acc_runs": [ + 0.5549510337323177 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011640934420031744, + "centered_acc": 0.06528835690968449, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06528835690968449 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26515151515151514, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26515151515151514 + ], + "acc": 0.26430976430976433, + "acc_ci": 0.0, + "acc_runs": [ + 0.26430976430976433 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009057621139172679, + "centered_acc": 0.020202020202020183, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020202020202020183 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26023890784982934, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26023890784982934 + ], + "acc": 0.20819112627986347, + "acc_ci": 0.0, + "acc_runs": [ + 0.20819112627986347 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012821930225112672, + "centered_acc": 0.013651877133105783, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013651877133105783 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.284, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.284 + ], + "acc": 0.156, + "acc_ci": 0.0, + "acc_runs": [ + 0.156 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020186703693570774, + "centered_acc": 0.0453333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0453333333333333 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5067087608524072, + "acc_ci": 0.0, + "acc_runs": [ + 0.5067087608524072 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051220692330247, + "centered_acc": 0.013417521704814472, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013417521704814472 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 5755669.654358351, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 5755669.654358351 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2970741795761639, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29438628053826277, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.002776073981693971, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2970741795761639 + ], + "mean_acc_runs": [ + 0.29438628053826277 + ], + "mean_centered_acc_runs": [ + -0.002776073981693971 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-2304.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-2304.json new file mode 100644 index 0000000000000000000000000000000000000000..27c677bfd57080f350f34b81286dcdd1b2b494c4 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-2304.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2304", + "checkpoint": "checkpoint-2304", + "global_step": 2304, + "tokens_seen": 2415919104, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25941047600079664, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25941047600079664 + ], + "acc": 0.2577175861382195, + "acc_ci": 0.0, + "acc_runs": [ + 0.2577175861382195 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004374153847826732, + "centered_acc": 0.012547301334395522, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012547301334395522 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5250272034820457, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5250272034820457 + ], + "acc": 0.5527747551686616, + "acc_ci": 0.0, + "acc_runs": [ + 0.5527747551686616 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011651200967623562, + "centered_acc": 0.05005440696409136, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05005440696409136 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2718855218855219, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2718855218855219 + ], + "acc": 0.26430976430976433, + "acc_ci": 0.0, + "acc_runs": [ + 0.26430976430976433 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009129795867310605, + "centered_acc": 0.02918069584736253, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02918069584736253 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26791808873720135, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26791808873720135 + ], + "acc": 0.20648464163822525, + "acc_ci": 0.0, + "acc_runs": [ + 0.20648464163822525 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012942030195136482, + "centered_acc": 0.023890784982935138, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023890784982935138 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.278, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.278 + ], + "acc": 0.148, + "acc_ci": 0.0, + "acc_runs": [ + 0.148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020055833888070865, + "centered_acc": 0.037333333333333364, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037333333333333364 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.49171270718232046, + "acc_ci": 0.0, + "acc_runs": [ + 0.49171270718232046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405055532282417, + "centered_acc": -0.016574585635359074, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.016574585635359074 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.62, + "acc_ci": 0.0, + "acc_runs": [ + 0.62 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04878317312145634, + "centered_acc": 0.24, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.24 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 8472343.17966308, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 8472343.17966308 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29720688168426096, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2965286916210891, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.0045563178193510495, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29720688168426096 + ], + "mean_acc_runs": [ + 0.2965286916210891 + ], + "mean_centered_acc_runs": [ + 0.0045563178193510495 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-2592.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-2592.json new file mode 100644 index 0000000000000000000000000000000000000000..1493e8e673094e292b6ce7c2c57639e660cbc9fe --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-2592.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2592", + "checkpoint": "checkpoint-2592", + "global_step": 2592, + "tokens_seen": 2717908992, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25960963951404104, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25960963951404104 + ], + "acc": 0.2559251145190201, + "acc_ci": 0.0, + "acc_runs": [ + 0.2559251145190201 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004375244237044889, + "centered_acc": 0.012812852685388046, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012812852685388046 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.529923830250272, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.529923830250272 + ], + "acc": 0.5462459194776932, + "acc_ci": 0.0, + "acc_runs": [ + 0.5462459194776932 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011644913435420134, + "centered_acc": 0.05984766050054402, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05984766050054402 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.27146464646464646, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27146464646464646 + ], + "acc": 0.2706228956228956, + "acc_ci": 0.0, + "acc_runs": [ + 0.2706228956228956 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00912536297036068, + "centered_acc": 0.028619528619528618, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.028619528619528618 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.257679180887372, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.257679180887372 + ], + "acc": 0.2022184300341297, + "acc_ci": 0.0, + "acc_runs": [ + 0.2022184300341297 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012780770562768334, + "centered_acc": 0.010238907849829356, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.010238907849829356 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.284, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.284 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020186703693570774, + "centered_acc": 0.0453333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0453333333333333 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4956590370955012, + "acc_ci": 0.0, + "acc_runs": [ + 0.4956590370955012 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051956064076884, + "centered_acc": -0.008681925808997626, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.008681925808997626 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.62, + "acc_ci": 0.0, + "acc_runs": [ + 0.62 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04878317312145634, + "centered_acc": 0.24, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.24 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 11561444.171015542, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 11561444.171015542 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.193, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.193 + ], + "acc": 0.048, + "acc_ci": 0.0, + "acc_runs": [ + 0.048 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012486268734370044, + "centered_acc": -0.076, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.076 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29927954951938857, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.296895885852294, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.0076383645252251515, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29927954951938857 + ], + "mean_acc_runs": [ + 0.296895885852294 + ], + "mean_centered_acc_runs": [ + 0.0076383645252251515 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-288.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-288.json new file mode 100644 index 0000000000000000000000000000000000000000..5aaece6882ec78acf987de167f4e16b99d50dec4 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-288.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-288", + "checkpoint": "checkpoint-288", + "global_step": 288, + "tokens_seen": 301989888, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25652260505875324, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25652260505875324 + ], + "acc": 0.2531368253335989, + "acc_ci": 0.0, + "acc_runs": [ + 0.2531368253335989 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043582106894420325, + "centered_acc": 0.008696806745004318, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008696806745004318 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5097932535364527, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5097932535364527 + ], + "acc": 0.5353645266594124, + "acc_ci": 0.0, + "acc_runs": [ + 0.5353645266594124 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011663586263283307, + "centered_acc": 0.019586507072905324, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019586507072905324 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.25462962962962965, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25462962962962965 + ], + "acc": 0.24831649831649832, + "acc_ci": 0.0, + "acc_runs": [ + 0.24831649831649832 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008939407288589512, + "centered_acc": 0.006172839506172867, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006172839506172867 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2721843003412969, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2721843003412969 + ], + "acc": 0.20733788395904437, + "acc_ci": 0.0, + "acc_runs": [ + 0.20733788395904437 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01300660040642361, + "centered_acc": 0.029579067121729235, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.029579067121729235 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.144, + "acc_ci": 0.0, + "acc_runs": [ + 0.144 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4909234411996843, + "acc_ci": 0.0, + "acc_runs": [ + 0.4909234411996843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050170094497657, + "centered_acc": -0.01815311760063143, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.01815311760063143 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 100213.09880327753, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 100213.09880327753 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.177, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.177 + ], + "acc": 0.039, + "acc_ci": 0.0, + "acc_runs": [ + 0.039 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012075463420375087, + "centered_acc": -0.09733333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09733333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2926882980943554, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2846366637241939, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.01631958966008359, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2926882980943554 + ], + "mean_acc_runs": [ + 0.2846366637241939 + ], + "mean_centered_acc_runs": [ + -0.01631958966008359 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-2880.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-2880.json new file mode 100644 index 0000000000000000000000000000000000000000..71b40aa517a8ad63e15e63a85b532442968ca1db --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-2880.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2880", + "checkpoint": "checkpoint-2880", + "global_step": 2880, + "tokens_seen": 3019898880, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2599083847839076, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2599083847839076 + ], + "acc": 0.25731925911173076, + "acc_ci": 0.0, + "acc_runs": [ + 0.25731925911173076 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004376877619233649, + "centered_acc": 0.013211179711876797, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013211179711876797 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5201305767138193, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5201305767138193 + ], + "acc": 0.5391730141458106, + "acc_ci": 0.0, + "acc_runs": [ + 0.5391730141458106 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011656365410780507, + "centered_acc": 0.040261153427638696, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040261153427638696 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26304713804713803, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26304713804713803 + ], + "acc": 0.2647306397306397, + "acc_ci": 0.0, + "acc_runs": [ + 0.2647306397306397 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009034514898865944, + "centered_acc": 0.017396184062850706, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017396184062850706 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.257679180887372, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.257679180887372 + ], + "acc": 0.20563139931740615, + "acc_ci": 0.0, + "acc_runs": [ + 0.20563139931740615 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012780770562768334, + "centered_acc": 0.010238907849829356, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.010238907849829356 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.148, + "acc_ci": 0.0, + "acc_runs": [ + 0.148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5082872928176796, + "acc_ci": 0.0, + "acc_runs": [ + 0.5082872928176796 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405055532282417, + "centered_acc": 0.016574585635359185, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.016574585635359185 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 12798579.36306335, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 12798579.36306335 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.19, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.19 + ], + "acc": 0.049, + "acc_ci": 0.0, + "acc_runs": [ + 0.049 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012411851354816256, + "centered_acc": -0.08, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29412754673870617, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29404290668969674, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0015270073072271835, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29412754673870617 + ], + "mean_acc_runs": [ + 0.29404290668969674 + ], + "mean_centered_acc_runs": [ + -0.0015270073072271835 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-3168.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-3168.json new file mode 100644 index 0000000000000000000000000000000000000000..4cccb926aaf61f040531fe5b2e0ad879fa662664 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-3168.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-3168", + "checkpoint": "checkpoint-3168", + "global_step": 3168, + "tokens_seen": 3321888768, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2581159131647082, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2581159131647082 + ], + "acc": 0.25731925911173076, + "acc_ci": 0.0, + "acc_runs": [ + 0.25731925911173076 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004367037632204837, + "centered_acc": 0.010821217552944296, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.010821217552944296 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.544613710554951, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.544613710554951 + ], + "acc": 0.5500544069640914, + "acc_ci": 0.0, + "acc_runs": [ + 0.5500544069640914 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011619292444156876, + "centered_acc": 0.089227421109902, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.089227421109902 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2676767676767677, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2676767676767677 + ], + "acc": 0.26936026936026936, + "acc_ci": 0.0, + "acc_runs": [ + 0.26936026936026936 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009085000147099275, + "centered_acc": 0.023569023569023573, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023569023569023573 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25426621160409557, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25426621160409557 + ], + "acc": 0.2022184300341297, + "acc_ci": 0.0, + "acc_runs": [ + 0.2022184300341297 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012724999945157845, + "centered_acc": 0.0056882821387940945, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0056882821387940945 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5067087608524072, + "acc_ci": 0.0, + "acc_runs": [ + 0.5067087608524072 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051220692330247, + "centered_acc": 0.013417521704814472, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013417521704814472 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 19793869.552791856, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 19793869.552791856 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.196, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.196 + ], + "acc": 0.05, + "acc_ci": 0.0, + "acc_runs": [ + 0.05 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012559527926707349, + "centered_acc": -0.072, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.072 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.3011121005000871, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2945948588096329, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.003922043291430997, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.3011121005000871 + ], + "mean_acc_runs": [ + 0.2945948588096329 + ], + "mean_centered_acc_runs": [ + 0.003922043291430997 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-3456.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-3456.json new file mode 100644 index 0000000000000000000000000000000000000000..67ff40c9eca2fd53dd6d57419b09a2d313e20bc3 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-3456.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-3456", + "checkpoint": "checkpoint-3456", + "global_step": 3456, + "tokens_seen": 3623878656, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2562238597888867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2562238597888867 + ], + "acc": 0.25692093208524197, + "acc_ci": 0.0, + "acc_runs": [ + 0.25692093208524197 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004356547185846709, + "centered_acc": 0.00829847971851557, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.00829847971851557 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5321001088139282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5321001088139282 + ], + "acc": 0.5495103373231773, + "acc_ci": 0.0, + "acc_runs": [ + 0.5495103373231773 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011641758014820223, + "centered_acc": 0.06420021762785644, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06420021762785644 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.27146464646464646, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27146464646464646 + ], + "acc": 0.27441077441077444, + "acc_ci": 0.0, + "acc_runs": [ + 0.27441077441077444 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00912536297036068, + "centered_acc": 0.028619528619528618, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.028619528619528618 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26023890784982934, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26023890784982934 + ], + "acc": 0.20477815699658702, + "acc_ci": 0.0, + "acc_runs": [ + 0.20477815699658702 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012821930225112672, + "centered_acc": 0.013651877133105783, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013651877133105783 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.276, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.276 + ], + "acc": 0.154, + "acc_ci": 0.0, + "acc_runs": [ + 0.154 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02001121929807355, + "centered_acc": 0.0346666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0346666666666667 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5114443567482242, + "acc_ci": 0.0, + "acc_runs": [ + 0.5114443567482242 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014048804199859391, + "centered_acc": 0.022888713496448387, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.022888713496448387 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.56, + "acc_ci": 0.0, + "acc_runs": [ + 0.56 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049888765156985884, + "centered_acc": 0.1200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1200000000000001 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 24398718.721468586, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 24398718.721468586 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.19, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.19 + ], + "acc": 0.05, + "acc_ci": 0.0, + "acc_runs": [ + 0.05 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012411851354816256, + "centered_acc": -0.08, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29767125381954845, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29393520193377054, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0034555103544975296, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29767125381954845 + ], + "mean_acc_runs": [ + 0.29393520193377054 + ], + "mean_centered_acc_runs": [ + -0.0034555103544975296 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-3744.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-3744.json new file mode 100644 index 0000000000000000000000000000000000000000..f8180368d00fc9e3f3be4c1f83422fd687ef8340 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-3744.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-3744", + "checkpoint": "checkpoint-3744", + "global_step": 3744, + "tokens_seen": 3925868544, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2610037841067516, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2610037841067516 + ], + "acc": 0.25642302330213107, + "acc_ci": 0.0, + "acc_runs": [ + 0.25642302330213107 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043828441286430965, + "centered_acc": 0.0146717121423355, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0146717121423355 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5315560391730142, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5315560391730142 + ], + "acc": 0.544613710554951, + "acc_ci": 0.0, + "acc_runs": [ + 0.544613710554951 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011642567710878958, + "centered_acc": 0.06311207834602839, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06311207834602839 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2718855218855219, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2718855218855219 + ], + "acc": 0.2697811447811448, + "acc_ci": 0.0, + "acc_runs": [ + 0.2697811447811448 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009129795867310605, + "centered_acc": 0.02918069584736253, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02918069584736253 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2508532423208191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2508532423208191 + ], + "acc": 0.1945392491467577, + "acc_ci": 0.0, + "acc_runs": [ + 0.1945392491467577 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012668198621315347, + "centered_acc": 0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.162, + "acc_ci": 0.0, + "acc_runs": [ + 0.162 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5019731649565904, + "acc_ci": 0.0, + "acc_runs": [ + 0.5019731649565904 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052376259225643, + "centered_acc": 0.003946329913180779, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.003946329913180779 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.61, + "acc_ci": 0.0, + "acc_runs": [ + 0.61 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04902071300001973, + "centered_acc": 0.21999999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.21999999999999997 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 19344082.739549376, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 19344082.739549376 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.186, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.186 + ], + "acc": 0.053, + "acc_ci": 0.0, + "acc_runs": [ + 0.053 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012310790208412926, + "centered_acc": -0.08533333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08533333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29688309791435114, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29706177545152757, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.004810006987859258, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29688309791435114 + ], + "mean_acc_runs": [ + 0.29706177545152757 + ], + "mean_centered_acc_runs": [ + 0.004810006987859258 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-4032.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-4032.json new file mode 100644 index 0000000000000000000000000000000000000000..0ba6002a630fb46d5f625a1af76c70d30ff6188d --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-4032.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4032", + "checkpoint": "checkpoint-4032", + "global_step": 4032, + "tokens_seen": 4227858432, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25851424019119695, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25851424019119695 + ], + "acc": 0.2559251145190201, + "acc_ci": 0.0, + "acc_runs": [ + 0.2559251145190201 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004369232540126326, + "centered_acc": 0.011352320254929271, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011352320254929271 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.529379760609358, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.529379760609358 + ], + "acc": 0.5375408052230686, + "acc_ci": 0.0, + "acc_runs": [ + 0.5375408052230686 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01164566756505079, + "centered_acc": 0.05875952121871597, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05875952121871597 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.273989898989899, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.273989898989899 + ], + "acc": 0.2697811447811448, + "acc_ci": 0.0, + "acc_runs": [ + 0.2697811447811448 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00915180590154416, + "centered_acc": 0.03198653198653201, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03198653198653201 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2619453924914676, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2619453924914676 + ], + "acc": 0.20477815699658702, + "acc_ci": 0.0, + "acc_runs": [ + 0.20477815699658702 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012849054826857987, + "centered_acc": 0.015927189988623452, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015927189988623452 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.278, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.278 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020055833888070865, + "centered_acc": 0.037333333333333364, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037333333333333364 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5035516969218626, + "acc_ci": 0.0, + "acc_runs": [ + 0.5035516969218626 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052131146915852, + "centered_acc": 0.00710339384372527, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.00710339384372527 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 22451524.063414067, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 22451524.063414067 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.19, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.19 + ], + "acc": 0.049, + "acc_ci": 0.0, + "acc_runs": [ + 0.049 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012411851354816256, + "centered_acc": -0.08, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29863821538032026, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29208643802153833, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0023291984251933473, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29863821538032026 + ], + "mean_acc_runs": [ + 0.29208643802153833 + ], + "mean_centered_acc_runs": [ + -0.0023291984251933473 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-4320.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-4320.json new file mode 100644 index 0000000000000000000000000000000000000000..c931144f79bdc8f4a1eeea3673099e5957257d6f --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-4320.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4320", + "checkpoint": "checkpoint-4320", + "global_step": 4320, + "tokens_seen": 4529848320, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26020713005377416, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26020713005377416 + ], + "acc": 0.25801633140808605, + "acc_ci": 0.0, + "acc_runs": [ + 0.25801633140808605 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004378508362084079, + "centered_acc": 0.013609506738365548, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013609506738365548 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5212187159956474, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5212187159956474 + ], + "acc": 0.544613710554951, + "acc_ci": 0.0, + "acc_runs": [ + 0.544613710554951 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011655314732288771, + "centered_acc": 0.042437431991294794, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.042437431991294794 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.27146464646464646, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27146464646464646 + ], + "acc": 0.2697811447811448, + "acc_ci": 0.0, + "acc_runs": [ + 0.2697811447811448 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00912536297036068, + "centered_acc": 0.028619528619528618, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.028619528619528618 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26023890784982934, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26023890784982934 + ], + "acc": 0.1962457337883959, + "acc_ci": 0.0, + "acc_runs": [ + 0.1962457337883959 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012821930225112672, + "centered_acc": 0.013651877133105783, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013651877133105783 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.278, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.278 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020055833888070865, + "centered_acc": 0.037333333333333364, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037333333333333364 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.49171270718232046, + "acc_ci": 0.0, + "acc_runs": [ + 0.49171270718232046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405055532282417, + "centered_acc": -0.016574585635359074, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.016574585635359074 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.6, + "acc_ci": 0.0, + "acc_runs": [ + 0.6 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0492365963917331, + "centered_acc": 0.19999999999999996, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.19999999999999996 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 42494483.9688464, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 42494483.9688464 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.19, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.19 + ], + "acc": 0.054, + "acc_ci": 0.0, + "acc_runs": [ + 0.054 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012411851354816256, + "centered_acc": -0.08, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2968549000606496, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29506570894885986, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.00048310936359226595, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2968549000606496 + ], + "mean_acc_runs": [ + 0.29506570894885986 + ], + "mean_centered_acc_runs": [ + -0.00048310936359226595 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-4608.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-4608.json new file mode 100644 index 0000000000000000000000000000000000000000..63c4637e25403ebf0ac656511183508c8d816cff --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-4608.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4608", + "checkpoint": "checkpoint-4608", + "global_step": 4608, + "tokens_seen": 4831838208, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.261202947619996, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.261202947619996 + ], + "acc": 0.2593108942441745, + "acc_ci": 0.0, + "acc_runs": [ + 0.2593108942441745 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004383925147478856, + "centered_acc": 0.014937263493328024, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014937263493328024 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5321001088139282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5321001088139282 + ], + "acc": 0.5495103373231773, + "acc_ci": 0.0, + "acc_runs": [ + 0.5495103373231773 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011641758014820223, + "centered_acc": 0.06420021762785644, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06420021762785644 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.265993265993266, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.265993265993266 + ], + "acc": 0.2676767676767677, + "acc_ci": 0.0, + "acc_runs": [ + 0.2676767676767677 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009066789565615746, + "centered_acc": 0.021324354657688005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021324354657688005 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2525597269624573, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2525597269624573 + ], + "acc": 0.2022184300341297, + "acc_ci": 0.0, + "acc_runs": [ + 0.2022184300341297 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012696728980207687, + "centered_acc": 0.003412969283276427, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.003412969283276427 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4846093133385951, + "acc_ci": 0.0, + "acc_runs": [ + 0.4846093133385951 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014045826789783585, + "centered_acc": -0.030781373322809835, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.030781373322809835 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 28706360.124189243, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 28706360.124189243 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.192, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.192 + ], + "acc": 0.052, + "acc_ci": 0.0, + "acc_runs": [ + 0.052 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01246159264666003, + "centered_acc": -0.07733333333333332, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07733333333333332 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29830934156494127, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29316132043905446, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.004407219782954861, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29830934156494127 + ], + "mean_acc_runs": [ + 0.29316132043905446 + ], + "mean_centered_acc_runs": [ + -0.004407219782954861 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-4896.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-4896.json new file mode 100644 index 0000000000000000000000000000000000000000..e20a25a75e9243b33d355162ad638f03babb5e09 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-4896.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4896", + "checkpoint": "checkpoint-4896", + "global_step": 4896, + "tokens_seen": 5133828096, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2614021111332404, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2614021111332404 + ], + "acc": 0.2592113124875523, + "acc_ci": 0.0, + "acc_runs": [ + 0.2592113124875523 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004385004998923735, + "centered_acc": 0.015202814844320548, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015202814844320548 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5348204570184983, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5348204570184983 + ], + "acc": 0.544069640914037, + "acc_ci": 0.0, + "acc_runs": [ + 0.544069640914037 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011637500993816005, + "centered_acc": 0.06964091403699668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06964091403699668 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2647306397306397, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2647306397306397 + ], + "acc": 0.2702020202020202, + "acc_ci": 0.0, + "acc_runs": [ + 0.2702020202020202 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009053021086174166, + "centered_acc": 0.019640852974186274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019640852974186274 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26706484641638223, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26706484641638223 + ], + "acc": 0.19880546075085323, + "acc_ci": 0.0, + "acc_runs": [ + 0.19880546075085323 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01292893319649643, + "centered_acc": 0.022753128555176305, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.022753128555176305 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.148, + "acc_ci": 0.0, + "acc_runs": [ + 0.148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4909234411996843, + "acc_ci": 0.0, + "acc_runs": [ + 0.4909234411996843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050170094497657, + "centered_acc": -0.01815311760063143, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.01815311760063143 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 55221746.34222039, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 55221746.34222039 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.193, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.193 + ], + "acc": 0.055, + "acc_ci": 0.0, + "acc_runs": [ + 0.055 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012486268734370044, + "centered_acc": -0.076, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.076 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29916967571646014, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2934499337327848, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.0001843907063832223, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29916967571646014 + ], + "mean_acc_runs": [ + 0.2934499337327848 + ], + "mean_centered_acc_runs": [ + 0.0001843907063832223 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-5184.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-5184.json new file mode 100644 index 0000000000000000000000000000000000000000..09ae858310a264db8187fe0e64acf7cec535afeb --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-5184.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-5184", + "checkpoint": "checkpoint-5184", + "global_step": 5184, + "tokens_seen": 5435817984, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2605058753236407, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2605058753236407 + ], + "acc": 0.26020713005377416, + "acc_ci": 0.0, + "acc_runs": [ + 0.26020713005377416 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004380136468544087, + "centered_acc": 0.014007833764854297, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014007833764854297 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5397170837867247, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5397170837867247 + ], + "acc": 0.5462459194776932, + "acc_ci": 0.0, + "acc_runs": [ + 0.5462459194776932 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011628961491718543, + "centered_acc": 0.07943416757344934, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.07943416757344934 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26515151515151514, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26515151515151514 + ], + "acc": 0.26262626262626265, + "acc_ci": 0.0, + "acc_runs": [ + 0.26262626262626265 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009057621139172679, + "centered_acc": 0.020202020202020183, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020202020202020183 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24829351535836178, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24829351535836178 + ], + "acc": 0.1945392491467577, + "acc_ci": 0.0, + "acc_runs": [ + 0.1945392491467577 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012624912868089679, + "centered_acc": -0.0022753128555176305, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0022753128555176305 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.264, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.264 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01973288558592215, + "centered_acc": 0.018666666666666682, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018666666666666682 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4909234411996843, + "acc_ci": 0.0, + "acc_runs": [ + 0.4909234411996843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050170094497657, + "centered_acc": -0.01815311760063143, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.01815311760063143 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.61, + "acc_ci": 0.0, + "acc_runs": [ + 0.61 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04902071300001973, + "centered_acc": 0.21999999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.21999999999999997 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 49500622.3597079, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 49500622.3597079 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.19, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.19 + ], + "acc": 0.049, + "acc_ci": 0.0, + "acc_runs": [ + 0.049 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012411851354816256, + "centered_acc": -0.08, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2946113316033737, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29438294642778723, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.0009396868109157833, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2946113316033737 + ], + "mean_acc_runs": [ + 0.29438294642778723 + ], + "mean_centered_acc_runs": [ + 0.0009396868109157833 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-5472.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-5472.json new file mode 100644 index 0000000000000000000000000000000000000000..97dc2c2a831553a9f9b532d743a7343d2997e27b --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-5472.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-5472", + "checkpoint": "checkpoint-5472", + "global_step": 5472, + "tokens_seen": 5737807872, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2622983469428401, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2622983469428401 + ], + "acc": 0.26070503883688506, + "acc_ci": 0.0, + "acc_runs": [ + 0.26070503883688506 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004389849907040485, + "centered_acc": 0.0163977959237868, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0163977959237868 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5364526659412405, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5364526659412405 + ], + "acc": 0.5418933623503809, + "acc_ci": 0.0, + "acc_runs": [ + 0.5418933623503809 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01163477983787243, + "centered_acc": 0.07290533188248105, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.07290533188248105 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.25841750841750843, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25841750841750843 + ], + "acc": 0.2689393939393939, + "acc_ci": 0.0, + "acc_runs": [ + 0.2689393939393939 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008982741341291388, + "centered_acc": 0.011223344556677914, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011223344556677914 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2525597269624573, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2525597269624573 + ], + "acc": 0.2030716723549488, + "acc_ci": 0.0, + "acc_runs": [ + 0.2030716723549488 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012696728980207687, + "centered_acc": 0.003412969283276427, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.003412969283276427 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.278, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.278 + ], + "acc": 0.15, + "acc_ci": 0.0, + "acc_runs": [ + 0.15 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020055833888070865, + "centered_acc": 0.037333333333333364, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037333333333333364 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4877663772691397, + "acc_ci": 0.0, + "acc_runs": [ + 0.4877663772691397 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01404827882040573, + "centered_acc": -0.024467245461720633, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.024467245461720633 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 46741535.6698555, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 46741535.6698555 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.197, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.197 + ], + "acc": 0.053, + "acc_ci": 0.0, + "acc_runs": [ + 0.053 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012583693787968194, + "centered_acc": -0.07066666666666666, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07066666666666666 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29745470804400775, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2933663306524449, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0019206904001590173, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29745470804400775 + ], + "mean_acc_runs": [ + 0.2933663306524449 + ], + "mean_centered_acc_runs": [ + -0.0019206904001590173 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-576.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-576.json new file mode 100644 index 0000000000000000000000000000000000000000..9e3076778c2b43fcfc9cb5f7de566eccb208c5c6 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-576.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-576", + "checkpoint": "checkpoint-576", + "global_step": 576, + "tokens_seen": 603979776, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2538338976299542, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2538338976299542 + ], + "acc": 0.2562238597888867, + "acc_ci": 0.0, + "acc_runs": [ + 0.2562238597888867 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004343142545094425, + "centered_acc": 0.005111863506605567, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005111863506605567 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5223068552774756, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5223068552774756 + ], + "acc": 0.5353645266594124, + "acc_ci": 0.0, + "acc_runs": [ + 0.5353645266594124 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011654208652596384, + "centered_acc": 0.044613710554951114, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.044613710554951114 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26346801346801346, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26346801346801346 + ], + "acc": 0.25336700336700335, + "acc_ci": 0.0, + "acc_runs": [ + 0.25336700336700335 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009039157374497906, + "centered_acc": 0.017957351290684615, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017957351290684615 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2883959044368601, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2883959044368601 + ], + "acc": 0.2022184300341297, + "acc_ci": 0.0, + "acc_runs": [ + 0.2022184300341297 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.013238394422428247, + "centered_acc": 0.05119453924914678, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05119453924914678 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.304, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.304 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02059164957122494, + "centered_acc": 0.072, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.072 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4846093133385951, + "acc_ci": 0.0, + "acc_runs": [ + 0.4846093133385951 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014045826789783585, + "centered_acc": -0.030781373322809835, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.030781373322809835 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 910792.405258101, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 910792.405258101 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.185, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.185 + ], + "acc": 0.041, + "acc_ci": 0.0, + "acc_runs": [ + 0.041 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012285191326386655, + "centered_acc": -0.08666666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08666666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.30283411180205055, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28530705949617274, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.007777294648965302, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.30283411180205055 + ], + "mean_acc_runs": [ + 0.28530705949617274 + ], + "mean_centered_acc_runs": [ + -0.007777294648965302 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-5760.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-5760.json new file mode 100644 index 0000000000000000000000000000000000000000..52780ab9a98c0d70bf7c61cc15bd0ab6808853b6 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-5760.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-5760", + "checkpoint": "checkpoint-5760", + "global_step": 5760, + "tokens_seen": 6039797760, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2620991834295957, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2620991834295957 + ], + "acc": 0.26110336586337385, + "acc_ci": 0.0, + "acc_runs": [ + 0.26110336586337385 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043887752982100766, + "centered_acc": 0.016132244572794274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.016132244572794274 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5457018498367792, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5457018498367792 + ], + "acc": 0.5571273122959739, + "acc_ci": 0.0, + "acc_runs": [ + 0.5571273122959739 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011616990094583295, + "centered_acc": 0.09140369967355833, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.09140369967355833 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2601010101010101, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2601010101010101 + ], + "acc": 0.26346801346801346, + "acc_ci": 0.0, + "acc_runs": [ + 0.26346801346801346 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009001718541079735, + "centered_acc": 0.013468013468013481, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013468013468013481 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25170648464163825, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25170648464163825 + ], + "acc": 0.20733788395904437, + "acc_ci": 0.0, + "acc_runs": [ + 0.20733788395904437 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012682496334042911, + "centered_acc": 0.0022753128555176674, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0022753128555176674 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.268, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.268 + ], + "acc": 0.15, + "acc_ci": 0.0, + "acc_runs": [ + 0.15 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019827714859587477, + "centered_acc": 0.02400000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02400000000000002 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48697711128650356, + "acc_ci": 0.0, + "acc_runs": [ + 0.48697711128650356 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014047718393997596, + "centered_acc": -0.026045777426992878, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.026045777426992878 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 60888305.54860191, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 60888305.54860191 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.19, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.19 + ], + "acc": 0.053, + "acc_ci": 0.0, + "acc_runs": [ + 0.053 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012411851354816256, + "centered_acc": -0.08, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2962680880015039, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29473011486466094, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.002465731478856504, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2962680880015039 + ], + "mean_acc_runs": [ + 0.29473011486466094 + ], + "mean_centered_acc_runs": [ + -0.002465731478856504 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-6048.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-6048.json new file mode 100644 index 0000000000000000000000000000000000000000..b20008c9a4e612e1843fea43805c5cf9958df2d2 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-6048.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6048", + "checkpoint": "checkpoint-6048", + "global_step": 6048, + "tokens_seen": 6341787648, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26319458275243973, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26319458275243973 + ], + "acc": 0.2614021111332404, + "acc_ci": 0.0, + "acc_runs": [ + 0.2614021111332404 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004394671271021473, + "centered_acc": 0.017592777003252975, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017592777003252975 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5353645266594124, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5353645266594124 + ], + "acc": 0.5478781284004353, + "acc_ci": 0.0, + "acc_runs": [ + 0.5478781284004353 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011636607860111725, + "centered_acc": 0.07072905331882473, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.07072905331882473 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26136363636363635, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26136363636363635 + ], + "acc": 0.2680976430976431, + "acc_ci": 0.0, + "acc_runs": [ + 0.2680976430976431 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009015838366608263, + "centered_acc": 0.015151515151515138, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015151515151515138 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2525597269624573, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2525597269624573 + ], + "acc": 0.19795221843003413, + "acc_ci": 0.0, + "acc_runs": [ + 0.19795221843003413 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012696728980207687, + "centered_acc": 0.003412969283276427, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.003412969283276427 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.276, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.276 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02001121929807355, + "centered_acc": 0.0346666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0346666666666667 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4846093133385951, + "acc_ci": 0.0, + "acc_runs": [ + 0.4846093133385951 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014045826789783585, + "centered_acc": -0.030781373322809835, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.030781373322809835 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 67012298.04645489, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 67012298.04645489 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.198, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.198 + ], + "acc": 0.057, + "acc_ci": 0.0, + "acc_runs": [ + 0.057 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012607733934175283, + "centered_acc": -0.06933333333333332, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.06933333333333332 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2977470787896576, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2937226876173648, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0024429779650229563, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2977470787896576 + ], + "mean_acc_runs": [ + 0.2937226876173648 + ], + "mean_centered_acc_runs": [ + -0.0024429779650229563 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-6336.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-6336.json new file mode 100644 index 0000000000000000000000000000000000000000..2ec2ae0c9bca8c736bed0a78ef5d9f5127403592 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-6336.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6336", + "checkpoint": "checkpoint-6336", + "global_step": 6336, + "tokens_seen": 6643777536, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2635929097789285, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2635929097789285 + ], + "acc": 0.26020713005377416, + "acc_ci": 0.0, + "acc_runs": [ + 0.26020713005377416 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004396806562351701, + "centered_acc": 0.018123879705238027, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018123879705238027 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5261153427638737, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5261153427638737 + ], + "acc": 0.5478781284004353, + "acc_ci": 0.0, + "acc_runs": [ + 0.5478781284004353 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01164990085426341, + "centered_acc": 0.052230685527747456, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.052230685527747456 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.265993265993266, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.265993265993266 + ], + "acc": 0.26851851851851855, + "acc_ci": 0.0, + "acc_runs": [ + 0.26851851851851855 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009066789565615746, + "centered_acc": 0.021324354657688005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021324354657688005 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24829351535836178, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24829351535836178 + ], + "acc": 0.19539249146757678, + "acc_ci": 0.0, + "acc_runs": [ + 0.19539249146757678 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012624912868089679, + "centered_acc": -0.0022753128555176305, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0022753128555176305 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.266, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.266 + ], + "acc": 0.15, + "acc_ci": 0.0, + "acc_runs": [ + 0.15 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019780559675655396, + "centered_acc": 0.021333333333333353, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021333333333333353 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4925019731649566, + "acc_ci": 0.0, + "acc_runs": [ + 0.4925019731649566 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050905521228646, + "centered_acc": -0.014996053670086829, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.014996053670086829 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 84311625.70162986, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 84311625.70162986 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.194, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.194 + ], + "acc": 0.051, + "acc_ci": 0.0, + "acc_runs": [ + 0.051 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012510816141264348, + "centered_acc": -0.07466666666666666, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07466666666666666 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29399917231573836, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29337857033789616, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0047056507134293, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29399917231573836 + ], + "mean_acc_runs": [ + 0.29337857033789616 + ], + "mean_centered_acc_runs": [ + -0.0047056507134293 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-6624.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-6624.json new file mode 100644 index 0000000000000000000000000000000000000000..02594ea7aa5b64b059aec7dc2963a777de3a3f95 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-6624.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6624", + "checkpoint": "checkpoint-6624", + "global_step": 6624, + "tokens_seen": 6945767424, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26399123680541725, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26399123680541725 + ], + "acc": 0.25960963951404104, + "acc_ci": 0.0, + "acc_runs": [ + 0.25960963951404104 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004398937225038117, + "centered_acc": 0.018654982407223002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018654982407223002 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5282916213275299, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5282916213275299 + ], + "acc": 0.5462459194776932, + "acc_ci": 0.0, + "acc_runs": [ + 0.5462459194776932 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0116471341727495, + "centered_acc": 0.05658324265505987, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05658324265505987 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26641414141414144, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26641414141414144 + ], + "acc": 0.2680976430976431, + "acc_ci": 0.0, + "acc_runs": [ + 0.2680976430976431 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009071357971078843, + "centered_acc": 0.021885521885521914, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021885521885521914 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25341296928327645, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25341296928327645 + ], + "acc": 0.19539249146757678, + "acc_ci": 0.0, + "acc_runs": [ + 0.19539249146757678 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012710896778378677, + "centered_acc": 0.004550625711035261, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.004550625711035261 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4980268350434096, + "acc_ci": 0.0, + "acc_runs": [ + 0.4980268350434096 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052376259225643, + "centered_acc": -0.003946329913180779, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.003946329913180779 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 91907972.31872264, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 91907972.31872264 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.201, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.201 + ], + "acc": 0.052, + "acc_ci": 0.0, + "acc_runs": [ + 0.052 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012679107214617274, + "centered_acc": -0.06533333333333331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.06533333333333331 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29885166147172754, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29456599903740643, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.000996625884414063, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29885166147172754 + ], + "mean_acc_runs": [ + 0.29456599903740643 + ], + "mean_centered_acc_runs": [ + 0.000996625884414063 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-6912.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-6912.json new file mode 100644 index 0000000000000000000000000000000000000000..92d2f915d386882c07f6c8b001e79294ed670e00 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-6912.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6912", + "checkpoint": "checkpoint-6912", + "global_step": 6912, + "tokens_seen": 7247757312, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2634933280223063, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2634933280223063 + ], + "acc": 0.2581159131647082, + "acc_ci": 0.0, + "acc_runs": [ + 0.2581159131647082 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004396273173717708, + "centered_acc": 0.017991104029741727, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017991104029741727 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5359085963003264, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5359085963003264 + ], + "acc": 0.5505984766050055, + "acc_ci": 0.0, + "acc_runs": [ + 0.5505984766050055 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011635700809215522, + "centered_acc": 0.07181719260065278, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.07181719260065278 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2676767676767677, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2676767676767677 + ], + "acc": 0.2697811447811448, + "acc_ci": 0.0, + "acc_runs": [ + 0.2697811447811448 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009085000147099275, + "centered_acc": 0.023569023569023573, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023569023569023573 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26109215017064846, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26109215017064846 + ], + "acc": 0.20051194539249148, + "acc_ci": 0.0, + "acc_runs": [ + 0.20051194539249148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012835523909473708, + "centered_acc": 0.014789533560864617, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014789533560864617 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.138, + "acc_ci": 0.0, + "acc_runs": [ + 0.138 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4846093133385951, + "acc_ci": 0.0, + "acc_runs": [ + 0.4846093133385951 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014045826789783585, + "centered_acc": -0.030781373322809835, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.030781373322809835 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 132609337.44550243, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 132609337.44550243 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.2, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2 + ], + "acc": 0.055, + "acc_ci": 0.0, + "acc_runs": [ + 0.055 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012655439943366655, + "centered_acc": -0.06666666666666665, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.06666666666666665 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.3003618070283415, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2904904255055646, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0045229180757548025, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.3003618070283415 + ], + "mean_acc_runs": [ + 0.2904904255055646 + ], + "mean_centered_acc_runs": [ + -0.0045229180757548025 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-72.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-72.json new file mode 100644 index 0000000000000000000000000000000000000000..2ffb97b87045dad82578a7f96c65b8a59c5a6fc8 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-72.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-72", + "checkpoint": "checkpoint-72", + "global_step": 72, + "tokens_seen": 75497472, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2619000199163513, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2619000199163513 + ], + "acc": 0.2570205138418642, + "acc_ci": 0.0, + "acc_runs": [ + 0.2570205138418642 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004387699525854858, + "centered_acc": 0.015866693221801748, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015866693221801748 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5070729053318824, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5070729053318824 + ], + "acc": 0.5261153427638737, + "acc_ci": 0.0, + "acc_runs": [ + 0.5261153427638737 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011664656918146136, + "centered_acc": 0.014145810663764857, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014145810663764857 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.25547138047138046, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25547138047138046 + ], + "acc": 0.24452861952861954, + "acc_ci": 0.0, + "acc_runs": [ + 0.24452861952861954 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008949113551665389, + "centered_acc": 0.0072951739618406135, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0072951739618406135 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2619453924914676, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2619453924914676 + ], + "acc": 0.19880546075085323, + "acc_ci": 0.0, + "acc_runs": [ + 0.19880546075085323 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012849054826857987, + "centered_acc": 0.015927189988623452, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015927189988623452 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4861878453038674, + "acc_ci": 0.0, + "acc_runs": [ + 0.4861878453038674 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014047122916440271, + "centered_acc": -0.027624309392265234, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.027624309392265234 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 26223.92304201877, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 26223.92304201877 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.171, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.171 + ], + "acc": 0.036, + "acc_ci": 0.0, + "acc_runs": [ + 0.036 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011912216456264588, + "centered_acc": -0.10533333333333332, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10533333333333332 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.288564949701847, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2828945243962778, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.021238650149129682, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.288564949701847 + ], + "mean_acc_runs": [ + 0.2828945243962778 + ], + "mean_centered_acc_runs": [ + -0.021238650149129682 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-7200.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-7200.json new file mode 100644 index 0000000000000000000000000000000000000000..748b3318cc3cf11c62f54ecb40e1c64cde5a21c8 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-7200.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-7200", + "checkpoint": "checkpoint-7200", + "global_step": 7200, + "tokens_seen": 7549747200, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26329416450906196, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26329416450906196 + ], + "acc": 0.2592113124875523, + "acc_ci": 0.0, + "acc_runs": [ + 0.2592113124875523 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004395205528158065, + "centered_acc": 0.017725552678749274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017725552678749274 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.528835690968444, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.528835690968444 + ], + "acc": 0.544613710554951, + "acc_ci": 0.0, + "acc_runs": [ + 0.544613710554951 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011646407809944788, + "centered_acc": 0.05767138193688792, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05767138193688792 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2706228956228956, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2706228956228956 + ], + "acc": 0.26641414141414144, + "acc_ci": 0.0, + "acc_runs": [ + 0.26641414141414144 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009116466166404035, + "centered_acc": 0.0274971941638608, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0274971941638608 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.23890784982935154, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.23890784982935154 + ], + "acc": 0.19368600682593856, + "acc_ci": 0.0, + "acc_runs": [ + 0.19368600682593856 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012461071376316715, + "centered_acc": -0.014789533560864617, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.014789533560864617 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.276, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.276 + ], + "acc": 0.15, + "acc_ci": 0.0, + "acc_runs": [ + 0.15 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02001121929807355, + "centered_acc": 0.0346666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0346666666666667 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48697711128650356, + "acc_ci": 0.0, + "acc_runs": [ + 0.48697711128650356 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014047718393997596, + "centered_acc": -0.026045777426992878, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.026045777426992878 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.6, + "acc_ci": 0.0, + "acc_runs": [ + 0.6 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0492365963917331, + "centered_acc": 0.19999999999999996, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.19999999999999996 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 124825446.74493426, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 124825446.74493426 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.198, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.198 + ], + "acc": 0.053, + "acc_ci": 0.0, + "acc_runs": [ + 0.053 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012607733934175283, + "centered_acc": -0.06933333333333332, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.06933333333333332 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29594343348829216, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2932189744342787, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0017814361475139483, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29594343348829216 + ], + "mean_acc_runs": [ + 0.2932189744342787 + ], + "mean_centered_acc_runs": [ + -0.0017814361475139483 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/checkpoint-864.json b/results/i5-fulle-lm/zero_shot_eval/checkpoint-864.json new file mode 100644 index 0000000000000000000000000000000000000000..511b4c0e8dd662400a5b2645b7161468066a3a57 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/checkpoint-864.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/i5-fulle-lm", + "run": "i5-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-864", + "checkpoint": "checkpoint-864", + "global_step": 864, + "tokens_seen": 905969664, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.255327623979287, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.255327623979287 + ], + "acc": 0.25652260505875324, + "acc_ci": 0.0, + "acc_runs": [ + 0.25652260505875324 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004351540603988709, + "centered_acc": 0.0071034986390493176, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0071034986390493176 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5190424374319913, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5190424374319913 + ], + "acc": 0.544613710554951, + "acc_ci": 0.0, + "acc_runs": [ + 0.544613710554951 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011657360703051524, + "centered_acc": 0.0380848748639826, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0380848748639826 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26725589225589225, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26725589225589225 + ], + "acc": 0.2617845117845118, + "acc_ci": 0.0, + "acc_runs": [ + 0.2617845117845118 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009080463246017426, + "centered_acc": 0.023007856341189663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023007856341189663 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.27559726962457337, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27559726962457337 + ], + "acc": 0.2167235494880546, + "acc_ci": 0.0, + "acc_runs": [ + 0.2167235494880546 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.013057169655761753, + "centered_acc": 0.03412969283276449, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03412969283276449 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.288, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.288 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020271503835075255, + "centered_acc": 0.05066666666666664, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05066666666666664 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4972375690607735, + "acc_ci": 0.0, + "acc_runs": [ + 0.4972375690607735 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052271211616401, + "centered_acc": -0.0055248618784530246, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0055248618784530246 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 2165596.3986663735, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 2165596.3986663735 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.185, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.185 + ], + "acc": 0.041, + "acc_ci": 0.0, + "acc_runs": [ + 0.041 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012285191326386655, + "centered_acc": -0.08666666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08666666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29837053721529067, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2934169407720745, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.002513779517118493, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29837053721529067 + ], + "mean_acc_runs": [ + 0.2934169407720745 + ], + "mean_centered_acc_runs": [ + -0.002513779517118493 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": true, + "max_next_i_train": 5, + "l2r_prefix_enabled": true +} diff --git a/results/i5-fulle-lm/zero_shot_eval/summary.jsonl b/results/i5-fulle-lm/zero_shot_eval/summary.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..8c3ea19975fffc16e4f2dfb5f323e27761838dd2 --- /dev/null +++ b/results/i5-fulle-lm/zero_shot_eval/summary.jsonl @@ -0,0 +1,26 @@ +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-72", "checkpoint": "checkpoint-72", "global_step": 72, "tokens_seen": 75497472, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2619000199163513, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2619000199163513], "acc": 0.2570205138418642, "acc_ci": 0.0, "acc_runs": [0.2570205138418642], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004387699525854858, "centered_acc": 0.015866693221801748, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015866693221801748], "ci": 0.0}, "piqa": {"acc_norm": 0.5070729053318824, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5070729053318824], "acc": 0.5261153427638737, "acc_ci": 0.0, "acc_runs": [0.5261153427638737], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011664656918146136, "centered_acc": 0.014145810663764857, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014145810663764857], "ci": 0.0}, "arc_easy": {"acc_norm": 0.25547138047138046, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25547138047138046], "acc": 0.24452861952861954, "acc_ci": 0.0, "acc_runs": [0.24452861952861954], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008949113551665389, "centered_acc": 0.0072951739618406135, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0072951739618406135], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2619453924914676, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2619453924914676], "acc": 0.19880546075085323, "acc_ci": 0.0, "acc_runs": [0.19880546075085323], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012849054826857987, "centered_acc": 0.015927189988623452, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015927189988623452], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4861878453038674, "acc_ci": 0.0, "acc_runs": [0.4861878453038674], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014047122916440271, "centered_acc": -0.027624309392265234, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.027624309392265234], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 26223.92304201877, "perplexity_ci": 0.0, "perplexity_runs": [26223.92304201877], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.171, "acc_norm_ci": 0.0, "acc_norm_runs": [0.171], "acc": 0.036, "acc_ci": 0.0, "acc_runs": [0.036], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011912216456264588, "centered_acc": -0.10533333333333332, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10533333333333332], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.288564949701847, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2828945243962778, "mean_acc_ci": 0.0, "mean_centered_acc": -0.021238650149129682, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.288564949701847], "mean_acc_runs": [0.2828945243962778], "mean_centered_acc_runs": [-0.021238650149129682], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-288", "checkpoint": "checkpoint-288", "global_step": 288, "tokens_seen": 301989888, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25652260505875324, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25652260505875324], "acc": 0.2531368253335989, "acc_ci": 0.0, "acc_runs": [0.2531368253335989], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043582106894420325, "centered_acc": 0.008696806745004318, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008696806745004318], "ci": 0.0}, "piqa": {"acc_norm": 0.5097932535364527, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5097932535364527], "acc": 0.5353645266594124, "acc_ci": 0.0, "acc_runs": [0.5353645266594124], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011663586263283307, "centered_acc": 0.019586507072905324, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019586507072905324], "ci": 0.0}, "arc_easy": {"acc_norm": 0.25462962962962965, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25462962962962965], "acc": 0.24831649831649832, "acc_ci": 0.0, "acc_runs": [0.24831649831649832], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008939407288589512, "centered_acc": 0.006172839506172867, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006172839506172867], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2721843003412969, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2721843003412969], "acc": 0.20733788395904437, "acc_ci": 0.0, "acc_runs": [0.20733788395904437], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01300660040642361, "centered_acc": 0.029579067121729235, "centered_acc_ci": 0.0, "centered_acc_runs": [0.029579067121729235], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.144, "acc_ci": 0.0, "acc_runs": [0.144], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4909234411996843, "acc_ci": 0.0, "acc_runs": [0.4909234411996843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050170094497657, "centered_acc": -0.01815311760063143, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.01815311760063143], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 100213.09880327753, "perplexity_ci": 0.0, "perplexity_runs": [100213.09880327753], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.177, "acc_norm_ci": 0.0, "acc_norm_runs": [0.177], "acc": 0.039, "acc_ci": 0.0, "acc_runs": [0.039], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012075463420375087, "centered_acc": -0.09733333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09733333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2926882980943554, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2846366637241939, "mean_acc_ci": 0.0, "mean_centered_acc": -0.01631958966008359, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2926882980943554], "mean_acc_runs": [0.2846366637241939], "mean_centered_acc_runs": [-0.01631958966008359], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-576", "checkpoint": "checkpoint-576", "global_step": 576, "tokens_seen": 603979776, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2538338976299542, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2538338976299542], "acc": 0.2562238597888867, "acc_ci": 0.0, "acc_runs": [0.2562238597888867], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004343142545094425, "centered_acc": 0.005111863506605567, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005111863506605567], "ci": 0.0}, "piqa": {"acc_norm": 0.5223068552774756, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5223068552774756], "acc": 0.5353645266594124, "acc_ci": 0.0, "acc_runs": [0.5353645266594124], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011654208652596384, "centered_acc": 0.044613710554951114, "centered_acc_ci": 0.0, "centered_acc_runs": [0.044613710554951114], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26346801346801346, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26346801346801346], "acc": 0.25336700336700335, "acc_ci": 0.0, "acc_runs": [0.25336700336700335], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009039157374497906, "centered_acc": 0.017957351290684615, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017957351290684615], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2883959044368601, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2883959044368601], "acc": 0.2022184300341297, "acc_ci": 0.0, "acc_runs": [0.2022184300341297], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.013238394422428247, "centered_acc": 0.05119453924914678, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05119453924914678], "ci": 0.0}, "openbookqa": {"acc_norm": 0.304, "acc_norm_ci": 0.0, "acc_norm_runs": [0.304], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02059164957122494, "centered_acc": 0.072, "centered_acc_ci": 0.0, "centered_acc_runs": [0.072], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4846093133385951, "acc_ci": 0.0, "acc_runs": [0.4846093133385951], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014045826789783585, "centered_acc": -0.030781373322809835, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.030781373322809835], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 910792.405258101, "perplexity_ci": 0.0, "perplexity_runs": [910792.405258101], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.185, "acc_norm_ci": 0.0, "acc_norm_runs": [0.185], "acc": 0.041, "acc_ci": 0.0, "acc_runs": [0.041], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012285191326386655, "centered_acc": -0.08666666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08666666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.30283411180205055, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28530705949617274, "mean_acc_ci": 0.0, "mean_centered_acc": -0.007777294648965302, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.30283411180205055], "mean_acc_runs": [0.28530705949617274], "mean_centered_acc_runs": [-0.007777294648965302], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-864", "checkpoint": "checkpoint-864", "global_step": 864, "tokens_seen": 905969664, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.255327623979287, "acc_norm_ci": 0.0, "acc_norm_runs": [0.255327623979287], "acc": 0.25652260505875324, "acc_ci": 0.0, "acc_runs": [0.25652260505875324], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004351540603988709, "centered_acc": 0.0071034986390493176, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0071034986390493176], "ci": 0.0}, "piqa": {"acc_norm": 0.5190424374319913, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5190424374319913], "acc": 0.544613710554951, "acc_ci": 0.0, "acc_runs": [0.544613710554951], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011657360703051524, "centered_acc": 0.0380848748639826, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0380848748639826], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26725589225589225, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26725589225589225], "acc": 0.2617845117845118, "acc_ci": 0.0, "acc_runs": [0.2617845117845118], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009080463246017426, "centered_acc": 0.023007856341189663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023007856341189663], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.27559726962457337, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27559726962457337], "acc": 0.2167235494880546, "acc_ci": 0.0, "acc_runs": [0.2167235494880546], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.013057169655761753, "centered_acc": 0.03412969283276449, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03412969283276449], "ci": 0.0}, "openbookqa": {"acc_norm": 0.288, "acc_norm_ci": 0.0, "acc_norm_runs": [0.288], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020271503835075255, "centered_acc": 0.05066666666666664, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05066666666666664], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4972375690607735, "acc_ci": 0.0, "acc_runs": [0.4972375690607735], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052271211616401, "centered_acc": -0.0055248618784530246, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0055248618784530246], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 2165596.3986663735, "perplexity_ci": 0.0, "perplexity_runs": [2165596.3986663735], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.185, "acc_norm_ci": 0.0, "acc_norm_runs": [0.185], "acc": 0.041, "acc_ci": 0.0, "acc_runs": [0.041], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012285191326386655, "centered_acc": -0.08666666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08666666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29837053721529067, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2934169407720745, "mean_acc_ci": 0.0, "mean_centered_acc": -0.002513779517118493, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29837053721529067], "mean_acc_runs": [0.2934169407720745], "mean_centered_acc_runs": [-0.002513779517118493], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-1152", "checkpoint": "checkpoint-1152", "global_step": 1152, "tokens_seen": 1207959552, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2560246962756423, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2560246962756423], "acc": 0.2568213503286198, "acc_ci": 0.0, "acc_runs": [0.2568213503286198], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004355436696716235, "centered_acc": 0.008032928367523043, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008032928367523043], "ci": 0.0}, "piqa": {"acc_norm": 0.5223068552774756, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5223068552774756], "acc": 0.5424374319912949, "acc_ci": 0.0, "acc_runs": [0.5424374319912949], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011654208652596384, "centered_acc": 0.044613710554951114, "centered_acc_ci": 0.0, "centered_acc_runs": [0.044613710554951114], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2689393939393939, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2689393939393939], "acc": 0.26515151515151514, "acc_ci": 0.0, "acc_runs": [0.26515151515151514], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009098548093009251, "centered_acc": 0.02525252525252523, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02525252525252523], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.27474402730375425, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27474402730375425], "acc": 0.22184300341296928, "acc_ci": 0.0, "acc_runs": [0.22184300341296928], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.013044617212771207, "centered_acc": 0.03299203640500566, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03299203640500566], "ci": 0.0}, "openbookqa": {"acc_norm": 0.284, "acc_norm_ci": 0.0, "acc_norm_runs": [0.284], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020186703693570774, "centered_acc": 0.0453333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0453333333333333], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.47592738752959746, "acc_ci": 0.0, "acc_runs": [0.47592738752959746], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014036189665395169, "centered_acc": -0.04814522494080509, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.04814522494080509], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 2547853.616903067, "perplexity_ci": 0.0, "perplexity_runs": [2547853.616903067], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.188, "acc_norm_ci": 0.0, "acc_norm_runs": [0.188], "acc": 0.04, "acc_ci": 0.0, "acc_runs": [0.04], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01236158601510376, "centered_acc": -0.08266666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08266666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29900249546604435, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2882468150187697, "mean_acc_ci": 0.0, "mean_centered_acc": -0.013112492682970298, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29900249546604435], "mean_acc_runs": [0.2882468150187697], "mean_centered_acc_runs": [-0.013112492682970298], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-1440", "checkpoint": "checkpoint-1440", "global_step": 1440, "tokens_seen": 1509949440, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25473013343955386, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25473013343955386], "acc": 0.2559251145190201, "acc_ci": 0.0, "acc_runs": [0.2559251145190201], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043481894593367845, "centered_acc": 0.006306844586071818, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006306844586071818], "ci": 0.0}, "piqa": {"acc_norm": 0.5250272034820457, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5250272034820457], "acc": 0.5505984766050055, "acc_ci": 0.0, "acc_runs": [0.5505984766050055], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011651200967623562, "centered_acc": 0.05005440696409136, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05005440696409136], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26936026936026936, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26936026936026936], "acc": 0.2596801346801347, "acc_ci": 0.0, "acc_runs": [0.2596801346801347], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009103043207756808, "centered_acc": 0.02581369248035914, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02581369248035914], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2781569965870307, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2781569965870307], "acc": 0.22098976109215018, "acc_ci": 0.0, "acc_runs": [0.22098976109215018], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.013094469919538899, "centered_acc": 0.037542662116040924, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037542662116040924], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4956590370955012, "acc_ci": 0.0, "acc_runs": [0.4956590370955012], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051956064076884, "centered_acc": -0.008681925808997626, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.008681925808997626], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 2931799.134003098, "perplexity_ci": 0.0, "perplexity_runs": [2931799.134003098], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29837910047815, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2908139985765512, "mean_acc_ci": 0.0, "mean_centered_acc": -0.009080303272040784, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29837910047815], "mean_acc_runs": [0.2908139985765512], "mean_centered_acc_runs": [-0.009080303272040784], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-1728", "checkpoint": "checkpoint-1728", "global_step": 1728, "tokens_seen": 1811939328, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26010754829715194, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26010754829715194], "acc": 0.2584146584345748, "acc_ci": 0.0, "acc_runs": [0.2584146584345748], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004377965074211591, "centered_acc": 0.013476731062869248, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013476731062869248], "ci": 0.0}, "piqa": {"acc_norm": 0.5239390642002176, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5239390642002176], "acc": 0.5505984766050055, "acc_ci": 0.0, "acc_runs": [0.5505984766050055], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165244562107948, "centered_acc": 0.04787812840043526, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04787812840043526], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2689393939393939, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2689393939393939], "acc": 0.2601010101010101, "acc_ci": 0.0, "acc_runs": [0.2601010101010101], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009098548093009251, "centered_acc": 0.02525252525252523, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02525252525252523], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2696245733788396, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2696245733788396], "acc": 0.21843003412969283, "acc_ci": 0.0, "acc_runs": [0.21843003412969283], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012968040686869096, "centered_acc": 0.026166097838452806, "centered_acc_ci": 0.0, "centered_acc_runs": [0.026166097838452806], "ci": 0.0}, "openbookqa": {"acc_norm": 0.294, "acc_norm_ci": 0.0, "acc_norm_runs": [0.294], "acc": 0.154, "acc_ci": 0.0, "acc_runs": [0.154], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02039509548493655, "centered_acc": 0.058666666666666645, "centered_acc_ci": 0.0, "centered_acc_runs": [0.058666666666666645], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.494869771112865, "acc_ci": 0.0, "acc_runs": [0.494869771112865], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051745961790525, "centered_acc": -0.010260457774269982, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.010260457774269982], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.56, "acc_ci": 0.0, "acc_runs": [0.56], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049888765156985884, "centered_acc": 0.1200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1200000000000001], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 6849435.695362101, "perplexity_ci": 0.0, "perplexity_runs": [6849435.695362101], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.182, "acc_norm_ci": 0.0, "acc_norm_runs": [0.182], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0122075806376622, "centered_acc": -0.09066666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09066666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29976842996926717, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29207014121568486, "mean_acc_ci": 0.0, "mean_centered_acc": -0.005879116852509636, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29976842996926717], "mean_acc_runs": [0.29207014121568486], "mean_centered_acc_runs": [-0.005879116852509636], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2016", "checkpoint": "checkpoint-2016", "global_step": 2016, "tokens_seen": 2113929216, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25941047600079664, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25941047600079664], "acc": 0.2584146584345748, "acc_ci": 0.0, "acc_runs": [0.2584146584345748], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004374153847826732, "centered_acc": 0.012547301334395522, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012547301334395522], "ci": 0.0}, "piqa": {"acc_norm": 0.5326441784548422, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5326441784548422], "acc": 0.5549510337323177, "acc_ci": 0.0, "acc_runs": [0.5549510337323177], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011640934420031744, "centered_acc": 0.06528835690968449, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06528835690968449], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26515151515151514, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26515151515151514], "acc": 0.26430976430976433, "acc_ci": 0.0, "acc_runs": [0.26430976430976433], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009057621139172679, "centered_acc": 0.020202020202020183, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020202020202020183], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26023890784982934, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26023890784982934], "acc": 0.20819112627986347, "acc_ci": 0.0, "acc_runs": [0.20819112627986347], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012821930225112672, "centered_acc": 0.013651877133105783, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013651877133105783], "ci": 0.0}, "openbookqa": {"acc_norm": 0.284, "acc_norm_ci": 0.0, "acc_norm_runs": [0.284], "acc": 0.156, "acc_ci": 0.0, "acc_runs": [0.156], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020186703693570774, "centered_acc": 0.0453333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0453333333333333], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5067087608524072, "acc_ci": 0.0, "acc_runs": [0.5067087608524072], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051220692330247, "centered_acc": 0.013417521704814472, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013417521704814472], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 5755669.654358351, "perplexity_ci": 0.0, "perplexity_runs": [5755669.654358351], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2970741795761639, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29438628053826277, "mean_acc_ci": 0.0, "mean_centered_acc": -0.002776073981693971, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2970741795761639], "mean_acc_runs": [0.29438628053826277], "mean_centered_acc_runs": [-0.002776073981693971], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2304", "checkpoint": "checkpoint-2304", "global_step": 2304, "tokens_seen": 2415919104, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25941047600079664, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25941047600079664], "acc": 0.2577175861382195, "acc_ci": 0.0, "acc_runs": [0.2577175861382195], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004374153847826732, "centered_acc": 0.012547301334395522, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012547301334395522], "ci": 0.0}, "piqa": {"acc_norm": 0.5250272034820457, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5250272034820457], "acc": 0.5527747551686616, "acc_ci": 0.0, "acc_runs": [0.5527747551686616], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011651200967623562, "centered_acc": 0.05005440696409136, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05005440696409136], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2718855218855219, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2718855218855219], "acc": 0.26430976430976433, "acc_ci": 0.0, "acc_runs": [0.26430976430976433], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009129795867310605, "centered_acc": 0.02918069584736253, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02918069584736253], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26791808873720135, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26791808873720135], "acc": 0.20648464163822525, "acc_ci": 0.0, "acc_runs": [0.20648464163822525], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012942030195136482, "centered_acc": 0.023890784982935138, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023890784982935138], "ci": 0.0}, "openbookqa": {"acc_norm": 0.278, "acc_norm_ci": 0.0, "acc_norm_runs": [0.278], "acc": 0.148, "acc_ci": 0.0, "acc_runs": [0.148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020055833888070865, "centered_acc": 0.037333333333333364, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037333333333333364], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.49171270718232046, "acc_ci": 0.0, "acc_runs": [0.49171270718232046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405055532282417, "centered_acc": -0.016574585635359074, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.016574585635359074], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.62, "acc_ci": 0.0, "acc_runs": [0.62], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04878317312145634, "centered_acc": 0.24, "centered_acc_ci": 0.0, "centered_acc_runs": [0.24], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 8472343.17966308, "perplexity_ci": 0.0, "perplexity_runs": [8472343.17966308], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29720688168426096, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2965286916210891, "mean_acc_ci": 0.0, "mean_centered_acc": 0.0045563178193510495, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29720688168426096], "mean_acc_runs": [0.2965286916210891], "mean_centered_acc_runs": [0.0045563178193510495], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2592", "checkpoint": "checkpoint-2592", "global_step": 2592, "tokens_seen": 2717908992, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25960963951404104, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25960963951404104], "acc": 0.2559251145190201, "acc_ci": 0.0, "acc_runs": [0.2559251145190201], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004375244237044889, "centered_acc": 0.012812852685388046, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012812852685388046], "ci": 0.0}, "piqa": {"acc_norm": 0.529923830250272, "acc_norm_ci": 0.0, "acc_norm_runs": [0.529923830250272], "acc": 0.5462459194776932, "acc_ci": 0.0, "acc_runs": [0.5462459194776932], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011644913435420134, "centered_acc": 0.05984766050054402, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05984766050054402], "ci": 0.0}, "arc_easy": {"acc_norm": 0.27146464646464646, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27146464646464646], "acc": 0.2706228956228956, "acc_ci": 0.0, "acc_runs": [0.2706228956228956], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00912536297036068, "centered_acc": 0.028619528619528618, "centered_acc_ci": 0.0, "centered_acc_runs": [0.028619528619528618], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.257679180887372, "acc_norm_ci": 0.0, "acc_norm_runs": [0.257679180887372], "acc": 0.2022184300341297, "acc_ci": 0.0, "acc_runs": [0.2022184300341297], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012780770562768334, "centered_acc": 0.010238907849829356, "centered_acc_ci": 0.0, "centered_acc_runs": [0.010238907849829356], "ci": 0.0}, "openbookqa": {"acc_norm": 0.284, "acc_norm_ci": 0.0, "acc_norm_runs": [0.284], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020186703693570774, "centered_acc": 0.0453333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0453333333333333], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4956590370955012, "acc_ci": 0.0, "acc_runs": [0.4956590370955012], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051956064076884, "centered_acc": -0.008681925808997626, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.008681925808997626], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.62, "acc_ci": 0.0, "acc_runs": [0.62], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04878317312145634, "centered_acc": 0.24, "centered_acc_ci": 0.0, "centered_acc_runs": [0.24], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 11561444.171015542, "perplexity_ci": 0.0, "perplexity_runs": [11561444.171015542], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.193, "acc_norm_ci": 0.0, "acc_norm_runs": [0.193], "acc": 0.048, "acc_ci": 0.0, "acc_runs": [0.048], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012486268734370044, "centered_acc": -0.076, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.076], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29927954951938857, "mean_acc_norm_ci": 0.0, "mean_acc": 0.296895885852294, "mean_acc_ci": 0.0, "mean_centered_acc": 0.0076383645252251515, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29927954951938857], "mean_acc_runs": [0.296895885852294], "mean_centered_acc_runs": [0.0076383645252251515], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-2880", "checkpoint": "checkpoint-2880", "global_step": 2880, "tokens_seen": 3019898880, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2599083847839076, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2599083847839076], "acc": 0.25731925911173076, "acc_ci": 0.0, "acc_runs": [0.25731925911173076], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004376877619233649, "centered_acc": 0.013211179711876797, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013211179711876797], "ci": 0.0}, "piqa": {"acc_norm": 0.5201305767138193, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5201305767138193], "acc": 0.5391730141458106, "acc_ci": 0.0, "acc_runs": [0.5391730141458106], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011656365410780507, "centered_acc": 0.040261153427638696, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040261153427638696], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26304713804713803, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26304713804713803], "acc": 0.2647306397306397, "acc_ci": 0.0, "acc_runs": [0.2647306397306397], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009034514898865944, "centered_acc": 0.017396184062850706, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017396184062850706], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.257679180887372, "acc_norm_ci": 0.0, "acc_norm_runs": [0.257679180887372], "acc": 0.20563139931740615, "acc_ci": 0.0, "acc_runs": [0.20563139931740615], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012780770562768334, "centered_acc": 0.010238907849829356, "centered_acc_ci": 0.0, "centered_acc_runs": [0.010238907849829356], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.148, "acc_ci": 0.0, "acc_runs": [0.148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5082872928176796, "acc_ci": 0.0, "acc_runs": [0.5082872928176796], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405055532282417, "centered_acc": 0.016574585635359185, "centered_acc_ci": 0.0, "centered_acc_runs": [0.016574585635359185], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 12798579.36306335, "perplexity_ci": 0.0, "perplexity_runs": [12798579.36306335], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.19, "acc_norm_ci": 0.0, "acc_norm_runs": [0.19], "acc": 0.049, "acc_ci": 0.0, "acc_runs": [0.049], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012411851354816256, "centered_acc": -0.08, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29412754673870617, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29404290668969674, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0015270073072271835, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29412754673870617], "mean_acc_runs": [0.29404290668969674], "mean_centered_acc_runs": [-0.0015270073072271835], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-3168", "checkpoint": "checkpoint-3168", "global_step": 3168, "tokens_seen": 3321888768, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2581159131647082, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2581159131647082], "acc": 0.25731925911173076, "acc_ci": 0.0, "acc_runs": [0.25731925911173076], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004367037632204837, "centered_acc": 0.010821217552944296, "centered_acc_ci": 0.0, "centered_acc_runs": [0.010821217552944296], "ci": 0.0}, "piqa": {"acc_norm": 0.544613710554951, "acc_norm_ci": 0.0, "acc_norm_runs": [0.544613710554951], "acc": 0.5500544069640914, "acc_ci": 0.0, "acc_runs": [0.5500544069640914], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011619292444156876, "centered_acc": 0.089227421109902, "centered_acc_ci": 0.0, "centered_acc_runs": [0.089227421109902], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2676767676767677, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2676767676767677], "acc": 0.26936026936026936, "acc_ci": 0.0, "acc_runs": [0.26936026936026936], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009085000147099275, "centered_acc": 0.023569023569023573, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023569023569023573], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25426621160409557, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25426621160409557], "acc": 0.2022184300341297, "acc_ci": 0.0, "acc_runs": [0.2022184300341297], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012724999945157845, "centered_acc": 0.0056882821387940945, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0056882821387940945], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5067087608524072, "acc_ci": 0.0, "acc_runs": [0.5067087608524072], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051220692330247, "centered_acc": 0.013417521704814472, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013417521704814472], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 19793869.552791856, "perplexity_ci": 0.0, "perplexity_runs": [19793869.552791856], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.196, "acc_norm_ci": 0.0, "acc_norm_runs": [0.196], "acc": 0.05, "acc_ci": 0.0, "acc_runs": [0.05], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012559527926707349, "centered_acc": -0.072, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.072], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.3011121005000871, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2945948588096329, "mean_acc_ci": 0.0, "mean_centered_acc": 0.003922043291430997, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.3011121005000871], "mean_acc_runs": [0.2945948588096329], "mean_centered_acc_runs": [0.003922043291430997], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-3456", "checkpoint": "checkpoint-3456", "global_step": 3456, "tokens_seen": 3623878656, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2562238597888867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2562238597888867], "acc": 0.25692093208524197, "acc_ci": 0.0, "acc_runs": [0.25692093208524197], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004356547185846709, "centered_acc": 0.00829847971851557, "centered_acc_ci": 0.0, "centered_acc_runs": [0.00829847971851557], "ci": 0.0}, "piqa": {"acc_norm": 0.5321001088139282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5321001088139282], "acc": 0.5495103373231773, "acc_ci": 0.0, "acc_runs": [0.5495103373231773], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011641758014820223, "centered_acc": 0.06420021762785644, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06420021762785644], "ci": 0.0}, "arc_easy": {"acc_norm": 0.27146464646464646, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27146464646464646], "acc": 0.27441077441077444, "acc_ci": 0.0, "acc_runs": [0.27441077441077444], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00912536297036068, "centered_acc": 0.028619528619528618, "centered_acc_ci": 0.0, "centered_acc_runs": [0.028619528619528618], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26023890784982934, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26023890784982934], "acc": 0.20477815699658702, "acc_ci": 0.0, "acc_runs": [0.20477815699658702], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012821930225112672, "centered_acc": 0.013651877133105783, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013651877133105783], "ci": 0.0}, "openbookqa": {"acc_norm": 0.276, "acc_norm_ci": 0.0, "acc_norm_runs": [0.276], "acc": 0.154, "acc_ci": 0.0, "acc_runs": [0.154], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02001121929807355, "centered_acc": 0.0346666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0346666666666667], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5114443567482242, "acc_ci": 0.0, "acc_runs": [0.5114443567482242], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014048804199859391, "centered_acc": 0.022888713496448387, "centered_acc_ci": 0.0, "centered_acc_runs": [0.022888713496448387], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.56, "acc_ci": 0.0, "acc_runs": [0.56], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049888765156985884, "centered_acc": 0.1200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1200000000000001], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 24398718.721468586, "perplexity_ci": 0.0, "perplexity_runs": [24398718.721468586], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.19, "acc_norm_ci": 0.0, "acc_norm_runs": [0.19], "acc": 0.05, "acc_ci": 0.0, "acc_runs": [0.05], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012411851354816256, "centered_acc": -0.08, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29767125381954845, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29393520193377054, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0034555103544975296, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29767125381954845], "mean_acc_runs": [0.29393520193377054], "mean_centered_acc_runs": [-0.0034555103544975296], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-3744", "checkpoint": "checkpoint-3744", "global_step": 3744, "tokens_seen": 3925868544, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2610037841067516, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2610037841067516], "acc": 0.25642302330213107, "acc_ci": 0.0, "acc_runs": [0.25642302330213107], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043828441286430965, "centered_acc": 0.0146717121423355, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0146717121423355], "ci": 0.0}, "piqa": {"acc_norm": 0.5315560391730142, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5315560391730142], "acc": 0.544613710554951, "acc_ci": 0.0, "acc_runs": [0.544613710554951], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011642567710878958, "centered_acc": 0.06311207834602839, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06311207834602839], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2718855218855219, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2718855218855219], "acc": 0.2697811447811448, "acc_ci": 0.0, "acc_runs": [0.2697811447811448], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009129795867310605, "centered_acc": 0.02918069584736253, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02918069584736253], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2508532423208191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2508532423208191], "acc": 0.1945392491467577, "acc_ci": 0.0, "acc_runs": [0.1945392491467577], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012668198621315347, "centered_acc": 0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.162, "acc_ci": 0.0, "acc_runs": [0.162], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5019731649565904, "acc_ci": 0.0, "acc_runs": [0.5019731649565904], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052376259225643, "centered_acc": 0.003946329913180779, "centered_acc_ci": 0.0, "centered_acc_runs": [0.003946329913180779], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.61, "acc_ci": 0.0, "acc_runs": [0.61], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04902071300001973, "centered_acc": 0.21999999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.21999999999999997], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 19344082.739549376, "perplexity_ci": 0.0, "perplexity_runs": [19344082.739549376], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.186, "acc_norm_ci": 0.0, "acc_norm_runs": [0.186], "acc": 0.053, "acc_ci": 0.0, "acc_runs": [0.053], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012310790208412926, "centered_acc": -0.08533333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08533333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29688309791435114, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29706177545152757, "mean_acc_ci": 0.0, "mean_centered_acc": 0.004810006987859258, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29688309791435114], "mean_acc_runs": [0.29706177545152757], "mean_centered_acc_runs": [0.004810006987859258], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4032", "checkpoint": "checkpoint-4032", "global_step": 4032, "tokens_seen": 4227858432, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25851424019119695, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25851424019119695], "acc": 0.2559251145190201, "acc_ci": 0.0, "acc_runs": [0.2559251145190201], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004369232540126326, "centered_acc": 0.011352320254929271, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011352320254929271], "ci": 0.0}, "piqa": {"acc_norm": 0.529379760609358, "acc_norm_ci": 0.0, "acc_norm_runs": [0.529379760609358], "acc": 0.5375408052230686, "acc_ci": 0.0, "acc_runs": [0.5375408052230686], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01164566756505079, "centered_acc": 0.05875952121871597, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05875952121871597], "ci": 0.0}, "arc_easy": {"acc_norm": 0.273989898989899, "acc_norm_ci": 0.0, "acc_norm_runs": [0.273989898989899], "acc": 0.2697811447811448, "acc_ci": 0.0, "acc_runs": [0.2697811447811448], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00915180590154416, "centered_acc": 0.03198653198653201, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03198653198653201], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2619453924914676, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2619453924914676], "acc": 0.20477815699658702, "acc_ci": 0.0, "acc_runs": [0.20477815699658702], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012849054826857987, "centered_acc": 0.015927189988623452, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015927189988623452], "ci": 0.0}, "openbookqa": {"acc_norm": 0.278, "acc_norm_ci": 0.0, "acc_norm_runs": [0.278], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020055833888070865, "centered_acc": 0.037333333333333364, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037333333333333364], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5035516969218626, "acc_ci": 0.0, "acc_runs": [0.5035516969218626], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052131146915852, "centered_acc": 0.00710339384372527, "centered_acc_ci": 0.0, "centered_acc_runs": [0.00710339384372527], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 22451524.063414067, "perplexity_ci": 0.0, "perplexity_runs": [22451524.063414067], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.19, "acc_norm_ci": 0.0, "acc_norm_runs": [0.19], "acc": 0.049, "acc_ci": 0.0, "acc_runs": [0.049], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012411851354816256, "centered_acc": -0.08, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29863821538032026, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29208643802153833, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0023291984251933473, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29863821538032026], "mean_acc_runs": [0.29208643802153833], "mean_centered_acc_runs": [-0.0023291984251933473], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4320", "checkpoint": "checkpoint-4320", "global_step": 4320, "tokens_seen": 4529848320, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26020713005377416, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26020713005377416], "acc": 0.25801633140808605, "acc_ci": 0.0, "acc_runs": [0.25801633140808605], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004378508362084079, "centered_acc": 0.013609506738365548, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013609506738365548], "ci": 0.0}, "piqa": {"acc_norm": 0.5212187159956474, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5212187159956474], "acc": 0.544613710554951, "acc_ci": 0.0, "acc_runs": [0.544613710554951], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011655314732288771, "centered_acc": 0.042437431991294794, "centered_acc_ci": 0.0, "centered_acc_runs": [0.042437431991294794], "ci": 0.0}, "arc_easy": {"acc_norm": 0.27146464646464646, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27146464646464646], "acc": 0.2697811447811448, "acc_ci": 0.0, "acc_runs": [0.2697811447811448], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00912536297036068, "centered_acc": 0.028619528619528618, "centered_acc_ci": 0.0, "centered_acc_runs": [0.028619528619528618], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26023890784982934, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26023890784982934], "acc": 0.1962457337883959, "acc_ci": 0.0, "acc_runs": [0.1962457337883959], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012821930225112672, "centered_acc": 0.013651877133105783, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013651877133105783], "ci": 0.0}, "openbookqa": {"acc_norm": 0.278, "acc_norm_ci": 0.0, "acc_norm_runs": [0.278], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020055833888070865, "centered_acc": 0.037333333333333364, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037333333333333364], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.49171270718232046, "acc_ci": 0.0, "acc_runs": [0.49171270718232046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405055532282417, "centered_acc": -0.016574585635359074, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.016574585635359074], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.6, "acc_ci": 0.0, "acc_runs": [0.6], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0492365963917331, "centered_acc": 0.19999999999999996, "centered_acc_ci": 0.0, "centered_acc_runs": [0.19999999999999996], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 42494483.9688464, "perplexity_ci": 0.0, "perplexity_runs": [42494483.9688464], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.19, "acc_norm_ci": 0.0, "acc_norm_runs": [0.19], "acc": 0.054, "acc_ci": 0.0, "acc_runs": [0.054], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012411851354816256, "centered_acc": -0.08, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2968549000606496, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29506570894885986, "mean_acc_ci": 0.0, "mean_centered_acc": -0.00048310936359226595, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2968549000606496], "mean_acc_runs": [0.29506570894885986], "mean_centered_acc_runs": [-0.00048310936359226595], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4608", "checkpoint": "checkpoint-4608", "global_step": 4608, "tokens_seen": 4831838208, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.261202947619996, "acc_norm_ci": 0.0, "acc_norm_runs": [0.261202947619996], "acc": 0.2593108942441745, "acc_ci": 0.0, "acc_runs": [0.2593108942441745], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004383925147478856, "centered_acc": 0.014937263493328024, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014937263493328024], "ci": 0.0}, "piqa": {"acc_norm": 0.5321001088139282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5321001088139282], "acc": 0.5495103373231773, "acc_ci": 0.0, "acc_runs": [0.5495103373231773], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011641758014820223, "centered_acc": 0.06420021762785644, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06420021762785644], "ci": 0.0}, "arc_easy": {"acc_norm": 0.265993265993266, "acc_norm_ci": 0.0, "acc_norm_runs": [0.265993265993266], "acc": 0.2676767676767677, "acc_ci": 0.0, "acc_runs": [0.2676767676767677], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009066789565615746, "centered_acc": 0.021324354657688005, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021324354657688005], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2525597269624573, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2525597269624573], "acc": 0.2022184300341297, "acc_ci": 0.0, "acc_runs": [0.2022184300341297], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012696728980207687, "centered_acc": 0.003412969283276427, "centered_acc_ci": 0.0, "centered_acc_runs": [0.003412969283276427], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4846093133385951, "acc_ci": 0.0, "acc_runs": [0.4846093133385951], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014045826789783585, "centered_acc": -0.030781373322809835, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.030781373322809835], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 28706360.124189243, "perplexity_ci": 0.0, "perplexity_runs": [28706360.124189243], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.192, "acc_norm_ci": 0.0, "acc_norm_runs": [0.192], "acc": 0.052, "acc_ci": 0.0, "acc_runs": [0.052], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01246159264666003, "centered_acc": -0.07733333333333332, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07733333333333332], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29830934156494127, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29316132043905446, "mean_acc_ci": 0.0, "mean_centered_acc": -0.004407219782954861, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29830934156494127], "mean_acc_runs": [0.29316132043905446], "mean_centered_acc_runs": [-0.004407219782954861], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-4896", "checkpoint": "checkpoint-4896", "global_step": 4896, "tokens_seen": 5133828096, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2614021111332404, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2614021111332404], "acc": 0.2592113124875523, "acc_ci": 0.0, "acc_runs": [0.2592113124875523], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004385004998923735, "centered_acc": 0.015202814844320548, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015202814844320548], "ci": 0.0}, "piqa": {"acc_norm": 0.5348204570184983, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5348204570184983], "acc": 0.544069640914037, "acc_ci": 0.0, "acc_runs": [0.544069640914037], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011637500993816005, "centered_acc": 0.06964091403699668, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06964091403699668], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2647306397306397, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2647306397306397], "acc": 0.2702020202020202, "acc_ci": 0.0, "acc_runs": [0.2702020202020202], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009053021086174166, "centered_acc": 0.019640852974186274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019640852974186274], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26706484641638223, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26706484641638223], "acc": 0.19880546075085323, "acc_ci": 0.0, "acc_runs": [0.19880546075085323], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01292893319649643, "centered_acc": 0.022753128555176305, "centered_acc_ci": 0.0, "centered_acc_runs": [0.022753128555176305], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.148, "acc_ci": 0.0, "acc_runs": [0.148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4909234411996843, "acc_ci": 0.0, "acc_runs": [0.4909234411996843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050170094497657, "centered_acc": -0.01815311760063143, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.01815311760063143], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 55221746.34222039, "perplexity_ci": 0.0, "perplexity_runs": [55221746.34222039], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.193, "acc_norm_ci": 0.0, "acc_norm_runs": [0.193], "acc": 0.055, "acc_ci": 0.0, "acc_runs": [0.055], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012486268734370044, "centered_acc": -0.076, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.076], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29916967571646014, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2934499337327848, "mean_acc_ci": 0.0, "mean_centered_acc": 0.0001843907063832223, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29916967571646014], "mean_acc_runs": [0.2934499337327848], "mean_centered_acc_runs": [0.0001843907063832223], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-5184", "checkpoint": "checkpoint-5184", "global_step": 5184, "tokens_seen": 5435817984, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2605058753236407, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2605058753236407], "acc": 0.26020713005377416, "acc_ci": 0.0, "acc_runs": [0.26020713005377416], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004380136468544087, "centered_acc": 0.014007833764854297, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014007833764854297], "ci": 0.0}, "piqa": {"acc_norm": 0.5397170837867247, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5397170837867247], "acc": 0.5462459194776932, "acc_ci": 0.0, "acc_runs": [0.5462459194776932], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011628961491718543, "centered_acc": 0.07943416757344934, "centered_acc_ci": 0.0, "centered_acc_runs": [0.07943416757344934], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26515151515151514, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26515151515151514], "acc": 0.26262626262626265, "acc_ci": 0.0, "acc_runs": [0.26262626262626265], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009057621139172679, "centered_acc": 0.020202020202020183, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020202020202020183], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24829351535836178, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24829351535836178], "acc": 0.1945392491467577, "acc_ci": 0.0, "acc_runs": [0.1945392491467577], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012624912868089679, "centered_acc": -0.0022753128555176305, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0022753128555176305], "ci": 0.0}, "openbookqa": {"acc_norm": 0.264, "acc_norm_ci": 0.0, "acc_norm_runs": [0.264], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01973288558592215, "centered_acc": 0.018666666666666682, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018666666666666682], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4909234411996843, "acc_ci": 0.0, "acc_runs": [0.4909234411996843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050170094497657, "centered_acc": -0.01815311760063143, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.01815311760063143], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.61, "acc_ci": 0.0, "acc_runs": [0.61], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04902071300001973, "centered_acc": 0.21999999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.21999999999999997], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 49500622.3597079, "perplexity_ci": 0.0, "perplexity_runs": [49500622.3597079], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.19, "acc_norm_ci": 0.0, "acc_norm_runs": [0.19], "acc": 0.049, "acc_ci": 0.0, "acc_runs": [0.049], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012411851354816256, "centered_acc": -0.08, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2946113316033737, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29438294642778723, "mean_acc_ci": 0.0, "mean_centered_acc": 0.0009396868109157833, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2946113316033737], "mean_acc_runs": [0.29438294642778723], "mean_centered_acc_runs": [0.0009396868109157833], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-5472", "checkpoint": "checkpoint-5472", "global_step": 5472, "tokens_seen": 5737807872, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2622983469428401, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2622983469428401], "acc": 0.26070503883688506, "acc_ci": 0.0, "acc_runs": [0.26070503883688506], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004389849907040485, "centered_acc": 0.0163977959237868, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0163977959237868], "ci": 0.0}, "piqa": {"acc_norm": 0.5364526659412405, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5364526659412405], "acc": 0.5418933623503809, "acc_ci": 0.0, "acc_runs": [0.5418933623503809], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01163477983787243, "centered_acc": 0.07290533188248105, "centered_acc_ci": 0.0, "centered_acc_runs": [0.07290533188248105], "ci": 0.0}, "arc_easy": {"acc_norm": 0.25841750841750843, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25841750841750843], "acc": 0.2689393939393939, "acc_ci": 0.0, "acc_runs": [0.2689393939393939], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008982741341291388, "centered_acc": 0.011223344556677914, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011223344556677914], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2525597269624573, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2525597269624573], "acc": 0.2030716723549488, "acc_ci": 0.0, "acc_runs": [0.2030716723549488], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012696728980207687, "centered_acc": 0.003412969283276427, "centered_acc_ci": 0.0, "centered_acc_runs": [0.003412969283276427], "ci": 0.0}, "openbookqa": {"acc_norm": 0.278, "acc_norm_ci": 0.0, "acc_norm_runs": [0.278], "acc": 0.15, "acc_ci": 0.0, "acc_runs": [0.15], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020055833888070865, "centered_acc": 0.037333333333333364, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037333333333333364], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4877663772691397, "acc_ci": 0.0, "acc_runs": [0.4877663772691397], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01404827882040573, "centered_acc": -0.024467245461720633, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.024467245461720633], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 46741535.6698555, "perplexity_ci": 0.0, "perplexity_runs": [46741535.6698555], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.197, "acc_norm_ci": 0.0, "acc_norm_runs": [0.197], "acc": 0.053, "acc_ci": 0.0, "acc_runs": [0.053], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012583693787968194, "centered_acc": -0.07066666666666666, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07066666666666666], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29745470804400775, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2933663306524449, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0019206904001590173, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29745470804400775], "mean_acc_runs": [0.2933663306524449], "mean_centered_acc_runs": [-0.0019206904001590173], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-5760", "checkpoint": "checkpoint-5760", "global_step": 5760, "tokens_seen": 6039797760, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2620991834295957, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2620991834295957], "acc": 0.26110336586337385, "acc_ci": 0.0, "acc_runs": [0.26110336586337385], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043887752982100766, "centered_acc": 0.016132244572794274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.016132244572794274], "ci": 0.0}, "piqa": {"acc_norm": 0.5457018498367792, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5457018498367792], "acc": 0.5571273122959739, "acc_ci": 0.0, "acc_runs": [0.5571273122959739], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011616990094583295, "centered_acc": 0.09140369967355833, "centered_acc_ci": 0.0, "centered_acc_runs": [0.09140369967355833], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2601010101010101, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2601010101010101], "acc": 0.26346801346801346, "acc_ci": 0.0, "acc_runs": [0.26346801346801346], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009001718541079735, "centered_acc": 0.013468013468013481, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013468013468013481], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25170648464163825, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25170648464163825], "acc": 0.20733788395904437, "acc_ci": 0.0, "acc_runs": [0.20733788395904437], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012682496334042911, "centered_acc": 0.0022753128555176674, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0022753128555176674], "ci": 0.0}, "openbookqa": {"acc_norm": 0.268, "acc_norm_ci": 0.0, "acc_norm_runs": [0.268], "acc": 0.15, "acc_ci": 0.0, "acc_runs": [0.15], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019827714859587477, "centered_acc": 0.02400000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02400000000000002], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48697711128650356, "acc_ci": 0.0, "acc_runs": [0.48697711128650356], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014047718393997596, "centered_acc": -0.026045777426992878, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.026045777426992878], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 60888305.54860191, "perplexity_ci": 0.0, "perplexity_runs": [60888305.54860191], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.19, "acc_norm_ci": 0.0, "acc_norm_runs": [0.19], "acc": 0.053, "acc_ci": 0.0, "acc_runs": [0.053], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012411851354816256, "centered_acc": -0.08, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2962680880015039, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29473011486466094, "mean_acc_ci": 0.0, "mean_centered_acc": -0.002465731478856504, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2962680880015039], "mean_acc_runs": [0.29473011486466094], "mean_centered_acc_runs": [-0.002465731478856504], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6048", "checkpoint": "checkpoint-6048", "global_step": 6048, "tokens_seen": 6341787648, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26319458275243973, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26319458275243973], "acc": 0.2614021111332404, "acc_ci": 0.0, "acc_runs": [0.2614021111332404], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004394671271021473, "centered_acc": 0.017592777003252975, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017592777003252975], "ci": 0.0}, "piqa": {"acc_norm": 0.5353645266594124, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5353645266594124], "acc": 0.5478781284004353, "acc_ci": 0.0, "acc_runs": [0.5478781284004353], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011636607860111725, "centered_acc": 0.07072905331882473, "centered_acc_ci": 0.0, "centered_acc_runs": [0.07072905331882473], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26136363636363635, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26136363636363635], "acc": 0.2680976430976431, "acc_ci": 0.0, "acc_runs": [0.2680976430976431], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009015838366608263, "centered_acc": 0.015151515151515138, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015151515151515138], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2525597269624573, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2525597269624573], "acc": 0.19795221843003413, "acc_ci": 0.0, "acc_runs": [0.19795221843003413], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012696728980207687, "centered_acc": 0.003412969283276427, "centered_acc_ci": 0.0, "centered_acc_runs": [0.003412969283276427], "ci": 0.0}, "openbookqa": {"acc_norm": 0.276, "acc_norm_ci": 0.0, "acc_norm_runs": [0.276], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02001121929807355, "centered_acc": 0.0346666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0346666666666667], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4846093133385951, "acc_ci": 0.0, "acc_runs": [0.4846093133385951], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014045826789783585, "centered_acc": -0.030781373322809835, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.030781373322809835], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 67012298.04645489, "perplexity_ci": 0.0, "perplexity_runs": [67012298.04645489], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.198, "acc_norm_ci": 0.0, "acc_norm_runs": [0.198], "acc": 0.057, "acc_ci": 0.0, "acc_runs": [0.057], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012607733934175283, "centered_acc": -0.06933333333333332, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.06933333333333332], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2977470787896576, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2937226876173648, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0024429779650229563, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2977470787896576], "mean_acc_runs": [0.2937226876173648], "mean_centered_acc_runs": [-0.0024429779650229563], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6336", "checkpoint": "checkpoint-6336", "global_step": 6336, "tokens_seen": 6643777536, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2635929097789285, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2635929097789285], "acc": 0.26020713005377416, "acc_ci": 0.0, "acc_runs": [0.26020713005377416], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004396806562351701, "centered_acc": 0.018123879705238027, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018123879705238027], "ci": 0.0}, "piqa": {"acc_norm": 0.5261153427638737, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5261153427638737], "acc": 0.5478781284004353, "acc_ci": 0.0, "acc_runs": [0.5478781284004353], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01164990085426341, "centered_acc": 0.052230685527747456, "centered_acc_ci": 0.0, "centered_acc_runs": [0.052230685527747456], "ci": 0.0}, "arc_easy": {"acc_norm": 0.265993265993266, "acc_norm_ci": 0.0, "acc_norm_runs": [0.265993265993266], "acc": 0.26851851851851855, "acc_ci": 0.0, "acc_runs": [0.26851851851851855], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009066789565615746, "centered_acc": 0.021324354657688005, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021324354657688005], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24829351535836178, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24829351535836178], "acc": 0.19539249146757678, "acc_ci": 0.0, "acc_runs": [0.19539249146757678], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012624912868089679, "centered_acc": -0.0022753128555176305, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0022753128555176305], "ci": 0.0}, "openbookqa": {"acc_norm": 0.266, "acc_norm_ci": 0.0, "acc_norm_runs": [0.266], "acc": 0.15, "acc_ci": 0.0, "acc_runs": [0.15], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019780559675655396, "centered_acc": 0.021333333333333353, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021333333333333353], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4925019731649566, "acc_ci": 0.0, "acc_runs": [0.4925019731649566], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050905521228646, "centered_acc": -0.014996053670086829, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.014996053670086829], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 84311625.70162986, "perplexity_ci": 0.0, "perplexity_runs": [84311625.70162986], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.194, "acc_norm_ci": 0.0, "acc_norm_runs": [0.194], "acc": 0.051, "acc_ci": 0.0, "acc_runs": [0.051], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012510816141264348, "centered_acc": -0.07466666666666666, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07466666666666666], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29399917231573836, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29337857033789616, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0047056507134293, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29399917231573836], "mean_acc_runs": [0.29337857033789616], "mean_centered_acc_runs": [-0.0047056507134293], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6624", "checkpoint": "checkpoint-6624", "global_step": 6624, "tokens_seen": 6945767424, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26399123680541725, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26399123680541725], "acc": 0.25960963951404104, "acc_ci": 0.0, "acc_runs": [0.25960963951404104], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004398937225038117, "centered_acc": 0.018654982407223002, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018654982407223002], "ci": 0.0}, "piqa": {"acc_norm": 0.5282916213275299, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5282916213275299], "acc": 0.5462459194776932, "acc_ci": 0.0, "acc_runs": [0.5462459194776932], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0116471341727495, "centered_acc": 0.05658324265505987, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05658324265505987], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26641414141414144, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26641414141414144], "acc": 0.2680976430976431, "acc_ci": 0.0, "acc_runs": [0.2680976430976431], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009071357971078843, "centered_acc": 0.021885521885521914, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021885521885521914], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25341296928327645, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25341296928327645], "acc": 0.19539249146757678, "acc_ci": 0.0, "acc_runs": [0.19539249146757678], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012710896778378677, "centered_acc": 0.004550625711035261, "centered_acc_ci": 0.0, "centered_acc_runs": [0.004550625711035261], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4980268350434096, "acc_ci": 0.0, "acc_runs": [0.4980268350434096], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052376259225643, "centered_acc": -0.003946329913180779, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.003946329913180779], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 91907972.31872264, "perplexity_ci": 0.0, "perplexity_runs": [91907972.31872264], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.201, "acc_norm_ci": 0.0, "acc_norm_runs": [0.201], "acc": 0.052, "acc_ci": 0.0, "acc_runs": [0.052], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012679107214617274, "centered_acc": -0.06533333333333331, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.06533333333333331], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29885166147172754, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29456599903740643, "mean_acc_ci": 0.0, "mean_centered_acc": 0.000996625884414063, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29885166147172754], "mean_acc_runs": [0.29456599903740643], "mean_centered_acc_runs": [0.000996625884414063], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-6912", "checkpoint": "checkpoint-6912", "global_step": 6912, "tokens_seen": 7247757312, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2634933280223063, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2634933280223063], "acc": 0.2581159131647082, "acc_ci": 0.0, "acc_runs": [0.2581159131647082], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004396273173717708, "centered_acc": 0.017991104029741727, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017991104029741727], "ci": 0.0}, "piqa": {"acc_norm": 0.5359085963003264, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5359085963003264], "acc": 0.5505984766050055, "acc_ci": 0.0, "acc_runs": [0.5505984766050055], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011635700809215522, "centered_acc": 0.07181719260065278, "centered_acc_ci": 0.0, "centered_acc_runs": [0.07181719260065278], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2676767676767677, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2676767676767677], "acc": 0.2697811447811448, "acc_ci": 0.0, "acc_runs": [0.2697811447811448], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009085000147099275, "centered_acc": 0.023569023569023573, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023569023569023573], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26109215017064846, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26109215017064846], "acc": 0.20051194539249148, "acc_ci": 0.0, "acc_runs": [0.20051194539249148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012835523909473708, "centered_acc": 0.014789533560864617, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014789533560864617], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.138, "acc_ci": 0.0, "acc_runs": [0.138], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4846093133385951, "acc_ci": 0.0, "acc_runs": [0.4846093133385951], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014045826789783585, "centered_acc": -0.030781373322809835, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.030781373322809835], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 132609337.44550243, "perplexity_ci": 0.0, "perplexity_runs": [132609337.44550243], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.2, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2], "acc": 0.055, "acc_ci": 0.0, "acc_runs": [0.055], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012655439943366655, "centered_acc": -0.06666666666666665, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.06666666666666665], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.3003618070283415, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2904904255055646, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0045229180757548025, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.3003618070283415], "mean_acc_runs": [0.2904904255055646], "mean_centered_acc_runs": [-0.0045229180757548025], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/i5-fulle-lm", "run": "i5-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/i5-fulle-lm/checkpoint-7200", "checkpoint": "checkpoint-7200", "global_step": 7200, "tokens_seen": 7549747200, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26329416450906196, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26329416450906196], "acc": 0.2592113124875523, "acc_ci": 0.0, "acc_runs": [0.2592113124875523], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004395205528158065, "centered_acc": 0.017725552678749274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017725552678749274], "ci": 0.0}, "piqa": {"acc_norm": 0.528835690968444, "acc_norm_ci": 0.0, "acc_norm_runs": [0.528835690968444], "acc": 0.544613710554951, "acc_ci": 0.0, "acc_runs": [0.544613710554951], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011646407809944788, "centered_acc": 0.05767138193688792, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05767138193688792], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2706228956228956, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2706228956228956], "acc": 0.26641414141414144, "acc_ci": 0.0, "acc_runs": [0.26641414141414144], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009116466166404035, "centered_acc": 0.0274971941638608, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0274971941638608], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.23890784982935154, "acc_norm_ci": 0.0, "acc_norm_runs": [0.23890784982935154], "acc": 0.19368600682593856, "acc_ci": 0.0, "acc_runs": [0.19368600682593856], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012461071376316715, "centered_acc": -0.014789533560864617, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.014789533560864617], "ci": 0.0}, "openbookqa": {"acc_norm": 0.276, "acc_norm_ci": 0.0, "acc_norm_runs": [0.276], "acc": 0.15, "acc_ci": 0.0, "acc_runs": [0.15], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02001121929807355, "centered_acc": 0.0346666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0346666666666667], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48697711128650356, "acc_ci": 0.0, "acc_runs": [0.48697711128650356], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014047718393997596, "centered_acc": -0.026045777426992878, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.026045777426992878], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.6, "acc_ci": 0.0, "acc_runs": [0.6], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0492365963917331, "centered_acc": 0.19999999999999996, "centered_acc_ci": 0.0, "centered_acc_runs": [0.19999999999999996], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 124825446.74493426, "perplexity_ci": 0.0, "perplexity_runs": [124825446.74493426], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.198, "acc_norm_ci": 0.0, "acc_norm_runs": [0.198], "acc": 0.053, "acc_ci": 0.0, "acc_runs": [0.053], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012607733934175283, "centered_acc": -0.06933333333333332, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.06933333333333332], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29594343348829216, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2932189744342787, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0017814361475139483, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29594343348829216], "mean_acc_runs": [0.2932189744342787], "mean_centered_acc_runs": [-0.0017814361475139483], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": true, "max_next_i_train": 5, "l2r_prefix_enabled": true} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-1152.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-1152.json new file mode 100644 index 0000000000000000000000000000000000000000..6fb9ae58e423a93593fc06431248a066ca7bb0f9 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-1152.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-1152", + "checkpoint": "checkpoint-1152", + "global_step": 1152, + "eval_loss": 4.016801834106445, + "metrics": { + "eval_loss": 4.016801834106445, + "eval_model_preparation_time": 0.0044, + "eval_runtime": 129.1729, + "eval_samples_per_second": 18.905, + "eval_steps_per_second": 2.369, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-1440.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-1440.json new file mode 100644 index 0000000000000000000000000000000000000000..ae29044cadbd6693e8252af425db38ae83b26e7c --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-1440.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-1440", + "checkpoint": "checkpoint-1440", + "global_step": 1440, + "eval_loss": 3.94815731048584, + "metrics": { + "eval_loss": 3.94815731048584, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.2196, + "eval_samples_per_second": 18.898, + "eval_steps_per_second": 2.368, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-1728.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-1728.json new file mode 100644 index 0000000000000000000000000000000000000000..a7da78a72de41c4c356627fdd7a3af1d353cc8a5 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-1728.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-1728", + "checkpoint": "checkpoint-1728", + "global_step": 1728, + "eval_loss": 3.91907000541687, + "metrics": { + "eval_loss": 3.91907000541687, + "eval_model_preparation_time": 0.004, + "eval_runtime": 129.1126, + "eval_samples_per_second": 18.914, + "eval_steps_per_second": 2.37, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-2016.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-2016.json new file mode 100644 index 0000000000000000000000000000000000000000..d64b3165346a0275967158f1dabd5db2d8a4850e --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-2016.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-2016", + "checkpoint": "checkpoint-2016", + "global_step": 2016, + "eval_loss": 3.910220146179199, + "metrics": { + "eval_loss": 3.910220146179199, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.109, + "eval_samples_per_second": 18.914, + "eval_steps_per_second": 2.37, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-2304.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-2304.json new file mode 100644 index 0000000000000000000000000000000000000000..70cb5697f17fcf6f0f1e11890704fea3a8e7481a --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-2304.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-2304", + "checkpoint": "checkpoint-2304", + "global_step": 2304, + "eval_loss": 3.909874200820923, + "metrics": { + "eval_loss": 3.909874200820923, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.128, + "eval_samples_per_second": 18.911, + "eval_steps_per_second": 2.37, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-2592.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-2592.json new file mode 100644 index 0000000000000000000000000000000000000000..e4d0ba7b58b3f9736690a0bedf2d03a22c579824 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-2592.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-2592", + "checkpoint": "checkpoint-2592", + "global_step": 2592, + "eval_loss": 3.9253413677215576, + "metrics": { + "eval_loss": 3.9253413677215576, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 129.0787, + "eval_samples_per_second": 18.919, + "eval_steps_per_second": 2.371, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-288.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-288.json new file mode 100644 index 0000000000000000000000000000000000000000..0fe2c5519a639c93e6b17c1d8f966c0e606a8675 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-288.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-288", + "checkpoint": "checkpoint-288", + "global_step": 288, + "eval_loss": 5.713050365447998, + "metrics": { + "eval_loss": 5.713050365447998, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 128.5467, + "eval_samples_per_second": 18.997, + "eval_steps_per_second": 2.38, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-2880.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-2880.json new file mode 100644 index 0000000000000000000000000000000000000000..a3a972e203c84bfd9753cc81ca6642004ecd9882 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-2880.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-2880", + "checkpoint": "checkpoint-2880", + "global_step": 2880, + "eval_loss": 3.939215660095215, + "metrics": { + "eval_loss": 3.939215660095215, + "eval_model_preparation_time": 0.0055, + "eval_runtime": 129.0363, + "eval_samples_per_second": 18.925, + "eval_steps_per_second": 2.371, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-3168.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-3168.json new file mode 100644 index 0000000000000000000000000000000000000000..856b90b3367bb6db1484d6f94062824fdefe1507 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-3168.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-3168", + "checkpoint": "checkpoint-3168", + "global_step": 3168, + "eval_loss": 3.9709622859954834, + "metrics": { + "eval_loss": 3.9709622859954834, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 127.5234, + "eval_samples_per_second": 19.149, + "eval_steps_per_second": 2.4, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-3456.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-3456.json new file mode 100644 index 0000000000000000000000000000000000000000..45ecca2d522132225c225becfe3b9cdb11843058 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-3456.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-3456", + "checkpoint": "checkpoint-3456", + "global_step": 3456, + "eval_loss": 3.997180700302124, + "metrics": { + "eval_loss": 3.997180700302124, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 128.714, + "eval_samples_per_second": 18.972, + "eval_steps_per_second": 2.377, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-3744.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-3744.json new file mode 100644 index 0000000000000000000000000000000000000000..c2f0ddb50817e254e32bf601ad32a18cd2af0cb9 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-3744.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-3744", + "checkpoint": "checkpoint-3744", + "global_step": 3744, + "eval_loss": 4.008206367492676, + "metrics": { + "eval_loss": 4.008206367492676, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.174, + "eval_samples_per_second": 18.905, + "eval_steps_per_second": 2.369, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-4032.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-4032.json new file mode 100644 index 0000000000000000000000000000000000000000..8e34ed80fb0689119f69b764579e98b7737ef393 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-4032.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-4032", + "checkpoint": "checkpoint-4032", + "global_step": 4032, + "eval_loss": 4.029216289520264, + "metrics": { + "eval_loss": 4.029216289520264, + "eval_model_preparation_time": 0.0052, + "eval_runtime": 127.6329, + "eval_samples_per_second": 19.133, + "eval_steps_per_second": 2.398, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-4320.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-4320.json new file mode 100644 index 0000000000000000000000000000000000000000..d6b147d4c6bb908b9b03050d77c3eb454cd3bc5c --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-4320.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-4320", + "checkpoint": "checkpoint-4320", + "global_step": 4320, + "eval_loss": 4.056842803955078, + "metrics": { + "eval_loss": 4.056842803955078, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 128.6999, + "eval_samples_per_second": 18.974, + "eval_steps_per_second": 2.378, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-4608.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-4608.json new file mode 100644 index 0000000000000000000000000000000000000000..7cfaf50f06e1208972180102ed135e8d454d263b --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-4608.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-4608", + "checkpoint": "checkpoint-4608", + "global_step": 4608, + "eval_loss": 4.077906131744385, + "metrics": { + "eval_loss": 4.077906131744385, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 128.9672, + "eval_samples_per_second": 18.935, + "eval_steps_per_second": 2.373, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-4896.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-4896.json new file mode 100644 index 0000000000000000000000000000000000000000..0b6c605180a6b064a5a13f5d4b6f80cca81afc01 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-4896.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-4896", + "checkpoint": "checkpoint-4896", + "global_step": 4896, + "eval_loss": 4.094040393829346, + "metrics": { + "eval_loss": 4.094040393829346, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 128.9995, + "eval_samples_per_second": 18.93, + "eval_steps_per_second": 2.372, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-5184.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-5184.json new file mode 100644 index 0000000000000000000000000000000000000000..b0a7d318d6ae687a312ca702ee65f334bb24e91d --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-5184.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-5184", + "checkpoint": "checkpoint-5184", + "global_step": 5184, + "eval_loss": 4.118155479431152, + "metrics": { + "eval_loss": 4.118155479431152, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 128.9892, + "eval_samples_per_second": 18.932, + "eval_steps_per_second": 2.372, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-5472.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-5472.json new file mode 100644 index 0000000000000000000000000000000000000000..82a370befa04a26e5a2bfb92e214e2784c45a670 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-5472.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-5472", + "checkpoint": "checkpoint-5472", + "global_step": 5472, + "eval_loss": 4.146016597747803, + "metrics": { + "eval_loss": 4.146016597747803, + "eval_model_preparation_time": 0.0047, + "eval_runtime": 128.9989, + "eval_samples_per_second": 18.93, + "eval_steps_per_second": 2.372, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-576.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-576.json new file mode 100644 index 0000000000000000000000000000000000000000..f1999a648258929db9682484b025a13241129bdf --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-576.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-576", + "checkpoint": "checkpoint-576", + "global_step": 576, + "eval_loss": 4.579164028167725, + "metrics": { + "eval_loss": 4.579164028167725, + "eval_model_preparation_time": 0.0038, + "eval_runtime": 128.9371, + "eval_samples_per_second": 18.939, + "eval_steps_per_second": 2.373, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-5760.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-5760.json new file mode 100644 index 0000000000000000000000000000000000000000..c19f627d016d71f14a8bf590f0251bc0d19c423e --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-5760.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-5760", + "checkpoint": "checkpoint-5760", + "global_step": 5760, + "eval_loss": 4.170133590698242, + "metrics": { + "eval_loss": 4.170133590698242, + "eval_model_preparation_time": 0.004, + "eval_runtime": 129.0287, + "eval_samples_per_second": 18.926, + "eval_steps_per_second": 2.372, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-6048.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-6048.json new file mode 100644 index 0000000000000000000000000000000000000000..fdfecccbdf37bf0af925204199bf5f2ff449f42e --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-6048.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-6048", + "checkpoint": "checkpoint-6048", + "global_step": 6048, + "eval_loss": 4.187249660491943, + "metrics": { + "eval_loss": 4.187249660491943, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 127.6885, + "eval_samples_per_second": 19.125, + "eval_steps_per_second": 2.396, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-6336.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-6336.json new file mode 100644 index 0000000000000000000000000000000000000000..fdafe9eaba8e8c71c281444e4c409ff1b4ce5af3 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-6336.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-6336", + "checkpoint": "checkpoint-6336", + "global_step": 6336, + "eval_loss": 4.210964679718018, + "metrics": { + "eval_loss": 4.210964679718018, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 128.6719, + "eval_samples_per_second": 18.978, + "eval_steps_per_second": 2.378, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-6624.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-6624.json new file mode 100644 index 0000000000000000000000000000000000000000..8ca5bcb4352d486e562260568953244d310e4802 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-6624.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-6624", + "checkpoint": "checkpoint-6624", + "global_step": 6624, + "eval_loss": 4.206256866455078, + "metrics": { + "eval_loss": 4.206256866455078, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 128.8726, + "eval_samples_per_second": 18.949, + "eval_steps_per_second": 2.374, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-6912.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-6912.json new file mode 100644 index 0000000000000000000000000000000000000000..6217633a5b9efb7585e3dbb3aa852a3bc2e91e27 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-6912.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-6912", + "checkpoint": "checkpoint-6912", + "global_step": 6912, + "eval_loss": 4.243518829345703, + "metrics": { + "eval_loss": 4.243518829345703, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 128.9396, + "eval_samples_per_second": 18.939, + "eval_steps_per_second": 2.373, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-72.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-72.json new file mode 100644 index 0000000000000000000000000000000000000000..3f90fea2c807630a58f80164bc70cdd33d549dcd --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-72.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-72", + "checkpoint": "checkpoint-72", + "global_step": 72, + "eval_loss": 7.504610061645508, + "metrics": { + "eval_loss": 7.504610061645508, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 126.677, + "eval_samples_per_second": 19.277, + "eval_steps_per_second": 2.416, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-7200.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-7200.json new file mode 100644 index 0000000000000000000000000000000000000000..508843b1a06a2fbd5620297a2a44686a756818eb --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-7200.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-7200", + "checkpoint": "checkpoint-7200", + "global_step": 7200, + "eval_loss": 4.263388156890869, + "metrics": { + "eval_loss": 4.263388156890869, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 129.0214, + "eval_samples_per_second": 18.927, + "eval_steps_per_second": 2.372, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-7488.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-7488.json new file mode 100644 index 0000000000000000000000000000000000000000..4fe12a5a61dcd80f699778bdaca029d156f0e996 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-7488.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-7488", + "checkpoint": "checkpoint-7488", + "global_step": 7488, + "eval_loss": 4.273472785949707, + "metrics": { + "eval_loss": 4.273472785949707, + "eval_model_preparation_time": 0.0039, + "eval_runtime": 129.0449, + "eval_samples_per_second": 18.924, + "eval_steps_per_second": 2.371, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-7776.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-7776.json new file mode 100644 index 0000000000000000000000000000000000000000..5f6121672752fcbf7125cc7bf00fe3a7cf1117f6 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-7776.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-7776", + "checkpoint": "checkpoint-7776", + "global_step": 7776, + "eval_loss": 4.297510623931885, + "metrics": { + "eval_loss": 4.297510623931885, + "eval_model_preparation_time": 0.004, + "eval_runtime": 129.0415, + "eval_samples_per_second": 18.924, + "eval_steps_per_second": 2.371, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-8064.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-8064.json new file mode 100644 index 0000000000000000000000000000000000000000..2596b5a6c0c4e589e07ff246be5a3eef2fd2bb2b --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-8064.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-8064", + "checkpoint": "checkpoint-8064", + "global_step": 8064, + "eval_loss": 4.313085556030273, + "metrics": { + "eval_loss": 4.313085556030273, + "eval_model_preparation_time": 0.0038, + "eval_runtime": 131.5187, + "eval_samples_per_second": 18.568, + "eval_steps_per_second": 4.646, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-8352.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-8352.json new file mode 100644 index 0000000000000000000000000000000000000000..8d84cd9d804ba917b0ffdbb41caf2df781c037d8 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-8352.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-8352", + "checkpoint": "checkpoint-8352", + "global_step": 8352, + "eval_loss": 4.332733631134033, + "metrics": { + "eval_loss": 4.332733631134033, + "eval_model_preparation_time": 0.0038, + "eval_runtime": 132.2285, + "eval_samples_per_second": 18.468, + "eval_steps_per_second": 4.621, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-864.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-864.json new file mode 100644 index 0000000000000000000000000000000000000000..03a6ff010efafcfd191117f569e4c5036d2257c1 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-864.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-864", + "checkpoint": "checkpoint-864", + "global_step": 864, + "eval_loss": 4.1703386306762695, + "metrics": { + "eval_loss": 4.1703386306762695, + "eval_model_preparation_time": 0.0055, + "eval_runtime": 129.0517, + "eval_samples_per_second": 18.923, + "eval_steps_per_second": 2.371, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-8640.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-8640.json new file mode 100644 index 0000000000000000000000000000000000000000..59b0a806e499dc4fde1172e6c297a0cd4921611b --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-8640.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-8640", + "checkpoint": "checkpoint-8640", + "global_step": 8640, + "eval_loss": 4.349362850189209, + "metrics": { + "eval_loss": 4.349362850189209, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 132.6311, + "eval_samples_per_second": 18.412, + "eval_steps_per_second": 4.607, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-8928.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-8928.json new file mode 100644 index 0000000000000000000000000000000000000000..13c51dd04fefa534cd957dffa6b28f674675b210 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-8928.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-8928", + "checkpoint": "checkpoint-8928", + "global_step": 8928, + "eval_loss": 4.363689422607422, + "metrics": { + "eval_loss": 4.363689422607422, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 132.7793, + "eval_samples_per_second": 18.391, + "eval_steps_per_second": 4.602, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-9216.json b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-9216.json new file mode 100644 index 0000000000000000000000000000000000000000..1117176dafe5dfab7d9d6a205ffeea197f8beffa --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/checkpoint-9216.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-9216", + "checkpoint": "checkpoint-9216", + "global_step": 9216, + "eval_loss": 4.39207649230957, + "metrics": { + "eval_loss": 4.39207649230957, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 132.8188, + "eval_samples_per_second": 18.386, + "eval_steps_per_second": 4.6, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": true, + "l2r_percent": 100.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 4, + "eval_gradient_accumulation_steps": 128, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/l2r50-fulle-lm/l2r_n1_eval/summary.jsonl b/results/l2r50-fulle-lm/l2r_n1_eval/summary.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..b1d7af9f87aadbea66bf70da2e8b173a65130121 --- /dev/null +++ b/results/l2r50-fulle-lm/l2r_n1_eval/summary.jsonl @@ -0,0 +1,33 @@ +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-72", "checkpoint": "checkpoint-72", "global_step": 72, "eval_loss": 7.504610061645508, "metrics": {"eval_loss": 7.504610061645508, "eval_model_preparation_time": 0.0035, "eval_runtime": 126.677, "eval_samples_per_second": 19.277, "eval_steps_per_second": 2.416, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-288", "checkpoint": "checkpoint-288", "global_step": 288, "eval_loss": 5.713050365447998, "metrics": {"eval_loss": 5.713050365447998, "eval_model_preparation_time": 0.0035, "eval_runtime": 128.5467, "eval_samples_per_second": 18.997, "eval_steps_per_second": 2.38, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-576", "checkpoint": "checkpoint-576", "global_step": 576, "eval_loss": 4.579164028167725, "metrics": {"eval_loss": 4.579164028167725, "eval_model_preparation_time": 0.0038, "eval_runtime": 128.9371, "eval_samples_per_second": 18.939, "eval_steps_per_second": 2.373, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-864", "checkpoint": "checkpoint-864", "global_step": 864, "eval_loss": 4.1703386306762695, "metrics": {"eval_loss": 4.1703386306762695, "eval_model_preparation_time": 0.0055, "eval_runtime": 129.0517, "eval_samples_per_second": 18.923, "eval_steps_per_second": 2.371, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-1152", "checkpoint": "checkpoint-1152", "global_step": 1152, "eval_loss": 4.016801834106445, "metrics": {"eval_loss": 4.016801834106445, "eval_model_preparation_time": 0.0044, "eval_runtime": 129.1729, "eval_samples_per_second": 18.905, "eval_steps_per_second": 2.369, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-1440", "checkpoint": "checkpoint-1440", "global_step": 1440, "eval_loss": 3.94815731048584, "metrics": {"eval_loss": 3.94815731048584, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.2196, "eval_samples_per_second": 18.898, "eval_steps_per_second": 2.368, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-1728", "checkpoint": "checkpoint-1728", "global_step": 1728, "eval_loss": 3.91907000541687, "metrics": {"eval_loss": 3.91907000541687, "eval_model_preparation_time": 0.004, "eval_runtime": 129.1126, "eval_samples_per_second": 18.914, "eval_steps_per_second": 2.37, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-2016", "checkpoint": "checkpoint-2016", "global_step": 2016, "eval_loss": 3.910220146179199, "metrics": {"eval_loss": 3.910220146179199, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.109, "eval_samples_per_second": 18.914, "eval_steps_per_second": 2.37, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-2304", "checkpoint": "checkpoint-2304", "global_step": 2304, "eval_loss": 3.909874200820923, "metrics": {"eval_loss": 3.909874200820923, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.128, "eval_samples_per_second": 18.911, "eval_steps_per_second": 2.37, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-2592", "checkpoint": "checkpoint-2592", "global_step": 2592, "eval_loss": 3.9253413677215576, "metrics": {"eval_loss": 3.9253413677215576, "eval_model_preparation_time": 0.0033, "eval_runtime": 129.0787, "eval_samples_per_second": 18.919, "eval_steps_per_second": 2.371, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-2880", "checkpoint": "checkpoint-2880", "global_step": 2880, "eval_loss": 3.939215660095215, "metrics": {"eval_loss": 3.939215660095215, "eval_model_preparation_time": 0.0055, "eval_runtime": 129.0363, "eval_samples_per_second": 18.925, "eval_steps_per_second": 2.371, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-3168", "checkpoint": "checkpoint-3168", "global_step": 3168, "eval_loss": 3.9709622859954834, "metrics": {"eval_loss": 3.9709622859954834, "eval_model_preparation_time": 0.0035, "eval_runtime": 127.5234, "eval_samples_per_second": 19.149, "eval_steps_per_second": 2.4, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-3456", "checkpoint": "checkpoint-3456", "global_step": 3456, "eval_loss": 3.997180700302124, "metrics": {"eval_loss": 3.997180700302124, "eval_model_preparation_time": 0.0037, "eval_runtime": 128.714, "eval_samples_per_second": 18.972, "eval_steps_per_second": 2.377, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-3744", "checkpoint": "checkpoint-3744", "global_step": 3744, "eval_loss": 4.008206367492676, "metrics": {"eval_loss": 4.008206367492676, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.174, "eval_samples_per_second": 18.905, "eval_steps_per_second": 2.369, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-4032", "checkpoint": "checkpoint-4032", "global_step": 4032, "eval_loss": 4.029216289520264, "metrics": {"eval_loss": 4.029216289520264, "eval_model_preparation_time": 0.0052, "eval_runtime": 127.6329, "eval_samples_per_second": 19.133, "eval_steps_per_second": 2.398, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-4320", "checkpoint": "checkpoint-4320", "global_step": 4320, "eval_loss": 4.056842803955078, "metrics": {"eval_loss": 4.056842803955078, "eval_model_preparation_time": 0.0033, "eval_runtime": 128.6999, "eval_samples_per_second": 18.974, "eval_steps_per_second": 2.378, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-4608", "checkpoint": "checkpoint-4608", "global_step": 4608, "eval_loss": 4.077906131744385, "metrics": {"eval_loss": 4.077906131744385, "eval_model_preparation_time": 0.0033, "eval_runtime": 128.9672, "eval_samples_per_second": 18.935, "eval_steps_per_second": 2.373, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-4896", "checkpoint": "checkpoint-4896", "global_step": 4896, "eval_loss": 4.094040393829346, "metrics": {"eval_loss": 4.094040393829346, "eval_model_preparation_time": 0.0033, "eval_runtime": 128.9995, "eval_samples_per_second": 18.93, "eval_steps_per_second": 2.372, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-5184", "checkpoint": "checkpoint-5184", "global_step": 5184, "eval_loss": 4.118155479431152, "metrics": {"eval_loss": 4.118155479431152, "eval_model_preparation_time": 0.0036, "eval_runtime": 128.9892, "eval_samples_per_second": 18.932, "eval_steps_per_second": 2.372, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-5472", "checkpoint": "checkpoint-5472", "global_step": 5472, "eval_loss": 4.146016597747803, "metrics": {"eval_loss": 4.146016597747803, "eval_model_preparation_time": 0.0047, "eval_runtime": 128.9989, "eval_samples_per_second": 18.93, "eval_steps_per_second": 2.372, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-5760", "checkpoint": "checkpoint-5760", "global_step": 5760, "eval_loss": 4.170133590698242, "metrics": {"eval_loss": 4.170133590698242, "eval_model_preparation_time": 0.004, "eval_runtime": 129.0287, "eval_samples_per_second": 18.926, "eval_steps_per_second": 2.372, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-6048", "checkpoint": "checkpoint-6048", "global_step": 6048, "eval_loss": 4.187249660491943, "metrics": {"eval_loss": 4.187249660491943, "eval_model_preparation_time": 0.0036, "eval_runtime": 127.6885, "eval_samples_per_second": 19.125, "eval_steps_per_second": 2.396, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-6336", "checkpoint": "checkpoint-6336", "global_step": 6336, "eval_loss": 4.210964679718018, "metrics": {"eval_loss": 4.210964679718018, "eval_model_preparation_time": 0.0033, "eval_runtime": 128.6719, "eval_samples_per_second": 18.978, "eval_steps_per_second": 2.378, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-6624", "checkpoint": "checkpoint-6624", "global_step": 6624, "eval_loss": 4.206256866455078, "metrics": {"eval_loss": 4.206256866455078, "eval_model_preparation_time": 0.0033, "eval_runtime": 128.8726, "eval_samples_per_second": 18.949, "eval_steps_per_second": 2.374, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-6912", "checkpoint": "checkpoint-6912", "global_step": 6912, "eval_loss": 4.243518829345703, "metrics": {"eval_loss": 4.243518829345703, "eval_model_preparation_time": 0.0033, "eval_runtime": 128.9396, "eval_samples_per_second": 18.939, "eval_steps_per_second": 2.373, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-7200", "checkpoint": "checkpoint-7200", "global_step": 7200, "eval_loss": 4.263388156890869, "metrics": {"eval_loss": 4.263388156890869, "eval_model_preparation_time": 0.0033, "eval_runtime": 129.0214, "eval_samples_per_second": 18.927, "eval_steps_per_second": 2.372, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-7488", "checkpoint": "checkpoint-7488", "global_step": 7488, "eval_loss": 4.273472785949707, "metrics": {"eval_loss": 4.273472785949707, "eval_model_preparation_time": 0.0039, "eval_runtime": 129.0449, "eval_samples_per_second": 18.924, "eval_steps_per_second": 2.371, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-7776", "checkpoint": "checkpoint-7776", "global_step": 7776, "eval_loss": 4.297510623931885, "metrics": {"eval_loss": 4.297510623931885, "eval_model_preparation_time": 0.004, "eval_runtime": 129.0415, "eval_samples_per_second": 18.924, "eval_steps_per_second": 2.371, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-8064", "checkpoint": "checkpoint-8064", "global_step": 8064, "eval_loss": 4.313085556030273, "metrics": {"eval_loss": 4.313085556030273, "eval_model_preparation_time": 0.0038, "eval_runtime": 131.5187, "eval_samples_per_second": 18.568, "eval_steps_per_second": 4.646, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-8352", "checkpoint": "checkpoint-8352", "global_step": 8352, "eval_loss": 4.332733631134033, "metrics": {"eval_loss": 4.332733631134033, "eval_model_preparation_time": 0.0038, "eval_runtime": 132.2285, "eval_samples_per_second": 18.468, "eval_steps_per_second": 4.621, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-8640", "checkpoint": "checkpoint-8640", "global_step": 8640, "eval_loss": 4.349362850189209, "metrics": {"eval_loss": 4.349362850189209, "eval_model_preparation_time": 0.0035, "eval_runtime": 132.6311, "eval_samples_per_second": 18.412, "eval_steps_per_second": 4.607, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-8928", "checkpoint": "checkpoint-8928", "global_step": 8928, "eval_loss": 4.363689422607422, "metrics": {"eval_loss": 4.363689422607422, "eval_model_preparation_time": 0.0035, "eval_runtime": 132.7793, "eval_samples_per_second": 18.391, "eval_steps_per_second": 4.602, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/l2r50-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/l2r50-fulle-lm/checkpoint-9216", "checkpoint": "checkpoint-9216", "global_step": 9216, "eval_loss": 4.39207649230957, "metrics": {"eval_loss": 4.39207649230957, "eval_model_preparation_time": 0.0036, "eval_runtime": 132.8188, "eval_samples_per_second": 18.386, "eval_steps_per_second": 4.6, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": true, "l2r_percent": 100.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 4, "eval_gradient_accumulation_steps": 128, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-10080.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-10080.json new file mode 100644 index 0000000000000000000000000000000000000000..fd628d6d811915fe11cd4303e9032c0eed4d504a --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-10080.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-10080", + "checkpoint": "checkpoint-10080", + "global_step": 10080, + "eval_loss": 4.687160015106201, + "metrics": { + "eval_loss": 4.687160015106201, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.0142, + "eval_samples_per_second": 18.928, + "eval_steps_per_second": 2.372, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-10368.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-10368.json new file mode 100644 index 0000000000000000000000000000000000000000..3acd3e763076dc6c20dfa9a1f8b689ac9528545f --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-10368.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-10368", + "checkpoint": "checkpoint-10368", + "global_step": 10368, + "eval_loss": 4.723207950592041, + "metrics": { + "eval_loss": 4.723207950592041, + "eval_model_preparation_time": 0.0043, + "eval_runtime": 129.1996, + "eval_samples_per_second": 18.901, + "eval_steps_per_second": 2.368, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-1152.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-1152.json new file mode 100644 index 0000000000000000000000000000000000000000..65584ec94274d504bb942fccf88426cb29fd7682 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-1152.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-1152", + "checkpoint": "checkpoint-1152", + "global_step": 1152, + "eval_loss": 4.138202667236328, + "metrics": { + "eval_loss": 4.138202667236328, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.2493, + "eval_samples_per_second": 18.894, + "eval_steps_per_second": 2.368, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-1440.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-1440.json new file mode 100644 index 0000000000000000000000000000000000000000..ad1f1c5b90a911c5bd0387e17deb97785a56e0d8 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-1440.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-1440", + "checkpoint": "checkpoint-1440", + "global_step": 1440, + "eval_loss": 3.9913787841796875, + "metrics": { + "eval_loss": 3.9913787841796875, + "eval_model_preparation_time": 0.0059, + "eval_runtime": 129.2595, + "eval_samples_per_second": 18.892, + "eval_steps_per_second": 2.367, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-1728.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-1728.json new file mode 100644 index 0000000000000000000000000000000000000000..2610317b2249792428fe3fecfe184772fa6f913d --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-1728.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-1728", + "checkpoint": "checkpoint-1728", + "global_step": 1728, + "eval_loss": 3.928884506225586, + "metrics": { + "eval_loss": 3.928884506225586, + "eval_model_preparation_time": 0.0057, + "eval_runtime": 129.08, + "eval_samples_per_second": 18.919, + "eval_steps_per_second": 2.371, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-2016.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-2016.json new file mode 100644 index 0000000000000000000000000000000000000000..e18a4d8e9bf056fd8aaff070635c264402f4faa7 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-2016.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2016", + "checkpoint": "checkpoint-2016", + "global_step": 2016, + "eval_loss": 3.9102258682250977, + "metrics": { + "eval_loss": 3.9102258682250977, + "eval_model_preparation_time": 0.0044, + "eval_runtime": 128.9893, + "eval_samples_per_second": 18.932, + "eval_steps_per_second": 2.372, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-2304.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-2304.json new file mode 100644 index 0000000000000000000000000000000000000000..be77f545622a57d451e61055f51f9bb23a9070dc --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-2304.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2304", + "checkpoint": "checkpoint-2304", + "global_step": 2304, + "eval_loss": 3.911813497543335, + "metrics": { + "eval_loss": 3.911813497543335, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 128.8471, + "eval_samples_per_second": 18.953, + "eval_steps_per_second": 2.375, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-2592.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-2592.json new file mode 100644 index 0000000000000000000000000000000000000000..a6ccecc2ae7f47ff461995a43b7b28a42d3c4086 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-2592.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2592", + "checkpoint": "checkpoint-2592", + "global_step": 2592, + "eval_loss": 3.9276037216186523, + "metrics": { + "eval_loss": 3.9276037216186523, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.0827, + "eval_samples_per_second": 18.918, + "eval_steps_per_second": 2.371, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-288.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-288.json new file mode 100644 index 0000000000000000000000000000000000000000..6a81efbe493ea82b981097d2b438e3fc848a9839 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-288.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-288", + "checkpoint": "checkpoint-288", + "global_step": 288, + "eval_loss": 6.486622333526611, + "metrics": { + "eval_loss": 6.486622333526611, + "eval_model_preparation_time": 0.0057, + "eval_runtime": 128.6597, + "eval_samples_per_second": 18.98, + "eval_steps_per_second": 2.378, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-2880.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-2880.json new file mode 100644 index 0000000000000000000000000000000000000000..cd861e9c1115683f52305641419999868986dcca --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-2880.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2880", + "checkpoint": "checkpoint-2880", + "global_step": 2880, + "eval_loss": 3.9492850303649902, + "metrics": { + "eval_loss": 3.9492850303649902, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.0449, + "eval_samples_per_second": 18.924, + "eval_steps_per_second": 2.371, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-3168.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-3168.json new file mode 100644 index 0000000000000000000000000000000000000000..a08b8867923d4244d76ef5283154fed6d7638ed8 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-3168.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-3168", + "checkpoint": "checkpoint-3168", + "global_step": 3168, + "eval_loss": 3.9791922569274902, + "metrics": { + "eval_loss": 3.9791922569274902, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 128.9926, + "eval_samples_per_second": 18.931, + "eval_steps_per_second": 2.372, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-3456.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-3456.json new file mode 100644 index 0000000000000000000000000000000000000000..71ddef203751049c27069af686cd05390b639dce --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-3456.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-3456", + "checkpoint": "checkpoint-3456", + "global_step": 3456, + "eval_loss": 4.006196022033691, + "metrics": { + "eval_loss": 4.006196022033691, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.007, + "eval_samples_per_second": 18.929, + "eval_steps_per_second": 2.372, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-3744.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-3744.json new file mode 100644 index 0000000000000000000000000000000000000000..f1dad7ac35406f38babb705ad3b2dfac884b154a --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-3744.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-3744", + "checkpoint": "checkpoint-3744", + "global_step": 3744, + "eval_loss": 4.035167217254639, + "metrics": { + "eval_loss": 4.035167217254639, + "eval_model_preparation_time": 0.0052, + "eval_runtime": 128.9832, + "eval_samples_per_second": 18.933, + "eval_steps_per_second": 2.372, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-4032.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-4032.json new file mode 100644 index 0000000000000000000000000000000000000000..74bcfa8ee44baca8a0e43e2917889e63b56055a0 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-4032.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4032", + "checkpoint": "checkpoint-4032", + "global_step": 4032, + "eval_loss": 4.076192378997803, + "metrics": { + "eval_loss": 4.076192378997803, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.1461, + "eval_samples_per_second": 18.909, + "eval_steps_per_second": 2.369, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-4320.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-4320.json new file mode 100644 index 0000000000000000000000000000000000000000..9332ff1fe3c6f3ee5d805074108000abd213ea68 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-4320.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4320", + "checkpoint": "checkpoint-4320", + "global_step": 4320, + "eval_loss": 4.109894752502441, + "metrics": { + "eval_loss": 4.109894752502441, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 128.919, + "eval_samples_per_second": 18.942, + "eval_steps_per_second": 2.374, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-4608.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-4608.json new file mode 100644 index 0000000000000000000000000000000000000000..8746c94807f67d2ce84b1fe5d52ea7a11b85c9f0 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-4608.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4608", + "checkpoint": "checkpoint-4608", + "global_step": 4608, + "eval_loss": 4.143605709075928, + "metrics": { + "eval_loss": 4.143605709075928, + "eval_model_preparation_time": 0.0041, + "eval_runtime": 129.1433, + "eval_samples_per_second": 18.909, + "eval_steps_per_second": 2.369, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-4896.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-4896.json new file mode 100644 index 0000000000000000000000000000000000000000..206a80a73228e8fae59a1d2e726f63de02a5fede --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-4896.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4896", + "checkpoint": "checkpoint-4896", + "global_step": 4896, + "eval_loss": 4.178235054016113, + "metrics": { + "eval_loss": 4.178235054016113, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.2611, + "eval_samples_per_second": 18.892, + "eval_steps_per_second": 2.367, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-5184.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-5184.json new file mode 100644 index 0000000000000000000000000000000000000000..31e366470c54cd22884d994df5339332a5469ee5 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-5184.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-5184", + "checkpoint": "checkpoint-5184", + "global_step": 5184, + "eval_loss": 4.207855224609375, + "metrics": { + "eval_loss": 4.207855224609375, + "eval_model_preparation_time": 0.0038, + "eval_runtime": 129.1778, + "eval_samples_per_second": 18.904, + "eval_steps_per_second": 2.369, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-5472.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-5472.json new file mode 100644 index 0000000000000000000000000000000000000000..0755b93b11a8a3e2ba1862494af334de20edd3e9 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-5472.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-5472", + "checkpoint": "checkpoint-5472", + "global_step": 5472, + "eval_loss": 4.2506256103515625, + "metrics": { + "eval_loss": 4.2506256103515625, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.113, + "eval_samples_per_second": 18.914, + "eval_steps_per_second": 2.37, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-576.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-576.json new file mode 100644 index 0000000000000000000000000000000000000000..e6a505e14105b0f4322c8d4f388acba9470a36f1 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-576.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-576", + "checkpoint": "checkpoint-576", + "global_step": 576, + "eval_loss": 5.569174766540527, + "metrics": { + "eval_loss": 5.569174766540527, + "eval_model_preparation_time": 0.0052, + "eval_runtime": 128.9688, + "eval_samples_per_second": 18.935, + "eval_steps_per_second": 2.373, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-5760.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-5760.json new file mode 100644 index 0000000000000000000000000000000000000000..87d12712ca2f2fcc112f30afc4a91d94b54155cd --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-5760.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-5760", + "checkpoint": "checkpoint-5760", + "global_step": 5760, + "eval_loss": 4.276021480560303, + "metrics": { + "eval_loss": 4.276021480560303, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.276, + "eval_samples_per_second": 18.89, + "eval_steps_per_second": 2.367, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-6048.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-6048.json new file mode 100644 index 0000000000000000000000000000000000000000..78fadeedc0e9d797225dcc97d6738be3b5a381fc --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-6048.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6048", + "checkpoint": "checkpoint-6048", + "global_step": 6048, + "eval_loss": 4.306779861450195, + "metrics": { + "eval_loss": 4.306779861450195, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.061, + "eval_samples_per_second": 18.921, + "eval_steps_per_second": 2.371, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-6336.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-6336.json new file mode 100644 index 0000000000000000000000000000000000000000..a02fcdda453ce63157592ba69aed5c39fee278af --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-6336.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6336", + "checkpoint": "checkpoint-6336", + "global_step": 6336, + "eval_loss": 4.3341827392578125, + "metrics": { + "eval_loss": 4.3341827392578125, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.0846, + "eval_samples_per_second": 18.918, + "eval_steps_per_second": 2.371, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-6624.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-6624.json new file mode 100644 index 0000000000000000000000000000000000000000..0e5c85a9b28c221f6c0dd4a9e19d77779b841f21 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-6624.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6624", + "checkpoint": "checkpoint-6624", + "global_step": 6624, + "eval_loss": 4.370403289794922, + "metrics": { + "eval_loss": 4.370403289794922, + "eval_model_preparation_time": 0.0056, + "eval_runtime": 129.038, + "eval_samples_per_second": 18.925, + "eval_steps_per_second": 2.371, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-6912.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-6912.json new file mode 100644 index 0000000000000000000000000000000000000000..6223e0c79b7cab65dddcd13baf4f053c30004d6d --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-6912.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6912", + "checkpoint": "checkpoint-6912", + "global_step": 6912, + "eval_loss": 4.416069507598877, + "metrics": { + "eval_loss": 4.416069507598877, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 128.9833, + "eval_samples_per_second": 18.933, + "eval_steps_per_second": 2.372, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-72.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-72.json new file mode 100644 index 0000000000000000000000000000000000000000..44cc4281cd911b0fd509c00822a7c350940874b2 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-72.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-72", + "checkpoint": "checkpoint-72", + "global_step": 72, + "eval_loss": 7.526080131530762, + "metrics": { + "eval_loss": 7.526080131530762, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 126.4927, + "eval_samples_per_second": 19.305, + "eval_steps_per_second": 2.419, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-7200.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-7200.json new file mode 100644 index 0000000000000000000000000000000000000000..9cbc0c7e59401c6e7caf2d34b3ccd17b6b000e32 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-7200.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-7200", + "checkpoint": "checkpoint-7200", + "global_step": 7200, + "eval_loss": 4.452248573303223, + "metrics": { + "eval_loss": 4.452248573303223, + "eval_model_preparation_time": 0.0054, + "eval_runtime": 129.2346, + "eval_samples_per_second": 18.896, + "eval_steps_per_second": 2.368, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-7488.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-7488.json new file mode 100644 index 0000000000000000000000000000000000000000..824675cbb84476332c3602b5d4598f6f377d999b --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-7488.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-7488", + "checkpoint": "checkpoint-7488", + "global_step": 7488, + "eval_loss": 4.478168487548828, + "metrics": { + "eval_loss": 4.478168487548828, + "eval_model_preparation_time": 0.0056, + "eval_runtime": 128.9748, + "eval_samples_per_second": 18.934, + "eval_steps_per_second": 2.373, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-7776.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-7776.json new file mode 100644 index 0000000000000000000000000000000000000000..f92e44baf5b3b274773b64cf7c40b299683a22f6 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-7776.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-7776", + "checkpoint": "checkpoint-7776", + "global_step": 7776, + "eval_loss": 4.4994401931762695, + "metrics": { + "eval_loss": 4.4994401931762695, + "eval_model_preparation_time": 0.0056, + "eval_runtime": 128.9996, + "eval_samples_per_second": 18.93, + "eval_steps_per_second": 2.372, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-8064.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-8064.json new file mode 100644 index 0000000000000000000000000000000000000000..f148e1c0484da5fc03fb2069cdc70198bf9697c1 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-8064.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8064", + "checkpoint": "checkpoint-8064", + "global_step": 8064, + "eval_loss": 4.5242133140563965, + "metrics": { + "eval_loss": 4.5242133140563965, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.0053, + "eval_samples_per_second": 18.929, + "eval_steps_per_second": 2.372, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-8352.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-8352.json new file mode 100644 index 0000000000000000000000000000000000000000..979975185c02753c9ed5dfbd62cec0faeda7e4cf --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-8352.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8352", + "checkpoint": "checkpoint-8352", + "global_step": 8352, + "eval_loss": 4.560777187347412, + "metrics": { + "eval_loss": 4.560777187347412, + "eval_model_preparation_time": 0.0054, + "eval_runtime": 128.9737, + "eval_samples_per_second": 18.934, + "eval_steps_per_second": 2.373, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-864.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-864.json new file mode 100644 index 0000000000000000000000000000000000000000..e6f2f15da9d83f3703be2a9ad6e250137e7deb9c --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-864.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-864", + "checkpoint": "checkpoint-864", + "global_step": 864, + "eval_loss": 4.530415058135986, + "metrics": { + "eval_loss": 4.530415058135986, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.1114, + "eval_samples_per_second": 18.914, + "eval_steps_per_second": 2.37, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-8640.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-8640.json new file mode 100644 index 0000000000000000000000000000000000000000..9728590602dec966fc06a77f7ee3d22b0156aaa4 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-8640.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8640", + "checkpoint": "checkpoint-8640", + "global_step": 8640, + "eval_loss": 4.578827381134033, + "metrics": { + "eval_loss": 4.578827381134033, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 128.9468, + "eval_samples_per_second": 18.938, + "eval_steps_per_second": 2.373, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-8928.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-8928.json new file mode 100644 index 0000000000000000000000000000000000000000..b1cd5daa1ba399e7f38630d439d8e89db5309e8a --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-8928.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8928", + "checkpoint": "checkpoint-8928", + "global_step": 8928, + "eval_loss": 4.612637996673584, + "metrics": { + "eval_loss": 4.612637996673584, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.0879, + "eval_samples_per_second": 18.917, + "eval_steps_per_second": 2.37, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-9216.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-9216.json new file mode 100644 index 0000000000000000000000000000000000000000..f4ef18d823b254d57fc96227389522cb830b0ed8 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-9216.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-9216", + "checkpoint": "checkpoint-9216", + "global_step": 9216, + "eval_loss": 4.631955623626709, + "metrics": { + "eval_loss": 4.631955623626709, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.2128, + "eval_samples_per_second": 18.899, + "eval_steps_per_second": 2.368, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-9504.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-9504.json new file mode 100644 index 0000000000000000000000000000000000000000..18e6bf2f4bbdb0c8a5ddad00567988a034e64357 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-9504.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-9504", + "checkpoint": "checkpoint-9504", + "global_step": 9504, + "eval_loss": 4.650114059448242, + "metrics": { + "eval_loss": 4.650114059448242, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.1763, + "eval_samples_per_second": 18.904, + "eval_steps_per_second": 2.369, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-9792.json b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-9792.json new file mode 100644 index 0000000000000000000000000000000000000000..2d0f3b8e81fb2d6f2ccabe4b4efc9cd15ff03a87 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/checkpoint-9792.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-9792", + "checkpoint": "checkpoint-9792", + "global_step": 9792, + "eval_loss": 4.661818504333496, + "metrics": { + "eval_loss": 4.661818504333496, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.1761, + "eval_samples_per_second": 18.904, + "eval_steps_per_second": 2.369, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask15-fulle-lm/l2r_n1_eval/summary.jsonl b/results/mask15-fulle-lm/l2r_n1_eval/summary.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..ab54c589c7d166fd2a268327a07887ca89ef5e37 --- /dev/null +++ b/results/mask15-fulle-lm/l2r_n1_eval/summary.jsonl @@ -0,0 +1,37 @@ +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-72", "checkpoint": "checkpoint-72", "global_step": 72, "eval_loss": 7.526080131530762, "metrics": {"eval_loss": 7.526080131530762, "eval_model_preparation_time": 0.0034, "eval_runtime": 126.4927, "eval_samples_per_second": 19.305, "eval_steps_per_second": 2.419, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-288", "checkpoint": "checkpoint-288", "global_step": 288, "eval_loss": 6.486622333526611, "metrics": {"eval_loss": 6.486622333526611, "eval_model_preparation_time": 0.0057, "eval_runtime": 128.6597, "eval_samples_per_second": 18.98, "eval_steps_per_second": 2.378, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-576", "checkpoint": "checkpoint-576", "global_step": 576, "eval_loss": 5.569174766540527, "metrics": {"eval_loss": 5.569174766540527, "eval_model_preparation_time": 0.0052, "eval_runtime": 128.9688, "eval_samples_per_second": 18.935, "eval_steps_per_second": 2.373, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-864", "checkpoint": "checkpoint-864", "global_step": 864, "eval_loss": 4.530415058135986, "metrics": {"eval_loss": 4.530415058135986, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.1114, "eval_samples_per_second": 18.914, "eval_steps_per_second": 2.37, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-1152", "checkpoint": "checkpoint-1152", "global_step": 1152, "eval_loss": 4.138202667236328, "metrics": {"eval_loss": 4.138202667236328, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.2493, "eval_samples_per_second": 18.894, "eval_steps_per_second": 2.368, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-1440", "checkpoint": "checkpoint-1440", "global_step": 1440, "eval_loss": 3.9913787841796875, "metrics": {"eval_loss": 3.9913787841796875, "eval_model_preparation_time": 0.0059, "eval_runtime": 129.2595, "eval_samples_per_second": 18.892, "eval_steps_per_second": 2.367, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-1728", "checkpoint": "checkpoint-1728", "global_step": 1728, "eval_loss": 3.928884506225586, "metrics": {"eval_loss": 3.928884506225586, "eval_model_preparation_time": 0.0057, "eval_runtime": 129.08, "eval_samples_per_second": 18.919, "eval_steps_per_second": 2.371, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2016", "checkpoint": "checkpoint-2016", "global_step": 2016, "eval_loss": 3.9102258682250977, "metrics": {"eval_loss": 3.9102258682250977, "eval_model_preparation_time": 0.0044, "eval_runtime": 128.9893, "eval_samples_per_second": 18.932, "eval_steps_per_second": 2.372, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2304", "checkpoint": "checkpoint-2304", "global_step": 2304, "eval_loss": 3.911813497543335, "metrics": {"eval_loss": 3.911813497543335, "eval_model_preparation_time": 0.0036, "eval_runtime": 128.8471, "eval_samples_per_second": 18.953, "eval_steps_per_second": 2.375, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2592", "checkpoint": "checkpoint-2592", "global_step": 2592, "eval_loss": 3.9276037216186523, "metrics": {"eval_loss": 3.9276037216186523, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.0827, "eval_samples_per_second": 18.918, "eval_steps_per_second": 2.371, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2880", "checkpoint": "checkpoint-2880", "global_step": 2880, "eval_loss": 3.9492850303649902, "metrics": {"eval_loss": 3.9492850303649902, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.0449, "eval_samples_per_second": 18.924, "eval_steps_per_second": 2.371, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-3168", "checkpoint": "checkpoint-3168", "global_step": 3168, "eval_loss": 3.9791922569274902, "metrics": {"eval_loss": 3.9791922569274902, "eval_model_preparation_time": 0.0035, "eval_runtime": 128.9926, "eval_samples_per_second": 18.931, "eval_steps_per_second": 2.372, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-3456", "checkpoint": "checkpoint-3456", "global_step": 3456, "eval_loss": 4.006196022033691, "metrics": {"eval_loss": 4.006196022033691, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.007, "eval_samples_per_second": 18.929, "eval_steps_per_second": 2.372, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-3744", "checkpoint": "checkpoint-3744", "global_step": 3744, "eval_loss": 4.035167217254639, "metrics": {"eval_loss": 4.035167217254639, "eval_model_preparation_time": 0.0052, "eval_runtime": 128.9832, "eval_samples_per_second": 18.933, "eval_steps_per_second": 2.372, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4032", "checkpoint": "checkpoint-4032", "global_step": 4032, "eval_loss": 4.076192378997803, "metrics": {"eval_loss": 4.076192378997803, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.1461, "eval_samples_per_second": 18.909, "eval_steps_per_second": 2.369, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4320", "checkpoint": "checkpoint-4320", "global_step": 4320, "eval_loss": 4.109894752502441, "metrics": {"eval_loss": 4.109894752502441, "eval_model_preparation_time": 0.0036, "eval_runtime": 128.919, "eval_samples_per_second": 18.942, "eval_steps_per_second": 2.374, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4608", "checkpoint": "checkpoint-4608", "global_step": 4608, "eval_loss": 4.143605709075928, "metrics": {"eval_loss": 4.143605709075928, "eval_model_preparation_time": 0.0041, "eval_runtime": 129.1433, "eval_samples_per_second": 18.909, "eval_steps_per_second": 2.369, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4896", "checkpoint": "checkpoint-4896", "global_step": 4896, "eval_loss": 4.178235054016113, "metrics": {"eval_loss": 4.178235054016113, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.2611, "eval_samples_per_second": 18.892, "eval_steps_per_second": 2.367, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-5184", "checkpoint": "checkpoint-5184", "global_step": 5184, "eval_loss": 4.207855224609375, "metrics": {"eval_loss": 4.207855224609375, "eval_model_preparation_time": 0.0038, "eval_runtime": 129.1778, "eval_samples_per_second": 18.904, "eval_steps_per_second": 2.369, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-5472", "checkpoint": "checkpoint-5472", "global_step": 5472, "eval_loss": 4.2506256103515625, "metrics": {"eval_loss": 4.2506256103515625, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.113, "eval_samples_per_second": 18.914, "eval_steps_per_second": 2.37, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-5760", "checkpoint": "checkpoint-5760", "global_step": 5760, "eval_loss": 4.276021480560303, "metrics": {"eval_loss": 4.276021480560303, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.276, "eval_samples_per_second": 18.89, "eval_steps_per_second": 2.367, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6048", "checkpoint": "checkpoint-6048", "global_step": 6048, "eval_loss": 4.306779861450195, "metrics": {"eval_loss": 4.306779861450195, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.061, "eval_samples_per_second": 18.921, "eval_steps_per_second": 2.371, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6336", "checkpoint": "checkpoint-6336", "global_step": 6336, "eval_loss": 4.3341827392578125, "metrics": {"eval_loss": 4.3341827392578125, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.0846, "eval_samples_per_second": 18.918, "eval_steps_per_second": 2.371, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6624", "checkpoint": "checkpoint-6624", "global_step": 6624, "eval_loss": 4.370403289794922, "metrics": {"eval_loss": 4.370403289794922, "eval_model_preparation_time": 0.0056, "eval_runtime": 129.038, "eval_samples_per_second": 18.925, "eval_steps_per_second": 2.371, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6912", "checkpoint": "checkpoint-6912", "global_step": 6912, "eval_loss": 4.416069507598877, "metrics": {"eval_loss": 4.416069507598877, "eval_model_preparation_time": 0.0036, "eval_runtime": 128.9833, "eval_samples_per_second": 18.933, "eval_steps_per_second": 2.372, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-7200", "checkpoint": "checkpoint-7200", "global_step": 7200, "eval_loss": 4.452248573303223, "metrics": {"eval_loss": 4.452248573303223, "eval_model_preparation_time": 0.0054, "eval_runtime": 129.2346, "eval_samples_per_second": 18.896, "eval_steps_per_second": 2.368, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-7488", "checkpoint": "checkpoint-7488", "global_step": 7488, "eval_loss": 4.478168487548828, "metrics": {"eval_loss": 4.478168487548828, "eval_model_preparation_time": 0.0056, "eval_runtime": 128.9748, "eval_samples_per_second": 18.934, "eval_steps_per_second": 2.373, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-7776", "checkpoint": "checkpoint-7776", "global_step": 7776, "eval_loss": 4.4994401931762695, "metrics": {"eval_loss": 4.4994401931762695, "eval_model_preparation_time": 0.0056, "eval_runtime": 128.9996, "eval_samples_per_second": 18.93, "eval_steps_per_second": 2.372, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8064", "checkpoint": "checkpoint-8064", "global_step": 8064, "eval_loss": 4.5242133140563965, "metrics": {"eval_loss": 4.5242133140563965, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.0053, "eval_samples_per_second": 18.929, "eval_steps_per_second": 2.372, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8352", "checkpoint": "checkpoint-8352", "global_step": 8352, "eval_loss": 4.560777187347412, "metrics": {"eval_loss": 4.560777187347412, "eval_model_preparation_time": 0.0054, "eval_runtime": 128.9737, "eval_samples_per_second": 18.934, "eval_steps_per_second": 2.373, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8640", "checkpoint": "checkpoint-8640", "global_step": 8640, "eval_loss": 4.578827381134033, "metrics": {"eval_loss": 4.578827381134033, "eval_model_preparation_time": 0.0035, "eval_runtime": 128.9468, "eval_samples_per_second": 18.938, "eval_steps_per_second": 2.373, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8928", "checkpoint": "checkpoint-8928", "global_step": 8928, "eval_loss": 4.612637996673584, "metrics": {"eval_loss": 4.612637996673584, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.0879, "eval_samples_per_second": 18.917, "eval_steps_per_second": 2.37, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-9216", "checkpoint": "checkpoint-9216", "global_step": 9216, "eval_loss": 4.631955623626709, "metrics": {"eval_loss": 4.631955623626709, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.2128, "eval_samples_per_second": 18.899, "eval_steps_per_second": 2.368, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-9504", "checkpoint": "checkpoint-9504", "global_step": 9504, "eval_loss": 4.650114059448242, "metrics": {"eval_loss": 4.650114059448242, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.1763, "eval_samples_per_second": 18.904, "eval_steps_per_second": 2.369, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-9792", "checkpoint": "checkpoint-9792", "global_step": 9792, "eval_loss": 4.661818504333496, "metrics": {"eval_loss": 4.661818504333496, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.1761, "eval_samples_per_second": 18.904, "eval_steps_per_second": 2.369, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-10080", "checkpoint": "checkpoint-10080", "global_step": 10080, "eval_loss": 4.687160015106201, "metrics": {"eval_loss": 4.687160015106201, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.0142, "eval_samples_per_second": 18.928, "eval_steps_per_second": 2.372, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-10368", "checkpoint": "checkpoint-10368", "global_step": 10368, "eval_loss": 4.723207950592041, "metrics": {"eval_loss": 4.723207950592041, "eval_model_preparation_time": 0.0043, "eval_runtime": 129.1996, "eval_samples_per_second": 18.901, "eval_steps_per_second": 2.368, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-10080.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-10080.json new file mode 100644 index 0000000000000000000000000000000000000000..9c31774e034739d4c907310fa8371afb65f26f1c --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-10080.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-10080", + "checkpoint": "checkpoint-10080", + "global_step": 10080, + "tokens_seen": 10569646080, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2538338976299542, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2538338976299542 + ], + "acc": 0.2533359888468433, + "acc_ci": 0.0, + "acc_runs": [ + 0.2533359888468433 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004343142545094425, + "centered_acc": 0.005111863506605567, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005111863506605567 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5250272034820457, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5250272034820457 + ], + "acc": 0.5397170837867247, + "acc_ci": 0.0, + "acc_runs": [ + 0.5397170837867247 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011651200967623562, + "centered_acc": 0.05005440696409136, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05005440696409136 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2689393939393939, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2689393939393939 + ], + "acc": 0.24831649831649832, + "acc_ci": 0.0, + "acc_runs": [ + 0.24831649831649832 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009098548093009251, + "centered_acc": 0.02525252525252523, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02525252525252523 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.257679180887372, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.257679180887372 + ], + "acc": 0.20051194539249148, + "acc_ci": 0.0, + "acc_runs": [ + 0.20051194539249148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012780770562768334, + "centered_acc": 0.010238907849829356, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.010238907849829356 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.182, + "acc_ci": 0.0, + "acc_runs": [ + 0.182 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4988161010260458, + "acc_ci": 0.0, + "acc_runs": [ + 0.4988161010260458 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052446290528871, + "centered_acc": -0.0023677979479084232, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0023677979479084232 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 5418371311.711865, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 5418371311.711865 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2934132793231276, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2926985079142304, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.006126130091939594, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2934132793231276 + ], + "mean_acc_runs": [ + 0.2926985079142304 + ], + "mean_centered_acc_runs": [ + -0.006126130091939594 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-10368.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-10368.json new file mode 100644 index 0000000000000000000000000000000000000000..54987e26f2323f86dedc6ec3d052e3fcef55a20e --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-10368.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-10368", + "checkpoint": "checkpoint-10368", + "global_step": 10368, + "tokens_seen": 10871635968, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25184226249751046, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25184226249751046 + ], + "acc": 0.2533359888468433, + "acc_ci": 0.0, + "acc_runs": [ + 0.2533359888468433 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004331840012787794, + "centered_acc": 0.0024563499966806144, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0024563499966806144 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.529923830250272, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.529923830250272 + ], + "acc": 0.5386289445048966, + "acc_ci": 0.0, + "acc_runs": [ + 0.5386289445048966 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011644913435420134, + "centered_acc": 0.05984766050054402, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05984766050054402 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26430976430976433, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26430976430976433 + ], + "acc": 0.2537878787878788, + "acc_ci": 0.0, + "acc_runs": [ + 0.2537878787878788 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009048410451862887, + "centered_acc": 0.019079685746352437, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019079685746352437 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2525597269624573, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2525597269624573 + ], + "acc": 0.197098976109215, + "acc_ci": 0.0, + "acc_runs": [ + 0.197098976109215 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012696728980207687, + "centered_acc": 0.003412969283276427, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.003412969283276427 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.178, + "acc_ci": 0.0, + "acc_runs": [ + 0.178 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4996053670086819, + "acc_ci": 0.0, + "acc_runs": [ + 0.4996053670086819 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405248130604961, + "centered_acc": -0.0007892659826361781, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0007892659826361781 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.64, + "acc_ci": 0.0, + "acc_runs": [ + 0.64 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.048241815132442176, + "centered_acc": 0.28, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.28 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 9574311667.190289, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 9574311667.190289 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2946059306700007, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29857446170312163, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.008805443306476068, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2946059306700007 + ], + "mean_acc_runs": [ + 0.29857446170312163 + ], + "mean_centered_acc_runs": [ + 0.008805443306476068 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-1152.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-1152.json new file mode 100644 index 0000000000000000000000000000000000000000..a48927edd5520194567800ce39c1a63e3a80232f --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-1152.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-1152", + "checkpoint": "checkpoint-1152", + "global_step": 1152, + "tokens_seen": 1207959552, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2574188408683529, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2574188408683529 + ], + "acc": 0.253734315873332, + "acc_ci": 0.0, + "acc_runs": [ + 0.253734315873332 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043631851720466585, + "centered_acc": 0.009891787824470569, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009891787824470569 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5282916213275299, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5282916213275299 + ], + "acc": 0.5391730141458106, + "acc_ci": 0.0, + "acc_runs": [ + 0.5391730141458106 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0116471341727495, + "centered_acc": 0.05658324265505987, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05658324265505987 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2697811447811448, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2697811447811448 + ], + "acc": 0.2601010101010101, + "acc_ci": 0.0, + "acc_runs": [ + 0.2601010101010101 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009107527914671116, + "centered_acc": 0.02637485970819305, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02637485970819305 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2593856655290102, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2593856655290102 + ], + "acc": 0.19880546075085323, + "acc_ci": 0.0, + "acc_runs": [ + 0.19880546075085323 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012808273573927224, + "centered_acc": 0.012514220705346949, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012514220705346949 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.156, + "acc_ci": 0.0, + "acc_runs": [ + 0.156 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.500394632991318, + "acc_ci": 0.0, + "acc_runs": [ + 0.500394632991318 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405248130604961, + "centered_acc": 0.0007892659826360671, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0007892659826360671 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 3908422.579860111, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 3908422.579860111 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.179, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.179 + ], + "acc": 0.042, + "acc_ci": 0.0, + "acc_runs": [ + 0.042 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012128730605719054, + "centered_acc": -0.09466666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09466666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29664621208433967, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2868495895636024, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0115487073603955, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29664621208433967 + ], + "mean_acc_runs": [ + 0.2868495895636024 + ], + "mean_centered_acc_runs": [ + -0.0115487073603955 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-1440.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-1440.json new file mode 100644 index 0000000000000000000000000000000000000000..a4b3977d80f18ba51cb1b27cf0c92551e14ab087 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-1440.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-1440", + "checkpoint": "checkpoint-1440", + "global_step": 1440, + "tokens_seen": 1509949440, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2526389165504879, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2526389165504879 + ], + "acc": 0.25492929695279826, + "acc_ci": 0.0, + "acc_runs": [ + 0.25492929695279826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004336375492801405, + "centered_acc": 0.0035185554006505657, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0035185554006505657 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.530467899891186, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.530467899891186 + ], + "acc": 0.5429815016322089, + "acc_ci": 0.0, + "acc_runs": [ + 0.5429815016322089 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011644145418355097, + "centered_acc": 0.06093579978237207, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06093579978237207 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2702020202020202, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2702020202020202 + ], + "acc": 0.25841750841750843, + "acc_ci": 0.0, + "acc_runs": [ + 0.25841750841750843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009112002229119622, + "centered_acc": 0.026936026936026963, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.026936026936026963 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26023890784982934, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26023890784982934 + ], + "acc": 0.21245733788395904, + "acc_ci": 0.0, + "acc_runs": [ + 0.21245733788395904 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012821930225112672, + "centered_acc": 0.013651877133105783, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013651877133105783 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.292, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.292 + ], + "acc": 0.154, + "acc_ci": 0.0, + "acc_runs": [ + 0.154 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020354375480530096, + "centered_acc": 0.05599999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05599999999999997 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5059194948697711, + "acc_ci": 0.0, + "acc_runs": [ + 0.5059194948697711 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051500838485923, + "centered_acc": 0.011838989739542116, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011838989739542116 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 6169027.649689698, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 6169027.649689698 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2980912907489206, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29429926015299457, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.0022358710228627446, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2980912907489206 + ], + "mean_acc_runs": [ + 0.29429926015299457 + ], + "mean_centered_acc_runs": [ + 0.0022358710228627446 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-1728.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-1728.json new file mode 100644 index 0000000000000000000000000000000000000000..6174d7b53c8f33fa37389df1e486bd1efb5d04f2 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-1728.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-1728", + "checkpoint": "checkpoint-1728", + "global_step": 1728, + "tokens_seen": 1811939328, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2523401712806214, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2523401712806214 + ], + "acc": 0.2555267874925314, + "acc_ci": 0.0, + "acc_runs": [ + 0.2555267874925314 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004334676952703991, + "centered_acc": 0.0031202283741618895, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0031202283741618895 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5277475516866159, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5277475516866159 + ], + "acc": 0.5408052230685527, + "acc_ci": 0.0, + "acc_runs": [ + 0.5408052230685527 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011647846656062001, + "centered_acc": 0.055495103373231824, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.055495103373231824 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2668350168350168, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2668350168350168 + ], + "acc": 0.26641414141414144, + "acc_ci": 0.0, + "acc_runs": [ + 0.26641414141414144 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009075915859266983, + "centered_acc": 0.02244668911335575, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02244668911335575 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2508532423208191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2508532423208191 + ], + "acc": 0.20477815699658702, + "acc_ci": 0.0, + "acc_runs": [ + 0.20477815699658702 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012668198621315347, + "centered_acc": 0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.156, + "acc_ci": 0.0, + "acc_runs": [ + 0.156 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5082872928176796, + "acc_ci": 0.0, + "acc_runs": [ + 0.5082872928176796 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405055532282417, + "centered_acc": 0.016574585635359185, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.016574585635359185 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 6938443.271501095, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 6938443.271501095 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.05, + "acc_ci": 0.0, + "acc_runs": [ + 0.05 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2921293303538456, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29500990635631924, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.002738979280970214, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2921293303538456 + ], + "mean_acc_runs": [ + 0.29500990635631924 + ], + "mean_centered_acc_runs": [ + -0.002738979280970214 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-2016.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-2016.json new file mode 100644 index 0000000000000000000000000000000000000000..dd5833705ffffdfd6f9121aebd48624036219584 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-2016.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2016", + "checkpoint": "checkpoint-2016", + "global_step": 2016, + "tokens_seen": 2113929216, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25542720573590916, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25542720573590916 + ], + "acc": 0.25642302330213107, + "acc_ci": 0.0, + "acc_runs": [ + 0.25642302330213107 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004352098082984506, + "centered_acc": 0.007236274314545543, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007236274314545543 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5331882480957563, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5331882480957563 + ], + "acc": 0.544613710554951, + "acc_ci": 0.0, + "acc_runs": [ + 0.544613710554951 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011640096923563308, + "centered_acc": 0.06637649619151254, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06637649619151254 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2676767676767677, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2676767676767677 + ], + "acc": 0.25715488215488214, + "acc_ci": 0.0, + "acc_runs": [ + 0.25715488215488214 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009085000147099275, + "centered_acc": 0.023569023569023573, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023569023569023573 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.257679180887372, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.257679180887372 + ], + "acc": 0.21160409556313994, + "acc_ci": 0.0, + "acc_runs": [ + 0.21160409556313994 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012780770562768334, + "centered_acc": 0.010238907849829356, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.010238907849829356 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.27, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27 + ], + "acc": 0.162, + "acc_ci": 0.0, + "acc_runs": [ + 0.162 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019874354831287473, + "centered_acc": 0.02666666666666669, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02666666666666669 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5043409629044988, + "acc_ci": 0.0, + "acc_runs": [ + 0.5043409629044988 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051956064076884, + "centered_acc": 0.008681925808997626, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008681925808997626 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.56, + "acc_ci": 0.0, + "acc_runs": [ + 0.56 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049888765156985884, + "centered_acc": 0.1200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1200000000000001 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 7943611.446547047, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 7943611.446547047 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2944952337326342, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2918424136253303, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.007776568376150807, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2944952337326342 + ], + "mean_acc_runs": [ + 0.2918424136253303 + ], + "mean_centered_acc_runs": [ + -0.007776568376150807 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-2304.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-2304.json new file mode 100644 index 0000000000000000000000000000000000000000..f2b97b9e36cee726ca37a40689ec89240a3c9c3d --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-2304.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2304", + "checkpoint": "checkpoint-2304", + "global_step": 2304, + "tokens_seen": 2415919104, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2531368253335989, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2531368253335989 + ], + "acc": 0.2551284604660426, + "acc_ci": 0.0, + "acc_runs": [ + 0.2551284604660426 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004339200363454809, + "centered_acc": 0.00418243377813184, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.00418243377813184 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5250272034820457, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5250272034820457 + ], + "acc": 0.544069640914037, + "acc_ci": 0.0, + "acc_runs": [ + 0.544069640914037 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011651200967623562, + "centered_acc": 0.05005440696409136, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05005440696409136 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2697811447811448, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2697811447811448 + ], + "acc": 0.2542087542087542, + "acc_ci": 0.0, + "acc_runs": [ + 0.2542087542087542 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009107527914671116, + "centered_acc": 0.02637485970819305, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02637485970819305 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24914675767918087, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24914675767918087 + ], + "acc": 0.21160409556313994, + "acc_ci": 0.0, + "acc_runs": [ + 0.21160409556313994 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012639407111926505, + "centered_acc": -0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.176, + "acc_ci": 0.0, + "acc_runs": [ + 0.176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4980268350434096, + "acc_ci": 0.0, + "acc_runs": [ + 0.4980268350434096 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052376259225643, + "centered_acc": -0.003946329913180779, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.003946329913180779 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 13514248.98151079, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 13514248.98151079 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.184, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.184 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01225945734093864, + "centered_acc": -0.08800000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08800000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29351532187932833, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2940325247969084, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.006210818038124752, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29351532187932833 + ], + "mean_acc_runs": [ + 0.2940325247969084 + ], + "mean_centered_acc_runs": [ + -0.006210818038124752 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-2592.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-2592.json new file mode 100644 index 0000000000000000000000000000000000000000..8a77c5b94440e96bc72c5fcb0b9a15f3c624f099 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-2592.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2592", + "checkpoint": "checkpoint-2592", + "global_step": 2592, + "tokens_seen": 2717908992, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2528380800637323, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2528380800637323 + ], + "acc": 0.25712009559848636, + "acc_ci": 0.0, + "acc_runs": [ + 0.25712009559848636 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043375063448996155, + "centered_acc": 0.0037841067516430904, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0037841067516430904 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5195865070729053, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5195865070729053 + ], + "acc": 0.5473340587595212, + "acc_ci": 0.0, + "acc_runs": [ + 0.5473340587595212 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165686997928841, + "centered_acc": 0.03917301414581065, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03917301414581065 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2697811447811448, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2697811447811448 + ], + "acc": 0.2537878787878788, + "acc_ci": 0.0, + "acc_runs": [ + 0.2537878787878788 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009107527914671116, + "centered_acc": 0.02637485970819305, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02637485970819305 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26023890784982934, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26023890784982934 + ], + "acc": 0.2167235494880546, + "acc_ci": 0.0, + "acc_runs": [ + 0.2167235494880546 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012821930225112672, + "centered_acc": 0.013651877133105783, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013651877133105783 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.278, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.278 + ], + "acc": 0.17, + "acc_ci": 0.0, + "acc_runs": [ + 0.17 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020055833888070865, + "centered_acc": 0.037333333333333364, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037333333333333364 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5209155485398579, + "acc_ci": 0.0, + "acc_runs": [ + 0.5209155485398579 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014040185494212997, + "centered_acc": 0.04183109707971577, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04183109707971577 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 13131296.934187809, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 13131296.934187809 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.178, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.178 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012102167676183571, + "centered_acc": -0.09600000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09600000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29307410662793526, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2967168592947499, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0019196431445330847, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29307410662793526 + ], + "mean_acc_runs": [ + 0.2967168592947499 + ], + "mean_centered_acc_runs": [ + -0.0019196431445330847 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-288.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-288.json new file mode 100644 index 0000000000000000000000000000000000000000..d6f80da41cf75093ede090b6fdd431b76de0efd4 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-288.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-288", + "checkpoint": "checkpoint-288", + "global_step": 288, + "tokens_seen": 301989888, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2545309699263095, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2545309699263095 + ], + "acc": 0.25672176857199763, + "acc_ci": 0.0, + "acc_runs": [ + 0.25672176857199763 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043470700195274055, + "centered_acc": 0.006041293235079366, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006041293235079366 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.514145810663765, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.514145810663765 + ], + "acc": 0.5261153427638737, + "acc_ci": 0.0, + "acc_runs": [ + 0.5261153427638737 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011661154475524689, + "centered_acc": 0.028291621327529937, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.028291621327529937 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2516835016835017, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2516835016835017 + ], + "acc": 0.24915824915824916, + "acc_ci": 0.0, + "acc_runs": [ + 0.24915824915824916 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008905088235948856, + "centered_acc": 0.002244668911335568, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.002244668911335568 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25597269624573377, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25597269624573377 + ], + "acc": 0.20477815699658702, + "acc_ci": 0.0, + "acc_runs": [ + 0.20477815699658702 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012753013241244613, + "centered_acc": 0.007963594994311688, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007963594994311688 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.164, + "acc_ci": 0.0, + "acc_runs": [ + 0.164 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4972375690607735, + "acc_ci": 0.0, + "acc_runs": [ + 0.4972375690607735 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052271211616401, + "centered_acc": -0.0055248618784530246, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0055248618784530246 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 238698.5004554275, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 238698.5004554275 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.174, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.174 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011994493230973449, + "centered_acc": -0.10133333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10133333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2887221630865517, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28692985483251815, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.018119491836606938, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2887221630865517 + ], + "mean_acc_runs": [ + 0.28692985483251815 + ], + "mean_centered_acc_runs": [ + -0.018119491836606938 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-2880.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-2880.json new file mode 100644 index 0000000000000000000000000000000000000000..ac445329e0710b96438aa9ade3e76a60e89bd756 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-2880.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2880", + "checkpoint": "checkpoint-2880", + "global_step": 2880, + "tokens_seen": 3019898880, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2535351523600876, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2535351523600876 + ], + "acc": 0.2550288787094204, + "acc_ci": 0.0, + "acc_runs": [ + 0.2550288787094204 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004341454841892553, + "centered_acc": 0.004713536480116816, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.004713536480116816 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5255712731229597, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5255712731229597 + ], + "acc": 0.545157780195865, + "acc_ci": 0.0, + "acc_runs": [ + 0.545157780195865 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165055784457337, + "centered_acc": 0.05114254624591941, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05114254624591941 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2689393939393939, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2689393939393939 + ], + "acc": 0.25925925925925924, + "acc_ci": 0.0, + "acc_runs": [ + 0.25925925925925924 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009098548093009251, + "centered_acc": 0.02525252525252523, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02525252525252523 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25426621160409557, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25426621160409557 + ], + "acc": 0.21245733788395904, + "acc_ci": 0.0, + "acc_runs": [ + 0.21245733788395904 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012724999945157845, + "centered_acc": 0.0056882821387940945, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0056882821387940945 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.272, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.272 + ], + "acc": 0.162, + "acc_ci": 0.0, + "acc_runs": [ + 0.162 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019920483209566107, + "centered_acc": 0.02933333333333336, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02933333333333336 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5019731649565904, + "acc_ci": 0.0, + "acc_runs": [ + 0.5019731649565904 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052376259225643, + "centered_acc": 0.003946329913180779, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.003946329913180779 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 14108333.68989918, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 14108333.68989918 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.18, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.18 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012155153135512024, + "centered_acc": -0.09333333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09333333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2923853385044228, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29491638827787947, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.004075761824673674, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2923853385044228 + ], + "mean_acc_runs": [ + 0.29491638827787947 + ], + "mean_centered_acc_runs": [ + -0.004075761824673674 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-3168.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-3168.json new file mode 100644 index 0000000000000000000000000000000000000000..ff9036dd14c47a18b5d9b366cca2e7f98741ceba --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-3168.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-3168", + "checkpoint": "checkpoint-3168", + "global_step": 3168, + "tokens_seen": 3321888768, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25582553276239794, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25582553276239794 + ], + "acc": 0.2559251145190201, + "acc_ci": 0.0, + "acc_runs": [ + 0.2559251145190201 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043543250171373706, + "centered_acc": 0.007767377016530593, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007767377016530593 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5179542981501633, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5179542981501633 + ], + "acc": 0.5402611534276387, + "acc_ci": 0.0, + "acc_runs": [ + 0.5402611534276387 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011658300623287121, + "centered_acc": 0.0359085963003265, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0359085963003265 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2647306397306397, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2647306397306397 + ], + "acc": 0.257996632996633, + "acc_ci": 0.0, + "acc_runs": [ + 0.257996632996633 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009053021086174166, + "centered_acc": 0.019640852974186274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019640852974186274 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24744027303754265, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24744027303754265 + ], + "acc": 0.21331058020477817, + "acc_ci": 0.0, + "acc_runs": [ + 0.21331058020477817 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01261035266329266, + "centered_acc": -0.0034129692832764644, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0034129692832764644 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.166, + "acc_ci": 0.0, + "acc_runs": [ + 0.166 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5074980268350434, + "acc_ci": 0.0, + "acc_runs": [ + 0.5074980268350434 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050905521228646, + "centered_acc": 0.014996053670086829, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014996053670086829 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 17457173.74951655, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 17457173.74951655 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.187, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.187 + ], + "acc": 0.048, + "acc_ci": 0.0, + "acc_runs": [ + 0.048 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012336254828074166, + "centered_acc": -0.084, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.084 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29249179061345726, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29472789697568136, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.00553983323423101, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29249179061345726 + ], + "mean_acc_runs": [ + 0.29472789697568136 + ], + "mean_centered_acc_runs": [ + -0.00553983323423101 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-3456.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-3456.json new file mode 100644 index 0000000000000000000000000000000000000000..a87412705195e3fca27effd41380a7935670a410 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-3456.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-3456", + "checkpoint": "checkpoint-3456", + "global_step": 3456, + "tokens_seen": 3623878656, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25194184425413263, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25194184425413263 + ], + "acc": 0.2581159131647082, + "acc_ci": 0.0, + "acc_runs": [ + 0.2581159131647082 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004332408005299815, + "centered_acc": 0.0025891256721768396, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0025891256721768396 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5223068552774756, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5223068552774756 + ], + "acc": 0.545157780195865, + "acc_ci": 0.0, + "acc_runs": [ + 0.545157780195865 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011654208652596384, + "centered_acc": 0.044613710554951114, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.044613710554951114 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26725589225589225, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26725589225589225 + ], + "acc": 0.25547138047138046, + "acc_ci": 0.0, + "acc_runs": [ + 0.25547138047138046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009080463246017426, + "centered_acc": 0.023007856341189663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023007856341189663 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25853242320819114, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25853242320819114 + ], + "acc": 0.2098976109215017, + "acc_ci": 0.0, + "acc_runs": [ + 0.2098976109215017 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012794553754288777, + "centered_acc": 0.011376564277588189, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011376564277588189 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.168, + "acc_ci": 0.0, + "acc_runs": [ + 0.168 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5074980268350434, + "acc_ci": 0.0, + "acc_runs": [ + 0.5074980268350434 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050905521228646, + "centered_acc": 0.014996053670086829, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014996053670086829 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 22249751.478349194, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 22249751.478349194 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.18, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.18 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012155153135512024, + "centered_acc": -0.09333333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09333333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2923395024992819, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29384281733621986, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.007575011029993348, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2923395024992819 + ], + "mean_acc_runs": [ + 0.29384281733621986 + ], + "mean_centered_acc_runs": [ + -0.007575011029993348 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-3744.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-3744.json new file mode 100644 index 0000000000000000000000000000000000000000..1d5e3f7ec18aea587fc30819b0701fd8c5e7b377 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-3744.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-3744", + "checkpoint": "checkpoint-3744", + "global_step": 3744, + "tokens_seen": 3925868544, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25692093208524197, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25692093208524197 + ], + "acc": 0.25692093208524197, + "acc_ci": 0.0, + "acc_runs": [ + 0.25692093208524197 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004360424536144852, + "centered_acc": 0.009227909446989294, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009227909446989294 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5223068552774756, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5223068552774756 + ], + "acc": 0.5413492927094669, + "acc_ci": 0.0, + "acc_runs": [ + 0.5413492927094669 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011654208652596384, + "centered_acc": 0.044613710554951114, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.044613710554951114 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2638888888888889, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2638888888888889 + ], + "acc": 0.2529461279461279, + "acc_ci": 0.0, + "acc_runs": [ + 0.2529461279461279 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009043789220055392, + "centered_acc": 0.018518518518518528, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018518518518518528 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24829351535836178, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24829351535836178 + ], + "acc": 0.2098976109215017, + "acc_ci": 0.0, + "acc_runs": [ + 0.2098976109215017 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012624912868089679, + "centered_acc": -0.0022753128555176305, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0022753128555176305 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.294, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.294 + ], + "acc": 0.174, + "acc_ci": 0.0, + "acc_runs": [ + 0.174 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02039509548493655, + "centered_acc": 0.058666666666666645, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.058666666666666645 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.500394632991318, + "acc_ci": 0.0, + "acc_runs": [ + 0.500394632991318 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405248130604961, + "centered_acc": 0.0007892659826360671, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0007892659826360671 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 20744945.13655991, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 20744945.13655991 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29473503193499473, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29287960584273565, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.007024183496854309, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29473503193499473 + ], + "mean_acc_runs": [ + 0.29287960584273565 + ], + "mean_centered_acc_runs": [ + -0.007024183496854309 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-4032.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-4032.json new file mode 100644 index 0000000000000000000000000000000000000000..85e995ab136e7f7dfae3c3a6c6c15984ac88d9f3 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-4032.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4032", + "checkpoint": "checkpoint-4032", + "global_step": 4032, + "tokens_seen": 4227858432, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2539334793865764, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2539334793865764 + ], + "acc": 0.25801633140808605, + "acc_ci": 0.0, + "acc_runs": [ + 0.25801633140808605 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004343704512379704, + "centered_acc": 0.005244639182101866, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005244639182101866 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5190424374319913, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5190424374319913 + ], + "acc": 0.544613710554951, + "acc_ci": 0.0, + "acc_runs": [ + 0.544613710554951 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011657360703051524, + "centered_acc": 0.0380848748639826, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0380848748639826 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2689393939393939, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2689393939393939 + ], + "acc": 0.2537878787878788, + "acc_ci": 0.0, + "acc_runs": [ + 0.2537878787878788 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009098548093009251, + "centered_acc": 0.02525252525252523, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02525252525252523 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2645051194539249, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2645051194539249 + ], + "acc": 0.2090443686006826, + "acc_ci": 0.0, + "acc_runs": [ + 0.2090443686006826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012889272949313479, + "centered_acc": 0.019340159271899877, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019340159271899877 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.176, + "acc_ci": 0.0, + "acc_runs": [ + 0.176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.49329123914759276, + "acc_ci": 0.0, + "acc_runs": [ + 0.49329123914759276 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051220692330247, + "centered_acc": -0.013417521704814472, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.013417521704814472 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 29776814.064322088, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 29776814.064322088 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.184, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.184 + ], + "acc": 0.048, + "acc_ci": 0.0, + "acc_runs": [ + 0.048 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01225945734093864, + "centered_acc": -0.08800000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08800000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2954034050353144, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2931040990272892, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.008250414768915306, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2954034050353144 + ], + "mean_acc_runs": [ + 0.2931040990272892 + ], + "mean_centered_acc_runs": [ + -0.008250414768915306 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-4320.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-4320.json new file mode 100644 index 0000000000000000000000000000000000000000..6d7d310440cc5d6399465fa0462cd5e3d9bd71f3 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-4320.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4320", + "checkpoint": "checkpoint-4320", + "global_step": 4320, + "tokens_seen": 4529848320, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25413264289982074, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25413264289982074 + ], + "acc": 0.25473013343955386, + "acc_ci": 0.0, + "acc_runs": [ + 0.25473013343955386 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004344827546976306, + "centered_acc": 0.005510190533094317, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005510190533094317 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5174102285092492, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5174102285092492 + ], + "acc": 0.5489662676822633, + "acc_ci": 0.0, + "acc_runs": [ + 0.5489662676822633 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011658749823107507, + "centered_acc": 0.03482045701849845, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03482045701849845 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2689393939393939, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2689393939393939 + ], + "acc": 0.2563131313131313, + "acc_ci": 0.0, + "acc_runs": [ + 0.2563131313131313 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009098548093009251, + "centered_acc": 0.02525252525252523, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02525252525252523 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26109215017064846, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26109215017064846 + ], + "acc": 0.2158703071672355, + "acc_ci": 0.0, + "acc_runs": [ + 0.2158703071672355 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012835523909473708, + "centered_acc": 0.014789533560864617, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014789533560864617 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.172, + "acc_ci": 0.0, + "acc_runs": [ + 0.172 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48539857932123126, + "acc_ci": 0.0, + "acc_runs": [ + 0.48539857932123126 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014046492383275955, + "centered_acc": -0.02920284135753748, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.02920284135753748 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 47715722.63763751, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 47715722.63763751 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2944290692531854, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29375658806971156, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.008769097567980113, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2944290692531854 + ], + "mean_acc_runs": [ + 0.29375658806971156 + ], + "mean_centered_acc_runs": [ + -0.008769097567980113 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-4608.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-4608.json new file mode 100644 index 0000000000000000000000000000000000000000..6a72a510a4cee06c8e85f36b250171ddb65889a6 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-4608.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4608", + "checkpoint": "checkpoint-4608", + "global_step": 4608, + "tokens_seen": 4831838208, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2531368253335989, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2531368253335989 + ], + "acc": 0.2551284604660426, + "acc_ci": 0.0, + "acc_runs": [ + 0.2551284604660426 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004339200363454809, + "centered_acc": 0.00418243377813184, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.00418243377813184 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5168661588683352, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5168661588683352 + ], + "acc": 0.5413492927094669, + "acc_ci": 0.0, + "acc_runs": [ + 0.5413492927094669 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011659185184878962, + "centered_acc": 0.0337323177366704, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0337323177366704 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26725589225589225, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26725589225589225 + ], + "acc": 0.25462962962962965, + "acc_ci": 0.0, + "acc_runs": [ + 0.25462962962962965 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009080463246017426, + "centered_acc": 0.023007856341189663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023007856341189663 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2551194539249147, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2551194539249147 + ], + "acc": 0.20733788395904437, + "acc_ci": 0.0, + "acc_runs": [ + 0.20733788395904437 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012739038695202038, + "centered_acc": 0.006825938566552929, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006825938566552929 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.174, + "acc_ci": 0.0, + "acc_runs": [ + 0.174 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4925019731649566, + "acc_ci": 0.0, + "acc_runs": [ + 0.4925019731649566 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050905521228646, + "centered_acc": -0.014996053670086829, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.014996053670086829 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 49952985.055417165, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 49952985.055417165 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.18, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.18 + ], + "acc": 0.042, + "acc_ci": 0.0, + "acc_runs": [ + 0.042 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012155153135512024, + "centered_acc": -0.09333333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09333333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29306305506379016, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.291523470170284, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.010667324114830536, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29306305506379016 + ], + "mean_acc_runs": [ + 0.291523470170284 + ], + "mean_centered_acc_runs": [ + -0.010667324114830536 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-4896.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-4896.json new file mode 100644 index 0000000000000000000000000000000000000000..fecef5ff5961e7d4c5299c8747d9953682d69dc0 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-4896.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4896", + "checkpoint": "checkpoint-4896", + "global_step": 4896, + "tokens_seen": 5133828096, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2529376618203545, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2529376618203545 + ], + "acc": 0.2551284604660426, + "acc_ci": 0.0, + "acc_runs": [ + 0.2551284604660426 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043380713189125845, + "centered_acc": 0.003916882427139316, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.003916882427139316 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5114254624591947, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5114254624591947 + ], + "acc": 0.5397170837867247, + "acc_ci": 0.0, + "acc_runs": [ + 0.5397170837867247 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011662778026451529, + "centered_acc": 0.02285092491838947, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02285092491838947 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.27230639730639733, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27230639730639733 + ], + "acc": 0.2521043771043771, + "acc_ci": 0.0, + "acc_runs": [ + 0.2521043771043771 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009134218447652479, + "centered_acc": 0.02974186307519644, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02974186307519644 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2508532423208191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2508532423208191 + ], + "acc": 0.20563139931740615, + "acc_ci": 0.0, + "acc_runs": [ + 0.20563139931740615 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012668198621315347, + "centered_acc": 0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.284, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.284 + ], + "acc": 0.166, + "acc_ci": 0.0, + "acc_runs": [ + 0.166 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020186703693570774, + "centered_acc": 0.0453333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0453333333333333 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5043409629044988, + "acc_ci": 0.0, + "acc_runs": [ + 0.5043409629044988 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051956064076884, + "centered_acc": 0.008681925808997626, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008681925808997626 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.61, + "acc_ci": 0.0, + "acc_runs": [ + 0.61 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04902071300001973, + "centered_acc": 0.21999999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.21999999999999997 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 44213339.58909345, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 44213339.58909345 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.188, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.188 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01236158601510376, + "centered_acc": -0.08266666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08266666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29325379398446094, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29572097453527496, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.0006189825412832152, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29325379398446094 + ], + "mean_acc_runs": [ + 0.29572097453527496 + ], + "mean_centered_acc_runs": [ + 0.0006189825412832152 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-5184.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-5184.json new file mode 100644 index 0000000000000000000000000000000000000000..860f24f5b9abed180e01c5c93dd624b62ed3e2e2 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-5184.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-5184", + "checkpoint": "checkpoint-5184", + "global_step": 5184, + "tokens_seen": 5435817984, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2561242780322645, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2561242780322645 + ], + "acc": 0.2535351523600876, + "acc_ci": 0.0, + "acc_runs": [ + 0.2535351523600876 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004355992090030594, + "centered_acc": 0.008165704043019343, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008165704043019343 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5108813928182807, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5108813928182807 + ], + "acc": 0.5375408052230686, + "acc_ci": 0.0, + "acc_runs": [ + 0.5375408052230686 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011663061261117717, + "centered_acc": 0.02176278563656142, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02176278563656142 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2647306397306397, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2647306397306397 + ], + "acc": 0.2474747474747475, + "acc_ci": 0.0, + "acc_runs": [ + 0.2474747474747475 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009053021086174166, + "centered_acc": 0.019640852974186274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019640852974186274 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2508532423208191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2508532423208191 + ], + "acc": 0.20563139931740615, + "acc_ci": 0.0, + "acc_runs": [ + 0.20563139931740615 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012668198621315347, + "centered_acc": 0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.288, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.288 + ], + "acc": 0.174, + "acc_ci": 0.0, + "acc_runs": [ + 0.174 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020271503835075255, + "centered_acc": 0.05066666666666664, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05066666666666664 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4988161010260458, + "acc_ci": 0.0, + "acc_runs": [ + 0.4988161010260458 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052446290528871, + "centered_acc": -0.0023677979479084232, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0023677979479084232 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.56, + "acc_ci": 0.0, + "acc_runs": [ + 0.56 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049888765156985884, + "centered_acc": 0.1200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1200000000000001 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 84466820.64095041, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 84466820.64095041 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.182, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.182 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0122075806376622, + "centered_acc": -0.09066666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09066666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2920982588170007, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2898285667175056, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.012787319479886872, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2920982588170007 + ], + "mean_acc_runs": [ + 0.2898285667175056 + ], + "mean_centered_acc_runs": [ + -0.012787319479886872 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-5472.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-5472.json new file mode 100644 index 0000000000000000000000000000000000000000..ac9305c729885a277207eea46ee1f332278195bb --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-5472.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-5472", + "checkpoint": "checkpoint-5472", + "global_step": 5472, + "tokens_seen": 5737807872, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2555267874925314, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2555267874925314 + ], + "acc": 0.25492929695279826, + "acc_ci": 0.0, + "acc_runs": [ + 0.25492929695279826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00435265526368279, + "centered_acc": 0.007369049990041843, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007369049990041843 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5184983677910773, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5184983677910773 + ], + "acc": 0.5364526659412405, + "acc_ci": 0.0, + "acc_runs": [ + 0.5364526659412405 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165783758381818, + "centered_acc": 0.03699673558215455, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03699673558215455 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26052188552188554, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26052188552188554 + ], + "acc": 0.25462962962962965, + "acc_ci": 0.0, + "acc_runs": [ + 0.25462962962962965 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00900643589033641, + "centered_acc": 0.01402918069584739, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01402918069584739 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26109215017064846, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26109215017064846 + ], + "acc": 0.21245733788395904, + "acc_ci": 0.0, + "acc_runs": [ + 0.21245733788395904 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012835523909473708, + "centered_acc": 0.014789533560864617, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014789533560864617 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.27, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27 + ], + "acc": 0.178, + "acc_ci": 0.0, + "acc_runs": [ + 0.178 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019874354831287473, + "centered_acc": 0.02666666666666669, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02666666666666669 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4988161010260458, + "acc_ci": 0.0, + "acc_runs": [ + 0.4988161010260458 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052446290528871, + "centered_acc": -0.0023677979479084232, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0023677979479084232 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 111355658.6881594, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 111355658.6881594 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.179, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.179 + ], + "acc": 0.042, + "acc_ci": 0.0, + "acc_runs": [ + 0.042 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012128730605719054, + "centered_acc": -0.09466666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09466666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2907731984960238, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29455724932073735, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.006734263841288827, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2907731984960238 + ], + "mean_acc_runs": [ + 0.29455724932073735 + ], + "mean_centered_acc_runs": [ + -0.006734263841288827 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-576.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-576.json new file mode 100644 index 0000000000000000000000000000000000000000..9061d2b9a35d4ddd6b4bd3cf5836c9f21509012b --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-576.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-576", + "checkpoint": "checkpoint-576", + "global_step": 576, + "tokens_seen": 603979776, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25761800438159727, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25761800438159727 + ], + "acc": 0.25652260505875324, + "acc_ci": 0.0, + "acc_runs": [ + 0.25652260505875324 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004364287353415334, + "centered_acc": 0.01015733917546302, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01015733917546302 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5 + ], + "acc": 0.529923830250272, + "acc_ci": 0.0, + "acc_runs": [ + 0.529923830250272 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665824165343952, + "centered_acc": 0.0, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26725589225589225, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26725589225589225 + ], + "acc": 0.25, + "acc_ci": 0.0, + "acc_runs": [ + 0.25 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009080463246017426, + "centered_acc": 0.023007856341189663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023007856341189663 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25 + ], + "acc": 0.189419795221843, + "acc_ci": 0.0, + "acc_runs": [ + 0.189419795221843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012653835621466646, + "centered_acc": 0.0, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.306, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.306 + ], + "acc": 0.174, + "acc_ci": 0.0, + "acc_runs": [ + 0.174 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020629569998345414, + "centered_acc": 0.07466666666666666, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.07466666666666666 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48539857932123126, + "acc_ci": 0.0, + "acc_runs": [ + 0.48539857932123126 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014046492383275955, + "centered_acc": -0.02920284135753748, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.02920284135753748 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 1577499.153202225, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 1577499.153202225 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.174, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.174 + ], + "acc": 0.041, + "acc_ci": 0.0, + "acc_runs": [ + 0.041 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011994493230973449, + "centered_acc": -0.10133333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10133333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2924789827729149, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28745522716258, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.014014376551127881, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2924789827729149 + ], + "mean_acc_runs": [ + 0.28745522716258 + ], + "mean_centered_acc_runs": [ + -0.014014376551127881 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-5760.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-5760.json new file mode 100644 index 0000000000000000000000000000000000000000..8c48b9bdf8332e77b2181327f96b1ab1271e5547 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-5760.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-5760", + "checkpoint": "checkpoint-5760", + "global_step": 5760, + "tokens_seen": 6039797760, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25562636924915355, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25562636924915355 + ], + "acc": 0.2538338976299542, + "acc_ci": 0.0, + "acc_runs": [ + 0.2538338976299542 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004353212146198103, + "centered_acc": 0.007501825665538068, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007501825665538068 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5228509249183896, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5228509249183896 + ], + "acc": 0.5495103373231773, + "acc_ci": 0.0, + "acc_runs": [ + 0.5495103373231773 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011653634832401168, + "centered_acc": 0.04570184983677916, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04570184983677916 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2601010101010101, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2601010101010101 + ], + "acc": 0.24621212121212122, + "acc_ci": 0.0, + "acc_runs": [ + 0.24621212121212122 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009001718541079735, + "centered_acc": 0.013468013468013481, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013468013468013481 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2696245733788396, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2696245733788396 + ], + "acc": 0.21331058020477817, + "acc_ci": 0.0, + "acc_runs": [ + 0.21331058020477817 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012968040686869096, + "centered_acc": 0.026166097838452806, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.026166097838452806 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.188, + "acc_ci": 0.0, + "acc_runs": [ + 0.188 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48382004735595896, + "acc_ci": 0.0, + "acc_runs": [ + 0.48382004735595896 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014045126130978676, + "centered_acc": -0.03235990528808208, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.03235990528808208 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 186823097.99509394, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 186823097.99509394 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.042, + "acc_ci": 0.0, + "acc_runs": [ + 0.042 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2952004796078988, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.293497444549969, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.008031169807247922, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2952004796078988 + ], + "mean_acc_runs": [ + 0.293497444549969 + ], + "mean_centered_acc_runs": [ + -0.008031169807247922 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-6048.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-6048.json new file mode 100644 index 0000000000000000000000000000000000000000..98dabee286d7070e21ad6ea1a712ab200e97934f --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-6048.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6048", + "checkpoint": "checkpoint-6048", + "global_step": 6048, + "tokens_seen": 6341787648, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25114519020115517, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25114519020115517 + ], + "acc": 0.25323640709022105, + "acc_ci": 0.0, + "acc_runs": [ + 0.25323640709022105 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004327855588466433, + "centered_acc": 0.0015269202682068883, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0015269202682068883 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5315560391730142, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5315560391730142 + ], + "acc": 0.5369967355821545, + "acc_ci": 0.0, + "acc_runs": [ + 0.5369967355821545 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011642567710878958, + "centered_acc": 0.06311207834602839, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06311207834602839 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.273989898989899, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.273989898989899 + ], + "acc": 0.2537878787878788, + "acc_ci": 0.0, + "acc_runs": [ + 0.2537878787878788 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00915180590154416, + "centered_acc": 0.03198653198653201, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03198653198653201 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25341296928327645, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25341296928327645 + ], + "acc": 0.20563139931740615, + "acc_ci": 0.0, + "acc_runs": [ + 0.20563139931740615 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012710896778378677, + "centered_acc": 0.004550625711035261, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.004550625711035261 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.29, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.29 + ], + "acc": 0.192, + "acc_ci": 0.0, + "acc_runs": [ + 0.192 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020313179231745138, + "centered_acc": 0.05333333333333331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05333333333333331 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4909234411996843, + "acc_ci": 0.0, + "acc_runs": [ + 0.4909234411996843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050170094497657, + "centered_acc": -0.01815311760063143, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.01815311760063143 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 161489972.21028882, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 161489972.21028882 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.042, + "acc_ci": 0.0, + "acc_runs": [ + 0.042 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2971840162745575, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2942863323751045, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0018224486379364812, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2971840162745575 + ], + "mean_acc_runs": [ + 0.2942863323751045 + ], + "mean_centered_acc_runs": [ + -0.0018224486379364812 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-6336.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-6336.json new file mode 100644 index 0000000000000000000000000000000000000000..7d96114e51a79536ce2d89b2047762d8ade72819 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-6336.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6336", + "checkpoint": "checkpoint-6336", + "global_step": 6336, + "tokens_seen": 6643777536, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2517426807408883, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2517426807408883 + ], + "acc": 0.25492929695279826, + "acc_ci": 0.0, + "acc_runs": [ + 0.25492929695279826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004331271717773648, + "centered_acc": 0.0023235743211843887, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0023235743211843887 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5255712731229597, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5255712731229597 + ], + "acc": 0.544069640914037, + "acc_ci": 0.0, + "acc_runs": [ + 0.544069640914037 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165055784457337, + "centered_acc": 0.05114254624591941, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05114254624591941 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2647306397306397, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2647306397306397 + ], + "acc": 0.24663299663299662, + "acc_ci": 0.0, + "acc_runs": [ + 0.24663299663299662 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009053021086174166, + "centered_acc": 0.019640852974186274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019640852974186274 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2593856655290102, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2593856655290102 + ], + "acc": 0.20563139931740615, + "acc_ci": 0.0, + "acc_runs": [ + 0.20563139931740615 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012808273573927224, + "centered_acc": 0.012514220705346949, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012514220705346949 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.284, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.284 + ], + "acc": 0.174, + "acc_ci": 0.0, + "acc_runs": [ + 0.174 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020186703693570774, + "centered_acc": 0.0453333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0453333333333333 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5043409629044988, + "acc_ci": 0.0, + "acc_runs": [ + 0.5043409629044988 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051956064076884, + "centered_acc": 0.008681925808997626, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008681925808997626 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.61, + "acc_ci": 0.0, + "acc_runs": [ + 0.61 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04902071300001973, + "centered_acc": 0.21999999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.21999999999999997 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 159404815.95343396, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 159404815.95343396 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.187, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.187 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012336254828074166, + "centered_acc": -0.084, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.084 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2954050431872497, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29628917584954373, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.0035790418818187264, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2954050431872497 + ], + "mean_acc_runs": [ + 0.29628917584954373 + ], + "mean_centered_acc_runs": [ + 0.0035790418818187264 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-6624.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-6624.json new file mode 100644 index 0000000000000000000000000000000000000000..6a762f9e1d9821aae4c68a5d1e2ee0e409d78f4b --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-6624.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6624", + "checkpoint": "checkpoint-6624", + "global_step": 6624, + "tokens_seen": 6945767424, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25473013343955386, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25473013343955386 + ], + "acc": 0.2528380800637323, + "acc_ci": 0.0, + "acc_runs": [ + 0.2528380800637323 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043481894593367845, + "centered_acc": 0.006306844586071818, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006306844586071818 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5206746463547334, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5206746463547334 + ], + "acc": 0.5326441784548422, + "acc_ci": 0.0, + "acc_runs": [ + 0.5326441784548422 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011655846995729838, + "centered_acc": 0.041349292709466745, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.041349292709466745 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2676767676767677, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2676767676767677 + ], + "acc": 0.2441077441077441, + "acc_ci": 0.0, + "acc_runs": [ + 0.2441077441077441 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009085000147099275, + "centered_acc": 0.023569023569023573, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023569023569023573 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2645051194539249, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2645051194539249 + ], + "acc": 0.20477815699658702, + "acc_ci": 0.0, + "acc_runs": [ + 0.20477815699658702 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012889272949313479, + "centered_acc": 0.019340159271899877, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019340159271899877 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.278, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.278 + ], + "acc": 0.178, + "acc_ci": 0.0, + "acc_runs": [ + 0.178 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020055833888070865, + "centered_acc": 0.037333333333333364, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037333333333333364 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48855564325177586, + "acc_ci": 0.0, + "acc_runs": [ + 0.48855564325177586 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014048804199859391, + "centered_acc": -0.022888713496448276, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.022888713496448276 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.62, + "acc_ci": 0.0, + "acc_runs": [ + 0.62 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04878317312145634, + "centered_acc": 0.24, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.24 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 324660306.4564743, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 324660306.4564743 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.039, + "acc_ci": 0.0, + "acc_runs": [ + 0.039 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29476444448749667, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29382112646483816, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.0013612811319349317, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29476444448749667 + ], + "mean_acc_runs": [ + 0.29382112646483816 + ], + "mean_centered_acc_runs": [ + 0.0013612811319349317 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-6912.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-6912.json new file mode 100644 index 0000000000000000000000000000000000000000..cc6e2a705745155c10554f290cb8c7e2207a35aa --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-6912.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6912", + "checkpoint": "checkpoint-6912", + "global_step": 6912, + "tokens_seen": 7247757312, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2546305516829317, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2546305516829317 + ], + "acc": 0.2550288787094204, + "acc_ci": 0.0, + "acc_runs": [ + 0.2550288787094204 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004347629889041179, + "centered_acc": 0.006174068910575592, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006174068910575592 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.514145810663765, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.514145810663765 + ], + "acc": 0.5429815016322089, + "acc_ci": 0.0, + "acc_runs": [ + 0.5429815016322089 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011661154475524689, + "centered_acc": 0.028291621327529937, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.028291621327529937 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2617845117845118, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2617845117845118 + ], + "acc": 0.24873737373737373, + "acc_ci": 0.0, + "acc_runs": [ + 0.24873737373737373 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009020523527210056, + "centered_acc": 0.015712682379349047, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015712682379349047 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26109215017064846, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26109215017064846 + ], + "acc": 0.20563139931740615, + "acc_ci": 0.0, + "acc_runs": [ + 0.20563139931740615 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012835523909473708, + "centered_acc": 0.014789533560864617, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014789533560864617 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.288, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.288 + ], + "acc": 0.176, + "acc_ci": 0.0, + "acc_runs": [ + 0.176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020271503835075255, + "centered_acc": 0.05066666666666664, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05066666666666664 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5090765588003157, + "acc_ci": 0.0, + "acc_runs": [ + 0.5090765588003157 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050170094497657, + "centered_acc": 0.01815311760063132, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01815311760063132 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 360058537.6461883, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 360058537.6461883 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.177, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.177 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012075463420375087, + "centered_acc": -0.09733333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09733333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2927755040503095, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2950743173970425, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.002996746593368403, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2927755040503095 + ], + "mean_acc_runs": [ + 0.2950743173970425 + ], + "mean_centered_acc_runs": [ + -0.002996746593368403 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-72.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-72.json new file mode 100644 index 0000000000000000000000000000000000000000..3fb2fb6bf9c145c4737c0d8bc71b1bef010014c6 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-72.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-72", + "checkpoint": "checkpoint-72", + "global_step": 72, + "tokens_seen": 75497472, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26199960167297354, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26199960167297354 + ], + "acc": 0.2561242780322645, + "acc_ci": 0.0, + "acc_runs": [ + 0.2561242780322645 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004388237557526529, + "centered_acc": 0.015999468897298048, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015999468897298048 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5103373231773667, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5103373231773667 + ], + "acc": 0.5223068552774756, + "acc_ci": 0.0, + "acc_runs": [ + 0.5223068552774756 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01166333067307569, + "centered_acc": 0.020674646354733373, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020674646354733373 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2558922558922559, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2558922558922559 + ], + "acc": 0.24116161616161616, + "acc_ci": 0.0, + "acc_runs": [ + 0.24116161616161616 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008953950243013958, + "centered_acc": 0.007856341189674524, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007856341189674524 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2508532423208191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2508532423208191 + ], + "acc": 0.20051194539249148, + "acc_ci": 0.0, + "acc_runs": [ + 0.20051194539249148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012668198621315347, + "centered_acc": 0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.15, + "acc_ci": 0.0, + "acc_runs": [ + 0.15 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5011838989739542, + "acc_ci": 0.0, + "acc_runs": [ + 0.5011838989739542 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052446290528871, + "centered_acc": 0.0023677979479084232, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0023677979479084232 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 16563.52414667361, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 16563.52414667361 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.172, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.172 + ], + "acc": 0.039, + "acc_ci": 0.0, + "acc_runs": [ + 0.039 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011939788882495308, + "centered_acc": -0.10400000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10400000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2895137371772359, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28285760556115025, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.01904324062613624, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2895137371772359 + ], + "mean_acc_runs": [ + 0.28285760556115025 + ], + "mean_centered_acc_runs": [ + -0.01904324062613624 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-7200.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-7200.json new file mode 100644 index 0000000000000000000000000000000000000000..e002c02bb9ccdd60c66029f0d3d60faa01aeaf57 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-7200.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-7200", + "checkpoint": "checkpoint-7200", + "global_step": 7200, + "tokens_seen": 7549747200, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2523401712806214, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2523401712806214 + ], + "acc": 0.25323640709022105, + "acc_ci": 0.0, + "acc_runs": [ + 0.25323640709022105 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004334676952703991, + "centered_acc": 0.0031202283741618895, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0031202283741618895 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.514145810663765, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.514145810663765 + ], + "acc": 0.5342763873775843, + "acc_ci": 0.0, + "acc_runs": [ + 0.5342763873775843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011661154475524689, + "centered_acc": 0.028291621327529937, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.028291621327529937 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26851851851851855, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26851851851851855 + ], + "acc": 0.2474747474747475, + "acc_ci": 0.0, + "acc_runs": [ + 0.2474747474747475 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009094042554994908, + "centered_acc": 0.024691358024691395, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.024691358024691395 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26023890784982934, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26023890784982934 + ], + "acc": 0.21331058020477817, + "acc_ci": 0.0, + "acc_runs": [ + 0.21331058020477817 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012821930225112672, + "centered_acc": 0.013651877133105783, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013651877133105783 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.296, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.296 + ], + "acc": 0.176, + "acc_ci": 0.0, + "acc_runs": [ + 0.176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020435342091896222, + "centered_acc": 0.061333333333333316, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.061333333333333316 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4909234411996843, + "acc_ci": 0.0, + "acc_runs": [ + 0.4909234411996843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050170094497657, + "centered_acc": -0.01815311760063143, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.01815311760063143 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.62, + "acc_ci": 0.0, + "acc_runs": [ + 0.62 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04878317312145634, + "centered_acc": 0.24, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.24 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 536351121.9164391, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 536351121.9164391 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.042, + "acc_ci": 0.0, + "acc_runs": [ + 0.042 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29570723471878907, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29555090251207156, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.00224187675625091, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29570723471878907 + ], + "mean_acc_runs": [ + 0.29555090251207156 + ], + "mean_centered_acc_runs": [ + 0.00224187675625091 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-7488.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-7488.json new file mode 100644 index 0000000000000000000000000000000000000000..bcd3cb8154315cd60b95a24fb83b77c1964f65aa --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-7488.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-7488", + "checkpoint": "checkpoint-7488", + "global_step": 7488, + "tokens_seen": 7851737088, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2544313881696873, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2544313881696873 + ], + "acc": 0.2524397530372436, + "acc_ci": 0.0, + "acc_runs": [ + 0.2524397530372436 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00434650985067984, + "centered_acc": 0.005908517559583067, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005908517559583067 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5272034820457019, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5272034820457019 + ], + "acc": 0.5402611534276387, + "acc_ci": 0.0, + "acc_runs": [ + 0.5402611534276387 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01164854526242911, + "centered_acc": 0.054406964091403776, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.054406964091403776 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2647306397306397, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2647306397306397 + ], + "acc": 0.24915824915824916, + "acc_ci": 0.0, + "acc_runs": [ + 0.24915824915824916 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009053021086174166, + "centered_acc": 0.019640852974186274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019640852974186274 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2627986348122867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2627986348122867 + ], + "acc": 0.20392491467576793, + "acc_ci": 0.0, + "acc_runs": [ + 0.20392491467576793 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012862523175351397, + "centered_acc": 0.017064846416382284, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017064846416382284 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.292, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.292 + ], + "acc": 0.18, + "acc_ci": 0.0, + "acc_runs": [ + 0.18 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020354375480530096, + "centered_acc": 0.05599999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05599999999999997 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48224151539068666, + "acc_ci": 0.0, + "acc_runs": [ + 0.48224151539068666 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014043619596174988, + "centered_acc": -0.03551696921862668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.03551696921862668 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.63, + "acc_ci": 0.0, + "acc_runs": [ + 0.63 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.048523658709390974, + "centered_acc": 0.26, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.26 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 575840534.3786378, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 575840534.3786378 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.041, + "acc_ci": 0.0, + "acc_runs": [ + 0.041 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29702735745971925, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29573130474632864, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.004675459485592147, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29702735745971925 + ], + "mean_acc_runs": [ + 0.29573130474632864 + ], + "mean_centered_acc_runs": [ + 0.004675459485592147 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-7776.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-7776.json new file mode 100644 index 0000000000000000000000000000000000000000..c4527714f9c0b2185249cbea8292165c80ce98f9 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-7776.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-7776", + "checkpoint": "checkpoint-7776", + "global_step": 7776, + "tokens_seen": 8153726976, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2535351523600876, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2535351523600876 + ], + "acc": 0.2535351523600876, + "acc_ci": 0.0, + "acc_runs": [ + 0.2535351523600876 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004341454841892553, + "centered_acc": 0.004713536480116816, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.004713536480116816 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.515233949945593, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.515233949945593 + ], + "acc": 0.5380848748639826, + "acc_ci": 0.0, + "acc_runs": [ + 0.5380848748639826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011660408257153443, + "centered_acc": 0.030467899891186034, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.030467899891186034 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.25841750841750843, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25841750841750843 + ], + "acc": 0.24621212121212122, + "acc_ci": 0.0, + "acc_runs": [ + 0.24621212121212122 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008982741341291388, + "centered_acc": 0.011223344556677914, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011223344556677914 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2627986348122867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2627986348122867 + ], + "acc": 0.20051194539249148, + "acc_ci": 0.0, + "acc_runs": [ + 0.20051194539249148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012862523175351397, + "centered_acc": 0.017064846416382284, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017064846416382284 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.294, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.294 + ], + "acc": 0.178, + "acc_ci": 0.0, + "acc_runs": [ + 0.178 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02039509548493655, + "centered_acc": 0.058666666666666645, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.058666666666666645 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5114443567482242, + "acc_ci": 0.0, + "acc_runs": [ + 0.5114443567482242 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014048804199859391, + "centered_acc": 0.022888713496448387, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.022888713496448387 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.63, + "acc_ci": 0.0, + "acc_runs": [ + 0.63 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.048523658709390974, + "centered_acc": 0.26, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.26 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 1420866287.5098183, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 1420866287.5098183 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.185, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.185 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012285191326386655, + "centered_acc": -0.08666666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08666666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29483087425591264, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2982075912350607, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.008325918265356898, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29483087425591264 + ], + "mean_acc_runs": [ + 0.2982075912350607 + ], + "mean_centered_acc_runs": [ + 0.008325918265356898 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-8064.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-8064.json new file mode 100644 index 0000000000000000000000000000000000000000..997d775c79c5bbf7a1852c6090ad10f3ff7e2b38 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-8064.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8064", + "checkpoint": "checkpoint-8064", + "global_step": 8064, + "tokens_seen": 8455716864, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2546305516829317, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2546305516829317 + ], + "acc": 0.25363473411670984, + "acc_ci": 0.0, + "acc_runs": [ + 0.25363473411670984 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004347629889041179, + "centered_acc": 0.006174068910575592, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006174068910575592 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5195865070729053, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5195865070729053 + ], + "acc": 0.5408052230685527, + "acc_ci": 0.0, + "acc_runs": [ + 0.5408052230685527 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165686997928841, + "centered_acc": 0.03917301414581065, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03917301414581065 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26851851851851855, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26851851851851855 + ], + "acc": 0.24621212121212122, + "acc_ci": 0.0, + "acc_runs": [ + 0.24621212121212122 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009094042554994908, + "centered_acc": 0.024691358024691395, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.024691358024691395 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2627986348122867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2627986348122867 + ], + "acc": 0.20819112627986347, + "acc_ci": 0.0, + "acc_runs": [ + 0.20819112627986347 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012862523175351397, + "centered_acc": 0.017064846416382284, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017064846416382284 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.184, + "acc_ci": 0.0, + "acc_runs": [ + 0.184 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48539857932123126, + "acc_ci": 0.0, + "acc_runs": [ + 0.48539857932123126 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014046492383275955, + "centered_acc": -0.02920284135753748, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.02920284135753748 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.66, + "acc_ci": 0.0, + "acc_runs": [ + 0.66 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04760952285695234, + "centered_acc": 0.32000000000000006, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.32000000000000006 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 1333346339.8321455, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 1333346339.8321455 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.182, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.182 + ], + "acc": 0.041, + "acc_ci": 0.0, + "acc_runs": [ + 0.041 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0122075806376622, + "centered_acc": -0.09066666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09066666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2955890353477737, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29975292457721786, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.010200967002295157, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2955890353477737 + ], + "mean_acc_runs": [ + 0.29975292457721786 + ], + "mean_centered_acc_runs": [ + 0.010200967002295157 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-8352.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-8352.json new file mode 100644 index 0000000000000000000000000000000000000000..65a4e167802a449855d24ac93577b177ac643a1c --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-8352.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8352", + "checkpoint": "checkpoint-8352", + "global_step": 8352, + "tokens_seen": 8757706752, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25423222465644296, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25423222465644296 + ], + "acc": 0.25403306114319857, + "acc_ci": 0.0, + "acc_runs": [ + 0.25403306114319857 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00434538861452022, + "centered_acc": 0.005642966208590616, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005642966208590616 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5277475516866159, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5277475516866159 + ], + "acc": 0.5402611534276387, + "acc_ci": 0.0, + "acc_runs": [ + 0.5402611534276387 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011647846656062001, + "centered_acc": 0.055495103373231824, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.055495103373231824 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2596801346801347, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2596801346801347 + ], + "acc": 0.25252525252525254, + "acc_ci": 0.0, + "acc_runs": [ + 0.25252525252525254 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008996990428562134, + "centered_acc": 0.01290684624017957, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01290684624017957 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26791808873720135, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26791808873720135 + ], + "acc": 0.21416382252559726, + "acc_ci": 0.0, + "acc_runs": [ + 0.21416382252559726 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012942030195136482, + "centered_acc": 0.023890784982935138, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023890784982935138 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.288, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.288 + ], + "acc": 0.184, + "acc_ci": 0.0, + "acc_runs": [ + 0.184 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020271503835075255, + "centered_acc": 0.05066666666666664, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05066666666666664 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48303078137332284, + "acc_ci": 0.0, + "acc_runs": [ + 0.48303078137332284 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014044390401613118, + "centered_acc": -0.033938437253354325, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.033938437253354325 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.62, + "acc_ci": 0.0, + "acc_runs": [ + 0.62 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04878317312145634, + "centered_acc": 0.24, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.24 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 2343983809.724925, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 2343983809.724925 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2964296666267325, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.296930153276871, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.0021376504184055633, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2964296666267325 + ], + "mean_acc_runs": [ + 0.296930153276871 + ], + "mean_centered_acc_runs": [ + 0.0021376504184055633 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-864.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-864.json new file mode 100644 index 0000000000000000000000000000000000000000..2ffa5e82b26286c55503b626cc25654e2de9c919 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-864.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-864", + "checkpoint": "checkpoint-864", + "global_step": 864, + "tokens_seen": 905969664, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2522405895239992, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2522405895239992 + ], + "acc": 0.2561242780322645, + "acc_ci": 0.0, + "acc_runs": [ + 0.2561242780322645 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004334110169012113, + "centered_acc": 0.00298745269866559, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.00298745269866559 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5255712731229597, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5255712731229597 + ], + "acc": 0.529379760609358, + "acc_ci": 0.0, + "acc_runs": [ + 0.529379760609358 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165055784457337, + "centered_acc": 0.05114254624591941, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05114254624591941 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2702020202020202, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2702020202020202 + ], + "acc": 0.2596801346801347, + "acc_ci": 0.0, + "acc_runs": [ + 0.2596801346801347 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009112002229119622, + "centered_acc": 0.026936026936026963, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.026936026936026963 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24658703071672355, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24658703071672355 + ], + "acc": 0.20136518771331058, + "acc_ci": 0.0, + "acc_runs": [ + 0.20136518771331058 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012595726268789992, + "centered_acc": -0.004550625711035261, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.004550625711035261 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.154, + "acc_ci": 0.0, + "acc_runs": [ + 0.154 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5138121546961326, + "acc_ci": 0.0, + "acc_runs": [ + 0.5138121546961326 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014047122916440271, + "centered_acc": 0.027624309392265234, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.027624309392265234 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 2074566.1803452503, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 2074566.1803452503 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.178, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.178 + ], + "acc": 0.042, + "acc_ci": 0.0, + "acc_runs": [ + 0.042 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012102167676183571, + "centered_acc": -0.09600000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09600000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2924334855942838, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29146489775049006, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0036243000248989887, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2924334855942838 + ], + "mean_acc_runs": [ + 0.29146489775049006 + ], + "mean_centered_acc_runs": [ + -0.0036243000248989887 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-8640.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-8640.json new file mode 100644 index 0000000000000000000000000000000000000000..c0093c6100427ff2cd72655739cb5396ad526542 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-8640.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8640", + "checkpoint": "checkpoint-8640", + "global_step": 8640, + "tokens_seen": 9059696640, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2517426807408883, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2517426807408883 + ], + "acc": 0.2504481179047998, + "acc_ci": 0.0, + "acc_runs": [ + 0.2504481179047998 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004331271717773648, + "centered_acc": 0.0023235743211843887, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0023235743211843887 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5130576713819369, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5130576713819369 + ], + "acc": 0.5364526659412405, + "acc_ci": 0.0, + "acc_runs": [ + 0.5364526659412405 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011661845375886444, + "centered_acc": 0.02611534276387384, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02611534276387384 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2718855218855219, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2718855218855219 + ], + "acc": 0.25462962962962965, + "acc_ci": 0.0, + "acc_runs": [ + 0.25462962962962965 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009129795867310605, + "centered_acc": 0.02918069584736253, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02918069584736253 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2627986348122867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2627986348122867 + ], + "acc": 0.20477815699658702, + "acc_ci": 0.0, + "acc_runs": [ + 0.20477815699658702 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012862523175351397, + "centered_acc": 0.017064846416382284, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017064846416382284 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.292, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.292 + ], + "acc": 0.184, + "acc_ci": 0.0, + "acc_runs": [ + 0.184 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020354375480530096, + "centered_acc": 0.05599999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05599999999999997 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4940805051302289, + "acc_ci": 0.0, + "acc_runs": [ + 0.4940805051302289 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051500838485925, + "centered_acc": -0.011838989739542227, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.011838989739542227 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 2731795470.477959, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 2731795470.477959 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.185, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.185 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012285191326386655, + "centered_acc": -0.08666666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08666666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29608075147010565, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29396765373761863, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0034718081677783666, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29608075147010565 + ], + "mean_acc_runs": [ + 0.29396765373761863 + ], + "mean_centered_acc_runs": [ + -0.0034718081677783666 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-8928.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-8928.json new file mode 100644 index 0000000000000000000000000000000000000000..424cd149550b623e72cf64641c11a255a24e99c2 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-8928.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8928", + "checkpoint": "checkpoint-8928", + "global_step": 8928, + "tokens_seen": 9361686528, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25253933479386576, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25253933479386576 + ], + "acc": 0.2539334793865764, + "acc_ci": 0.0, + "acc_runs": [ + 0.2539334793865764 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004335809614480365, + "centered_acc": 0.00338577972515434, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.00338577972515434 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5179542981501633, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5179542981501633 + ], + "acc": 0.5402611534276387, + "acc_ci": 0.0, + "acc_runs": [ + 0.5402611534276387 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011658300623287121, + "centered_acc": 0.0359085963003265, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0359085963003265 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2622053872053872, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2622053872053872 + ], + "acc": 0.24957912457912457, + "acc_ci": 0.0, + "acc_runs": [ + 0.24957912457912457 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009025197991724937, + "centered_acc": 0.01627384960718296, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01627384960718296 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26023890784982934, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26023890784982934 + ], + "acc": 0.2150170648464164, + "acc_ci": 0.0, + "acc_runs": [ + 0.2150170648464164 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012821930225112672, + "centered_acc": 0.013651877133105783, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013651877133105783 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.198, + "acc_ci": 0.0, + "acc_runs": [ + 0.198 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.47513812154696133, + "acc_ci": 0.0, + "acc_runs": [ + 0.47513812154696133 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01403510288362781, + "centered_acc": -0.04972375690607733, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.04972375690607733 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.64, + "acc_ci": 0.0, + "acc_runs": [ + 0.64 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.048241815132442176, + "centered_acc": 0.28, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.28 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 2380239310.07631, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 2380239310.07631 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.04, + "acc_ci": 0.0, + "acc_runs": [ + 0.04 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2936563213332076, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2990216405560418, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.0016375484526399498, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2936563213332076 + ], + "mean_acc_runs": [ + 0.2990216405560418 + ], + "mean_centered_acc_runs": [ + 0.0016375484526399498 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-9216.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-9216.json new file mode 100644 index 0000000000000000000000000000000000000000..ac6866ec57341fab1eb0aefa839756bbb246a416 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-9216.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-9216", + "checkpoint": "checkpoint-9216", + "global_step": 9216, + "tokens_seen": 9663676416, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.252141007767377, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.252141007767377 + ], + "acc": 0.2513443537143995, + "acc_ci": 0.0, + "acc_runs": [ + 0.2513443537143995 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004333543083293202, + "centered_acc": 0.0028546770231693643, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0028546770231693643 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5212187159956474, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5212187159956474 + ], + "acc": 0.5348204570184983, + "acc_ci": 0.0, + "acc_runs": [ + 0.5348204570184983 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011655314732288771, + "centered_acc": 0.042437431991294794, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.042437431991294794 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26557239057239057, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26557239057239057 + ], + "acc": 0.2478956228956229, + "acc_ci": 0.0, + "acc_runs": [ + 0.2478956228956229 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00906221062697191, + "centered_acc": 0.020763187429854096, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020763187429854096 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2525597269624573, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2525597269624573 + ], + "acc": 0.20819112627986347, + "acc_ci": 0.0, + "acc_runs": [ + 0.20819112627986347 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012696728980207687, + "centered_acc": 0.003412969283276427, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.003412969283276427 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.278, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.278 + ], + "acc": 0.176, + "acc_ci": 0.0, + "acc_runs": [ + 0.176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020055833888070865, + "centered_acc": 0.037333333333333364, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037333333333333364 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4877663772691397, + "acc_ci": 0.0, + "acc_runs": [ + 0.4877663772691397 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01404827882040573, + "centered_acc": -0.024467245461720633, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.024467245461720633 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.61, + "acc_ci": 0.0, + "acc_runs": [ + 0.61 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04902071300001973, + "centered_acc": 0.21999999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.21999999999999997 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 3063566637.8900723, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 3063566637.8900723 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.18, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.18 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012155153135512024, + "centered_acc": -0.09333333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09333333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2915819735496454, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29373053989512243, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0038248951318583684, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2915819735496454 + ], + "mean_acc_runs": [ + 0.29373053989512243 + ], + "mean_centered_acc_runs": [ + -0.0038248951318583684 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-9504.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-9504.json new file mode 100644 index 0000000000000000000000000000000000000000..7ba18a5b924f3333685769b41ac9a901a9c1c67f --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-9504.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-9504", + "checkpoint": "checkpoint-9504", + "global_step": 9504, + "tokens_seen": 9965666304, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25114519020115517, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25114519020115517 + ], + "acc": 0.25184226249751046, + "acc_ci": 0.0, + "acc_runs": [ + 0.25184226249751046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004327855588466433, + "centered_acc": 0.0015269202682068883, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0015269202682068883 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.529379760609358, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.529379760609358 + ], + "acc": 0.5391730141458106, + "acc_ci": 0.0, + "acc_runs": [ + 0.5391730141458106 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01164566756505079, + "centered_acc": 0.05875952121871597, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05875952121871597 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2676767676767677, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2676767676767677 + ], + "acc": 0.25547138047138046, + "acc_ci": 0.0, + "acc_runs": [ + 0.25547138047138046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009085000147099275, + "centered_acc": 0.023569023569023573, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023569023569023573 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2721843003412969, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2721843003412969 + ], + "acc": 0.20051194539249148, + "acc_ci": 0.0, + "acc_runs": [ + 0.20051194539249148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01300660040642361, + "centered_acc": 0.029579067121729235, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.029579067121729235 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.18, + "acc_ci": 0.0, + "acc_runs": [ + 0.18 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.49329123914759276, + "acc_ci": 0.0, + "acc_runs": [ + 0.49329123914759276 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051220692330247, + "centered_acc": -0.013417521704814472, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.013417521704814472 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.63, + "acc_ci": 0.0, + "acc_runs": [ + 0.63 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.048523658709390974, + "centered_acc": 0.26, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.26 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 3305910352.2299867, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 3305910352.2299867 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29756433647142966, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2973577303428486, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.007769474150399461, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29756433647142966 + ], + "mean_acc_runs": [ + 0.2973577303428486 + ], + "mean_centered_acc_runs": [ + 0.007769474150399461 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/checkpoint-9792.json b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-9792.json new file mode 100644 index 0000000000000000000000000000000000000000..7cce594901c416d0b8524cb545e9826c4f47713b --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/checkpoint-9792.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", + "run": "mask15-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-9792", + "checkpoint": "checkpoint-9792", + "global_step": 9792, + "tokens_seen": 10267656192, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2533359888468433, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2533359888468433 + ], + "acc": 0.2524397530372436, + "acc_ci": 0.0, + "acc_runs": [ + 0.2524397530372436 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004340328204135388, + "centered_acc": 0.0044479851291243655, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0044479851291243655 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5315560391730142, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5315560391730142 + ], + "acc": 0.5310119695321001, + "acc_ci": 0.0, + "acc_runs": [ + 0.5310119695321001 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011642567710878958, + "centered_acc": 0.06311207834602839, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06311207834602839 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26515151515151514, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26515151515151514 + ], + "acc": 0.24621212121212122, + "acc_ci": 0.0, + "acc_runs": [ + 0.24621212121212122 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009057621139172679, + "centered_acc": 0.020202020202020183, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020202020202020183 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26706484641638223, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26706484641638223 + ], + "acc": 0.1962457337883959, + "acc_ci": 0.0, + "acc_runs": [ + 0.1962457337883959 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01292893319649643, + "centered_acc": 0.022753128555176305, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.022753128555176305 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.192, + "acc_ci": 0.0, + "acc_runs": [ + 0.192 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48539857932123126, + "acc_ci": 0.0, + "acc_runs": [ + 0.48539857932123126 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014046492383275955, + "centered_acc": -0.02920284135753748, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.02920284135753748 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.66, + "acc_ci": 0.0, + "acc_runs": [ + 0.66 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04760952285695234, + "centered_acc": 0.32000000000000006, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.32000000000000006 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 3783817234.3022566, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 3783817234.3022566 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.184, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.184 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01225945734093864, + "centered_acc": -0.08800000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08800000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2978513982646258, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29875956186647923, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.013098588269134714, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2978513982646258 + ], + "mean_acc_runs": [ + 0.29875956186647923 + ], + "mean_centered_acc_runs": [ + 0.013098588269134714 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask15-fulle-lm/zero_shot_eval/summary.jsonl b/results/mask15-fulle-lm/zero_shot_eval/summary.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..b71cfd762f3324c06a819df0c69a2cdff8ee9a23 --- /dev/null +++ b/results/mask15-fulle-lm/zero_shot_eval/summary.jsonl @@ -0,0 +1,37 @@ +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-72", "checkpoint": "checkpoint-72", "global_step": 72, "tokens_seen": 75497472, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26199960167297354, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26199960167297354], "acc": 0.2561242780322645, "acc_ci": 0.0, "acc_runs": [0.2561242780322645], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004388237557526529, "centered_acc": 0.015999468897298048, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015999468897298048], "ci": 0.0}, "piqa": {"acc_norm": 0.5103373231773667, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5103373231773667], "acc": 0.5223068552774756, "acc_ci": 0.0, "acc_runs": [0.5223068552774756], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01166333067307569, "centered_acc": 0.020674646354733373, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020674646354733373], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2558922558922559, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2558922558922559], "acc": 0.24116161616161616, "acc_ci": 0.0, "acc_runs": [0.24116161616161616], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008953950243013958, "centered_acc": 0.007856341189674524, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007856341189674524], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2508532423208191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2508532423208191], "acc": 0.20051194539249148, "acc_ci": 0.0, "acc_runs": [0.20051194539249148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012668198621315347, "centered_acc": 0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.15, "acc_ci": 0.0, "acc_runs": [0.15], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5011838989739542, "acc_ci": 0.0, "acc_runs": [0.5011838989739542], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052446290528871, "centered_acc": 0.0023677979479084232, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0023677979479084232], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 16563.52414667361, "perplexity_ci": 0.0, "perplexity_runs": [16563.52414667361], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.172, "acc_norm_ci": 0.0, "acc_norm_runs": [0.172], "acc": 0.039, "acc_ci": 0.0, "acc_runs": [0.039], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011939788882495308, "centered_acc": -0.10400000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10400000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2895137371772359, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28285760556115025, "mean_acc_ci": 0.0, "mean_centered_acc": -0.01904324062613624, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2895137371772359], "mean_acc_runs": [0.28285760556115025], "mean_centered_acc_runs": [-0.01904324062613624], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-288", "checkpoint": "checkpoint-288", "global_step": 288, "tokens_seen": 301989888, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2545309699263095, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2545309699263095], "acc": 0.25672176857199763, "acc_ci": 0.0, "acc_runs": [0.25672176857199763], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043470700195274055, "centered_acc": 0.006041293235079366, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006041293235079366], "ci": 0.0}, "piqa": {"acc_norm": 0.514145810663765, "acc_norm_ci": 0.0, "acc_norm_runs": [0.514145810663765], "acc": 0.5261153427638737, "acc_ci": 0.0, "acc_runs": [0.5261153427638737], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011661154475524689, "centered_acc": 0.028291621327529937, "centered_acc_ci": 0.0, "centered_acc_runs": [0.028291621327529937], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2516835016835017, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2516835016835017], "acc": 0.24915824915824916, "acc_ci": 0.0, "acc_runs": [0.24915824915824916], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008905088235948856, "centered_acc": 0.002244668911335568, "centered_acc_ci": 0.0, "centered_acc_runs": [0.002244668911335568], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25597269624573377, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25597269624573377], "acc": 0.20477815699658702, "acc_ci": 0.0, "acc_runs": [0.20477815699658702], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012753013241244613, "centered_acc": 0.007963594994311688, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007963594994311688], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.164, "acc_ci": 0.0, "acc_runs": [0.164], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4972375690607735, "acc_ci": 0.0, "acc_runs": [0.4972375690607735], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052271211616401, "centered_acc": -0.0055248618784530246, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0055248618784530246], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 238698.5004554275, "perplexity_ci": 0.0, "perplexity_runs": [238698.5004554275], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.174, "acc_norm_ci": 0.0, "acc_norm_runs": [0.174], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011994493230973449, "centered_acc": -0.10133333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10133333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2887221630865517, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28692985483251815, "mean_acc_ci": 0.0, "mean_centered_acc": -0.018119491836606938, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2887221630865517], "mean_acc_runs": [0.28692985483251815], "mean_centered_acc_runs": [-0.018119491836606938], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-576", "checkpoint": "checkpoint-576", "global_step": 576, "tokens_seen": 603979776, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25761800438159727, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25761800438159727], "acc": 0.25652260505875324, "acc_ci": 0.0, "acc_runs": [0.25652260505875324], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004364287353415334, "centered_acc": 0.01015733917546302, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01015733917546302], "ci": 0.0}, "piqa": {"acc_norm": 0.5, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5], "acc": 0.529923830250272, "acc_ci": 0.0, "acc_runs": [0.529923830250272], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665824165343952, "centered_acc": 0.0, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26725589225589225, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26725589225589225], "acc": 0.25, "acc_ci": 0.0, "acc_runs": [0.25], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009080463246017426, "centered_acc": 0.023007856341189663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023007856341189663], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25], "acc": 0.189419795221843, "acc_ci": 0.0, "acc_runs": [0.189419795221843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012653835621466646, "centered_acc": 0.0, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0], "ci": 0.0}, "openbookqa": {"acc_norm": 0.306, "acc_norm_ci": 0.0, "acc_norm_runs": [0.306], "acc": 0.174, "acc_ci": 0.0, "acc_runs": [0.174], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020629569998345414, "centered_acc": 0.07466666666666666, "centered_acc_ci": 0.0, "centered_acc_runs": [0.07466666666666666], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48539857932123126, "acc_ci": 0.0, "acc_runs": [0.48539857932123126], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014046492383275955, "centered_acc": -0.02920284135753748, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.02920284135753748], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 1577499.153202225, "perplexity_ci": 0.0, "perplexity_runs": [1577499.153202225], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.174, "acc_norm_ci": 0.0, "acc_norm_runs": [0.174], "acc": 0.041, "acc_ci": 0.0, "acc_runs": [0.041], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011994493230973449, "centered_acc": -0.10133333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10133333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2924789827729149, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28745522716258, "mean_acc_ci": 0.0, "mean_centered_acc": -0.014014376551127881, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2924789827729149], "mean_acc_runs": [0.28745522716258], "mean_centered_acc_runs": [-0.014014376551127881], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-864", "checkpoint": "checkpoint-864", "global_step": 864, "tokens_seen": 905969664, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2522405895239992, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2522405895239992], "acc": 0.2561242780322645, "acc_ci": 0.0, "acc_runs": [0.2561242780322645], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004334110169012113, "centered_acc": 0.00298745269866559, "centered_acc_ci": 0.0, "centered_acc_runs": [0.00298745269866559], "ci": 0.0}, "piqa": {"acc_norm": 0.5255712731229597, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5255712731229597], "acc": 0.529379760609358, "acc_ci": 0.0, "acc_runs": [0.529379760609358], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165055784457337, "centered_acc": 0.05114254624591941, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05114254624591941], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2702020202020202, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2702020202020202], "acc": 0.2596801346801347, "acc_ci": 0.0, "acc_runs": [0.2596801346801347], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009112002229119622, "centered_acc": 0.026936026936026963, "centered_acc_ci": 0.0, "centered_acc_runs": [0.026936026936026963], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24658703071672355, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24658703071672355], "acc": 0.20136518771331058, "acc_ci": 0.0, "acc_runs": [0.20136518771331058], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012595726268789992, "centered_acc": -0.004550625711035261, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.004550625711035261], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.154, "acc_ci": 0.0, "acc_runs": [0.154], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5138121546961326, "acc_ci": 0.0, "acc_runs": [0.5138121546961326], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014047122916440271, "centered_acc": 0.027624309392265234, "centered_acc_ci": 0.0, "centered_acc_runs": [0.027624309392265234], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 2074566.1803452503, "perplexity_ci": 0.0, "perplexity_runs": [2074566.1803452503], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.178, "acc_norm_ci": 0.0, "acc_norm_runs": [0.178], "acc": 0.042, "acc_ci": 0.0, "acc_runs": [0.042], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012102167676183571, "centered_acc": -0.09600000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09600000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2924334855942838, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29146489775049006, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0036243000248989887, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2924334855942838], "mean_acc_runs": [0.29146489775049006], "mean_centered_acc_runs": [-0.0036243000248989887], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-1152", "checkpoint": "checkpoint-1152", "global_step": 1152, "tokens_seen": 1207959552, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2574188408683529, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2574188408683529], "acc": 0.253734315873332, "acc_ci": 0.0, "acc_runs": [0.253734315873332], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043631851720466585, "centered_acc": 0.009891787824470569, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009891787824470569], "ci": 0.0}, "piqa": {"acc_norm": 0.5282916213275299, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5282916213275299], "acc": 0.5391730141458106, "acc_ci": 0.0, "acc_runs": [0.5391730141458106], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0116471341727495, "centered_acc": 0.05658324265505987, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05658324265505987], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2697811447811448, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2697811447811448], "acc": 0.2601010101010101, "acc_ci": 0.0, "acc_runs": [0.2601010101010101], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009107527914671116, "centered_acc": 0.02637485970819305, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02637485970819305], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2593856655290102, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2593856655290102], "acc": 0.19880546075085323, "acc_ci": 0.0, "acc_runs": [0.19880546075085323], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012808273573927224, "centered_acc": 0.012514220705346949, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012514220705346949], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.156, "acc_ci": 0.0, "acc_runs": [0.156], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.500394632991318, "acc_ci": 0.0, "acc_runs": [0.500394632991318], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405248130604961, "centered_acc": 0.0007892659826360671, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0007892659826360671], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 3908422.579860111, "perplexity_ci": 0.0, "perplexity_runs": [3908422.579860111], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.179, "acc_norm_ci": 0.0, "acc_norm_runs": [0.179], "acc": 0.042, "acc_ci": 0.0, "acc_runs": [0.042], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012128730605719054, "centered_acc": -0.09466666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09466666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29664621208433967, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2868495895636024, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0115487073603955, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29664621208433967], "mean_acc_runs": [0.2868495895636024], "mean_centered_acc_runs": [-0.0115487073603955], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-1440", "checkpoint": "checkpoint-1440", "global_step": 1440, "tokens_seen": 1509949440, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2526389165504879, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2526389165504879], "acc": 0.25492929695279826, "acc_ci": 0.0, "acc_runs": [0.25492929695279826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004336375492801405, "centered_acc": 0.0035185554006505657, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0035185554006505657], "ci": 0.0}, "piqa": {"acc_norm": 0.530467899891186, "acc_norm_ci": 0.0, "acc_norm_runs": [0.530467899891186], "acc": 0.5429815016322089, "acc_ci": 0.0, "acc_runs": [0.5429815016322089], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011644145418355097, "centered_acc": 0.06093579978237207, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06093579978237207], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2702020202020202, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2702020202020202], "acc": 0.25841750841750843, "acc_ci": 0.0, "acc_runs": [0.25841750841750843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009112002229119622, "centered_acc": 0.026936026936026963, "centered_acc_ci": 0.0, "centered_acc_runs": [0.026936026936026963], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26023890784982934, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26023890784982934], "acc": 0.21245733788395904, "acc_ci": 0.0, "acc_runs": [0.21245733788395904], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012821930225112672, "centered_acc": 0.013651877133105783, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013651877133105783], "ci": 0.0}, "openbookqa": {"acc_norm": 0.292, "acc_norm_ci": 0.0, "acc_norm_runs": [0.292], "acc": 0.154, "acc_ci": 0.0, "acc_runs": [0.154], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020354375480530096, "centered_acc": 0.05599999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05599999999999997], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5059194948697711, "acc_ci": 0.0, "acc_runs": [0.5059194948697711], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051500838485923, "centered_acc": 0.011838989739542116, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011838989739542116], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 6169027.649689698, "perplexity_ci": 0.0, "perplexity_runs": [6169027.649689698], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2980912907489206, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29429926015299457, "mean_acc_ci": 0.0, "mean_centered_acc": 0.0022358710228627446, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2980912907489206], "mean_acc_runs": [0.29429926015299457], "mean_centered_acc_runs": [0.0022358710228627446], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-1728", "checkpoint": "checkpoint-1728", "global_step": 1728, "tokens_seen": 1811939328, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2523401712806214, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2523401712806214], "acc": 0.2555267874925314, "acc_ci": 0.0, "acc_runs": [0.2555267874925314], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004334676952703991, "centered_acc": 0.0031202283741618895, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0031202283741618895], "ci": 0.0}, "piqa": {"acc_norm": 0.5277475516866159, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5277475516866159], "acc": 0.5408052230685527, "acc_ci": 0.0, "acc_runs": [0.5408052230685527], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011647846656062001, "centered_acc": 0.055495103373231824, "centered_acc_ci": 0.0, "centered_acc_runs": [0.055495103373231824], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2668350168350168, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2668350168350168], "acc": 0.26641414141414144, "acc_ci": 0.0, "acc_runs": [0.26641414141414144], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009075915859266983, "centered_acc": 0.02244668911335575, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02244668911335575], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2508532423208191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2508532423208191], "acc": 0.20477815699658702, "acc_ci": 0.0, "acc_runs": [0.20477815699658702], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012668198621315347, "centered_acc": 0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.156, "acc_ci": 0.0, "acc_runs": [0.156], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5082872928176796, "acc_ci": 0.0, "acc_runs": [0.5082872928176796], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405055532282417, "centered_acc": 0.016574585635359185, "centered_acc_ci": 0.0, "centered_acc_runs": [0.016574585635359185], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 6938443.271501095, "perplexity_ci": 0.0, "perplexity_runs": [6938443.271501095], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.05, "acc_ci": 0.0, "acc_runs": [0.05], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2921293303538456, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29500990635631924, "mean_acc_ci": 0.0, "mean_centered_acc": -0.002738979280970214, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2921293303538456], "mean_acc_runs": [0.29500990635631924], "mean_centered_acc_runs": [-0.002738979280970214], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2016", "checkpoint": "checkpoint-2016", "global_step": 2016, "tokens_seen": 2113929216, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25542720573590916, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25542720573590916], "acc": 0.25642302330213107, "acc_ci": 0.0, "acc_runs": [0.25642302330213107], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004352098082984506, "centered_acc": 0.007236274314545543, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007236274314545543], "ci": 0.0}, "piqa": {"acc_norm": 0.5331882480957563, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5331882480957563], "acc": 0.544613710554951, "acc_ci": 0.0, "acc_runs": [0.544613710554951], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011640096923563308, "centered_acc": 0.06637649619151254, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06637649619151254], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2676767676767677, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2676767676767677], "acc": 0.25715488215488214, "acc_ci": 0.0, "acc_runs": [0.25715488215488214], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009085000147099275, "centered_acc": 0.023569023569023573, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023569023569023573], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.257679180887372, "acc_norm_ci": 0.0, "acc_norm_runs": [0.257679180887372], "acc": 0.21160409556313994, "acc_ci": 0.0, "acc_runs": [0.21160409556313994], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012780770562768334, "centered_acc": 0.010238907849829356, "centered_acc_ci": 0.0, "centered_acc_runs": [0.010238907849829356], "ci": 0.0}, "openbookqa": {"acc_norm": 0.27, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27], "acc": 0.162, "acc_ci": 0.0, "acc_runs": [0.162], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019874354831287473, "centered_acc": 0.02666666666666669, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02666666666666669], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5043409629044988, "acc_ci": 0.0, "acc_runs": [0.5043409629044988], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051956064076884, "centered_acc": 0.008681925808997626, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008681925808997626], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.56, "acc_ci": 0.0, "acc_runs": [0.56], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049888765156985884, "centered_acc": 0.1200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1200000000000001], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 7943611.446547047, "perplexity_ci": 0.0, "perplexity_runs": [7943611.446547047], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2944952337326342, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2918424136253303, "mean_acc_ci": 0.0, "mean_centered_acc": -0.007776568376150807, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2944952337326342], "mean_acc_runs": [0.2918424136253303], "mean_centered_acc_runs": [-0.007776568376150807], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2304", "checkpoint": "checkpoint-2304", "global_step": 2304, "tokens_seen": 2415919104, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2531368253335989, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2531368253335989], "acc": 0.2551284604660426, "acc_ci": 0.0, "acc_runs": [0.2551284604660426], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004339200363454809, "centered_acc": 0.00418243377813184, "centered_acc_ci": 0.0, "centered_acc_runs": [0.00418243377813184], "ci": 0.0}, "piqa": {"acc_norm": 0.5250272034820457, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5250272034820457], "acc": 0.544069640914037, "acc_ci": 0.0, "acc_runs": [0.544069640914037], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011651200967623562, "centered_acc": 0.05005440696409136, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05005440696409136], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2697811447811448, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2697811447811448], "acc": 0.2542087542087542, "acc_ci": 0.0, "acc_runs": [0.2542087542087542], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009107527914671116, "centered_acc": 0.02637485970819305, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02637485970819305], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24914675767918087, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24914675767918087], "acc": 0.21160409556313994, "acc_ci": 0.0, "acc_runs": [0.21160409556313994], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012639407111926505, "centered_acc": -0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.176, "acc_ci": 0.0, "acc_runs": [0.176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4980268350434096, "acc_ci": 0.0, "acc_runs": [0.4980268350434096], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052376259225643, "centered_acc": -0.003946329913180779, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.003946329913180779], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 13514248.98151079, "perplexity_ci": 0.0, "perplexity_runs": [13514248.98151079], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.184, "acc_norm_ci": 0.0, "acc_norm_runs": [0.184], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01225945734093864, "centered_acc": -0.08800000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08800000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29351532187932833, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2940325247969084, "mean_acc_ci": 0.0, "mean_centered_acc": -0.006210818038124752, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29351532187932833], "mean_acc_runs": [0.2940325247969084], "mean_centered_acc_runs": [-0.006210818038124752], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2592", "checkpoint": "checkpoint-2592", "global_step": 2592, "tokens_seen": 2717908992, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2528380800637323, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2528380800637323], "acc": 0.25712009559848636, "acc_ci": 0.0, "acc_runs": [0.25712009559848636], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043375063448996155, "centered_acc": 0.0037841067516430904, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0037841067516430904], "ci": 0.0}, "piqa": {"acc_norm": 0.5195865070729053, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5195865070729053], "acc": 0.5473340587595212, "acc_ci": 0.0, "acc_runs": [0.5473340587595212], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165686997928841, "centered_acc": 0.03917301414581065, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03917301414581065], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2697811447811448, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2697811447811448], "acc": 0.2537878787878788, "acc_ci": 0.0, "acc_runs": [0.2537878787878788], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009107527914671116, "centered_acc": 0.02637485970819305, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02637485970819305], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26023890784982934, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26023890784982934], "acc": 0.2167235494880546, "acc_ci": 0.0, "acc_runs": [0.2167235494880546], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012821930225112672, "centered_acc": 0.013651877133105783, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013651877133105783], "ci": 0.0}, "openbookqa": {"acc_norm": 0.278, "acc_norm_ci": 0.0, "acc_norm_runs": [0.278], "acc": 0.17, "acc_ci": 0.0, "acc_runs": [0.17], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020055833888070865, "centered_acc": 0.037333333333333364, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037333333333333364], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5209155485398579, "acc_ci": 0.0, "acc_runs": [0.5209155485398579], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014040185494212997, "centered_acc": 0.04183109707971577, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04183109707971577], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 13131296.934187809, "perplexity_ci": 0.0, "perplexity_runs": [13131296.934187809], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.178, "acc_norm_ci": 0.0, "acc_norm_runs": [0.178], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012102167676183571, "centered_acc": -0.09600000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09600000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29307410662793526, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2967168592947499, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0019196431445330847, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29307410662793526], "mean_acc_runs": [0.2967168592947499], "mean_centered_acc_runs": [-0.0019196431445330847], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-2880", "checkpoint": "checkpoint-2880", "global_step": 2880, "tokens_seen": 3019898880, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2535351523600876, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2535351523600876], "acc": 0.2550288787094204, "acc_ci": 0.0, "acc_runs": [0.2550288787094204], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004341454841892553, "centered_acc": 0.004713536480116816, "centered_acc_ci": 0.0, "centered_acc_runs": [0.004713536480116816], "ci": 0.0}, "piqa": {"acc_norm": 0.5255712731229597, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5255712731229597], "acc": 0.545157780195865, "acc_ci": 0.0, "acc_runs": [0.545157780195865], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165055784457337, "centered_acc": 0.05114254624591941, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05114254624591941], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2689393939393939, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2689393939393939], "acc": 0.25925925925925924, "acc_ci": 0.0, "acc_runs": [0.25925925925925924], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009098548093009251, "centered_acc": 0.02525252525252523, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02525252525252523], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25426621160409557, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25426621160409557], "acc": 0.21245733788395904, "acc_ci": 0.0, "acc_runs": [0.21245733788395904], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012724999945157845, "centered_acc": 0.0056882821387940945, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0056882821387940945], "ci": 0.0}, "openbookqa": {"acc_norm": 0.272, "acc_norm_ci": 0.0, "acc_norm_runs": [0.272], "acc": 0.162, "acc_ci": 0.0, "acc_runs": [0.162], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019920483209566107, "centered_acc": 0.02933333333333336, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02933333333333336], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5019731649565904, "acc_ci": 0.0, "acc_runs": [0.5019731649565904], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052376259225643, "centered_acc": 0.003946329913180779, "centered_acc_ci": 0.0, "centered_acc_runs": [0.003946329913180779], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 14108333.68989918, "perplexity_ci": 0.0, "perplexity_runs": [14108333.68989918], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.18, "acc_norm_ci": 0.0, "acc_norm_runs": [0.18], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012155153135512024, "centered_acc": -0.09333333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09333333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2923853385044228, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29491638827787947, "mean_acc_ci": 0.0, "mean_centered_acc": -0.004075761824673674, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2923853385044228], "mean_acc_runs": [0.29491638827787947], "mean_centered_acc_runs": [-0.004075761824673674], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-3168", "checkpoint": "checkpoint-3168", "global_step": 3168, "tokens_seen": 3321888768, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25582553276239794, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25582553276239794], "acc": 0.2559251145190201, "acc_ci": 0.0, "acc_runs": [0.2559251145190201], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043543250171373706, "centered_acc": 0.007767377016530593, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007767377016530593], "ci": 0.0}, "piqa": {"acc_norm": 0.5179542981501633, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5179542981501633], "acc": 0.5402611534276387, "acc_ci": 0.0, "acc_runs": [0.5402611534276387], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011658300623287121, "centered_acc": 0.0359085963003265, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0359085963003265], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2647306397306397, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2647306397306397], "acc": 0.257996632996633, "acc_ci": 0.0, "acc_runs": [0.257996632996633], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009053021086174166, "centered_acc": 0.019640852974186274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019640852974186274], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24744027303754265, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24744027303754265], "acc": 0.21331058020477817, "acc_ci": 0.0, "acc_runs": [0.21331058020477817], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01261035266329266, "centered_acc": -0.0034129692832764644, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0034129692832764644], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.166, "acc_ci": 0.0, "acc_runs": [0.166], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5074980268350434, "acc_ci": 0.0, "acc_runs": [0.5074980268350434], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050905521228646, "centered_acc": 0.014996053670086829, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014996053670086829], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 17457173.74951655, "perplexity_ci": 0.0, "perplexity_runs": [17457173.74951655], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.187, "acc_norm_ci": 0.0, "acc_norm_runs": [0.187], "acc": 0.048, "acc_ci": 0.0, "acc_runs": [0.048], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012336254828074166, "centered_acc": -0.084, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.084], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29249179061345726, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29472789697568136, "mean_acc_ci": 0.0, "mean_centered_acc": -0.00553983323423101, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29249179061345726], "mean_acc_runs": [0.29472789697568136], "mean_centered_acc_runs": [-0.00553983323423101], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-3456", "checkpoint": "checkpoint-3456", "global_step": 3456, "tokens_seen": 3623878656, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25194184425413263, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25194184425413263], "acc": 0.2581159131647082, "acc_ci": 0.0, "acc_runs": [0.2581159131647082], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004332408005299815, "centered_acc": 0.0025891256721768396, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0025891256721768396], "ci": 0.0}, "piqa": {"acc_norm": 0.5223068552774756, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5223068552774756], "acc": 0.545157780195865, "acc_ci": 0.0, "acc_runs": [0.545157780195865], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011654208652596384, "centered_acc": 0.044613710554951114, "centered_acc_ci": 0.0, "centered_acc_runs": [0.044613710554951114], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26725589225589225, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26725589225589225], "acc": 0.25547138047138046, "acc_ci": 0.0, "acc_runs": [0.25547138047138046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009080463246017426, "centered_acc": 0.023007856341189663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023007856341189663], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25853242320819114, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25853242320819114], "acc": 0.2098976109215017, "acc_ci": 0.0, "acc_runs": [0.2098976109215017], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012794553754288777, "centered_acc": 0.011376564277588189, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011376564277588189], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.168, "acc_ci": 0.0, "acc_runs": [0.168], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5074980268350434, "acc_ci": 0.0, "acc_runs": [0.5074980268350434], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050905521228646, "centered_acc": 0.014996053670086829, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014996053670086829], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 22249751.478349194, "perplexity_ci": 0.0, "perplexity_runs": [22249751.478349194], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.18, "acc_norm_ci": 0.0, "acc_norm_runs": [0.18], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012155153135512024, "centered_acc": -0.09333333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09333333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2923395024992819, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29384281733621986, "mean_acc_ci": 0.0, "mean_centered_acc": -0.007575011029993348, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2923395024992819], "mean_acc_runs": [0.29384281733621986], "mean_centered_acc_runs": [-0.007575011029993348], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-3744", "checkpoint": "checkpoint-3744", "global_step": 3744, "tokens_seen": 3925868544, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25692093208524197, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25692093208524197], "acc": 0.25692093208524197, "acc_ci": 0.0, "acc_runs": [0.25692093208524197], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004360424536144852, "centered_acc": 0.009227909446989294, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009227909446989294], "ci": 0.0}, "piqa": {"acc_norm": 0.5223068552774756, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5223068552774756], "acc": 0.5413492927094669, "acc_ci": 0.0, "acc_runs": [0.5413492927094669], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011654208652596384, "centered_acc": 0.044613710554951114, "centered_acc_ci": 0.0, "centered_acc_runs": [0.044613710554951114], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2638888888888889, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2638888888888889], "acc": 0.2529461279461279, "acc_ci": 0.0, "acc_runs": [0.2529461279461279], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009043789220055392, "centered_acc": 0.018518518518518528, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018518518518518528], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24829351535836178, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24829351535836178], "acc": 0.2098976109215017, "acc_ci": 0.0, "acc_runs": [0.2098976109215017], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012624912868089679, "centered_acc": -0.0022753128555176305, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0022753128555176305], "ci": 0.0}, "openbookqa": {"acc_norm": 0.294, "acc_norm_ci": 0.0, "acc_norm_runs": [0.294], "acc": 0.174, "acc_ci": 0.0, "acc_runs": [0.174], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02039509548493655, "centered_acc": 0.058666666666666645, "centered_acc_ci": 0.0, "centered_acc_runs": [0.058666666666666645], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.500394632991318, "acc_ci": 0.0, "acc_runs": [0.500394632991318], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405248130604961, "centered_acc": 0.0007892659826360671, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0007892659826360671], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 20744945.13655991, "perplexity_ci": 0.0, "perplexity_runs": [20744945.13655991], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29473503193499473, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29287960584273565, "mean_acc_ci": 0.0, "mean_centered_acc": -0.007024183496854309, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29473503193499473], "mean_acc_runs": [0.29287960584273565], "mean_centered_acc_runs": [-0.007024183496854309], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4032", "checkpoint": "checkpoint-4032", "global_step": 4032, "tokens_seen": 4227858432, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2539334793865764, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2539334793865764], "acc": 0.25801633140808605, "acc_ci": 0.0, "acc_runs": [0.25801633140808605], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004343704512379704, "centered_acc": 0.005244639182101866, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005244639182101866], "ci": 0.0}, "piqa": {"acc_norm": 0.5190424374319913, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5190424374319913], "acc": 0.544613710554951, "acc_ci": 0.0, "acc_runs": [0.544613710554951], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011657360703051524, "centered_acc": 0.0380848748639826, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0380848748639826], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2689393939393939, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2689393939393939], "acc": 0.2537878787878788, "acc_ci": 0.0, "acc_runs": [0.2537878787878788], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009098548093009251, "centered_acc": 0.02525252525252523, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02525252525252523], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2645051194539249, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2645051194539249], "acc": 0.2090443686006826, "acc_ci": 0.0, "acc_runs": [0.2090443686006826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012889272949313479, "centered_acc": 0.019340159271899877, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019340159271899877], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.176, "acc_ci": 0.0, "acc_runs": [0.176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.49329123914759276, "acc_ci": 0.0, "acc_runs": [0.49329123914759276], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051220692330247, "centered_acc": -0.013417521704814472, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.013417521704814472], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 29776814.064322088, "perplexity_ci": 0.0, "perplexity_runs": [29776814.064322088], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.184, "acc_norm_ci": 0.0, "acc_norm_runs": [0.184], "acc": 0.048, "acc_ci": 0.0, "acc_runs": [0.048], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01225945734093864, "centered_acc": -0.08800000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08800000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2954034050353144, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2931040990272892, "mean_acc_ci": 0.0, "mean_centered_acc": -0.008250414768915306, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2954034050353144], "mean_acc_runs": [0.2931040990272892], "mean_centered_acc_runs": [-0.008250414768915306], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4320", "checkpoint": "checkpoint-4320", "global_step": 4320, "tokens_seen": 4529848320, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25413264289982074, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25413264289982074], "acc": 0.25473013343955386, "acc_ci": 0.0, "acc_runs": [0.25473013343955386], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004344827546976306, "centered_acc": 0.005510190533094317, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005510190533094317], "ci": 0.0}, "piqa": {"acc_norm": 0.5174102285092492, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5174102285092492], "acc": 0.5489662676822633, "acc_ci": 0.0, "acc_runs": [0.5489662676822633], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011658749823107507, "centered_acc": 0.03482045701849845, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03482045701849845], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2689393939393939, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2689393939393939], "acc": 0.2563131313131313, "acc_ci": 0.0, "acc_runs": [0.2563131313131313], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009098548093009251, "centered_acc": 0.02525252525252523, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02525252525252523], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26109215017064846, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26109215017064846], "acc": 0.2158703071672355, "acc_ci": 0.0, "acc_runs": [0.2158703071672355], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012835523909473708, "centered_acc": 0.014789533560864617, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014789533560864617], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.172, "acc_ci": 0.0, "acc_runs": [0.172], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48539857932123126, "acc_ci": 0.0, "acc_runs": [0.48539857932123126], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014046492383275955, "centered_acc": -0.02920284135753748, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.02920284135753748], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 47715722.63763751, "perplexity_ci": 0.0, "perplexity_runs": [47715722.63763751], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2944290692531854, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29375658806971156, "mean_acc_ci": 0.0, "mean_centered_acc": -0.008769097567980113, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2944290692531854], "mean_acc_runs": [0.29375658806971156], "mean_centered_acc_runs": [-0.008769097567980113], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4608", "checkpoint": "checkpoint-4608", "global_step": 4608, "tokens_seen": 4831838208, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2531368253335989, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2531368253335989], "acc": 0.2551284604660426, "acc_ci": 0.0, "acc_runs": [0.2551284604660426], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004339200363454809, "centered_acc": 0.00418243377813184, "centered_acc_ci": 0.0, "centered_acc_runs": [0.00418243377813184], "ci": 0.0}, "piqa": {"acc_norm": 0.5168661588683352, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5168661588683352], "acc": 0.5413492927094669, "acc_ci": 0.0, "acc_runs": [0.5413492927094669], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011659185184878962, "centered_acc": 0.0337323177366704, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0337323177366704], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26725589225589225, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26725589225589225], "acc": 0.25462962962962965, "acc_ci": 0.0, "acc_runs": [0.25462962962962965], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009080463246017426, "centered_acc": 0.023007856341189663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023007856341189663], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2551194539249147, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2551194539249147], "acc": 0.20733788395904437, "acc_ci": 0.0, "acc_runs": [0.20733788395904437], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012739038695202038, "centered_acc": 0.006825938566552929, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006825938566552929], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.174, "acc_ci": 0.0, "acc_runs": [0.174], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4925019731649566, "acc_ci": 0.0, "acc_runs": [0.4925019731649566], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050905521228646, "centered_acc": -0.014996053670086829, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.014996053670086829], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 49952985.055417165, "perplexity_ci": 0.0, "perplexity_runs": [49952985.055417165], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.18, "acc_norm_ci": 0.0, "acc_norm_runs": [0.18], "acc": 0.042, "acc_ci": 0.0, "acc_runs": [0.042], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012155153135512024, "centered_acc": -0.09333333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09333333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29306305506379016, "mean_acc_norm_ci": 0.0, "mean_acc": 0.291523470170284, "mean_acc_ci": 0.0, "mean_centered_acc": -0.010667324114830536, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29306305506379016], "mean_acc_runs": [0.291523470170284], "mean_centered_acc_runs": [-0.010667324114830536], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-4896", "checkpoint": "checkpoint-4896", "global_step": 4896, "tokens_seen": 5133828096, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2529376618203545, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2529376618203545], "acc": 0.2551284604660426, "acc_ci": 0.0, "acc_runs": [0.2551284604660426], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043380713189125845, "centered_acc": 0.003916882427139316, "centered_acc_ci": 0.0, "centered_acc_runs": [0.003916882427139316], "ci": 0.0}, "piqa": {"acc_norm": 0.5114254624591947, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5114254624591947], "acc": 0.5397170837867247, "acc_ci": 0.0, "acc_runs": [0.5397170837867247], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011662778026451529, "centered_acc": 0.02285092491838947, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02285092491838947], "ci": 0.0}, "arc_easy": {"acc_norm": 0.27230639730639733, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27230639730639733], "acc": 0.2521043771043771, "acc_ci": 0.0, "acc_runs": [0.2521043771043771], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009134218447652479, "centered_acc": 0.02974186307519644, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02974186307519644], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2508532423208191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2508532423208191], "acc": 0.20563139931740615, "acc_ci": 0.0, "acc_runs": [0.20563139931740615], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012668198621315347, "centered_acc": 0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.284, "acc_norm_ci": 0.0, "acc_norm_runs": [0.284], "acc": 0.166, "acc_ci": 0.0, "acc_runs": [0.166], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020186703693570774, "centered_acc": 0.0453333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0453333333333333], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5043409629044988, "acc_ci": 0.0, "acc_runs": [0.5043409629044988], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051956064076884, "centered_acc": 0.008681925808997626, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008681925808997626], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.61, "acc_ci": 0.0, "acc_runs": [0.61], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04902071300001973, "centered_acc": 0.21999999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.21999999999999997], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 44213339.58909345, "perplexity_ci": 0.0, "perplexity_runs": [44213339.58909345], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.188, "acc_norm_ci": 0.0, "acc_norm_runs": [0.188], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01236158601510376, "centered_acc": -0.08266666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08266666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29325379398446094, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29572097453527496, "mean_acc_ci": 0.0, "mean_centered_acc": 0.0006189825412832152, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29325379398446094], "mean_acc_runs": [0.29572097453527496], "mean_centered_acc_runs": [0.0006189825412832152], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-5184", "checkpoint": "checkpoint-5184", "global_step": 5184, "tokens_seen": 5435817984, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2561242780322645, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2561242780322645], "acc": 0.2535351523600876, "acc_ci": 0.0, "acc_runs": [0.2535351523600876], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004355992090030594, "centered_acc": 0.008165704043019343, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008165704043019343], "ci": 0.0}, "piqa": {"acc_norm": 0.5108813928182807, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5108813928182807], "acc": 0.5375408052230686, "acc_ci": 0.0, "acc_runs": [0.5375408052230686], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011663061261117717, "centered_acc": 0.02176278563656142, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02176278563656142], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2647306397306397, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2647306397306397], "acc": 0.2474747474747475, "acc_ci": 0.0, "acc_runs": [0.2474747474747475], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009053021086174166, "centered_acc": 0.019640852974186274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019640852974186274], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2508532423208191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2508532423208191], "acc": 0.20563139931740615, "acc_ci": 0.0, "acc_runs": [0.20563139931740615], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012668198621315347, "centered_acc": 0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.288, "acc_norm_ci": 0.0, "acc_norm_runs": [0.288], "acc": 0.174, "acc_ci": 0.0, "acc_runs": [0.174], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020271503835075255, "centered_acc": 0.05066666666666664, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05066666666666664], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4988161010260458, "acc_ci": 0.0, "acc_runs": [0.4988161010260458], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052446290528871, "centered_acc": -0.0023677979479084232, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0023677979479084232], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.56, "acc_ci": 0.0, "acc_runs": [0.56], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049888765156985884, "centered_acc": 0.1200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1200000000000001], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 84466820.64095041, "perplexity_ci": 0.0, "perplexity_runs": [84466820.64095041], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.182, "acc_norm_ci": 0.0, "acc_norm_runs": [0.182], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0122075806376622, "centered_acc": -0.09066666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09066666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2920982588170007, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2898285667175056, "mean_acc_ci": 0.0, "mean_centered_acc": -0.012787319479886872, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2920982588170007], "mean_acc_runs": [0.2898285667175056], "mean_centered_acc_runs": [-0.012787319479886872], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-5472", "checkpoint": "checkpoint-5472", "global_step": 5472, "tokens_seen": 5737807872, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2555267874925314, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2555267874925314], "acc": 0.25492929695279826, "acc_ci": 0.0, "acc_runs": [0.25492929695279826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00435265526368279, "centered_acc": 0.007369049990041843, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007369049990041843], "ci": 0.0}, "piqa": {"acc_norm": 0.5184983677910773, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5184983677910773], "acc": 0.5364526659412405, "acc_ci": 0.0, "acc_runs": [0.5364526659412405], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165783758381818, "centered_acc": 0.03699673558215455, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03699673558215455], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26052188552188554, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26052188552188554], "acc": 0.25462962962962965, "acc_ci": 0.0, "acc_runs": [0.25462962962962965], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00900643589033641, "centered_acc": 0.01402918069584739, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01402918069584739], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26109215017064846, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26109215017064846], "acc": 0.21245733788395904, "acc_ci": 0.0, "acc_runs": [0.21245733788395904], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012835523909473708, "centered_acc": 0.014789533560864617, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014789533560864617], "ci": 0.0}, "openbookqa": {"acc_norm": 0.27, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27], "acc": 0.178, "acc_ci": 0.0, "acc_runs": [0.178], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019874354831287473, "centered_acc": 0.02666666666666669, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02666666666666669], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4988161010260458, "acc_ci": 0.0, "acc_runs": [0.4988161010260458], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052446290528871, "centered_acc": -0.0023677979479084232, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0023677979479084232], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 111355658.6881594, "perplexity_ci": 0.0, "perplexity_runs": [111355658.6881594], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.179, "acc_norm_ci": 0.0, "acc_norm_runs": [0.179], "acc": 0.042, "acc_ci": 0.0, "acc_runs": [0.042], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012128730605719054, "centered_acc": -0.09466666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09466666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2907731984960238, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29455724932073735, "mean_acc_ci": 0.0, "mean_centered_acc": -0.006734263841288827, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2907731984960238], "mean_acc_runs": [0.29455724932073735], "mean_centered_acc_runs": [-0.006734263841288827], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-5760", "checkpoint": "checkpoint-5760", "global_step": 5760, "tokens_seen": 6039797760, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25562636924915355, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25562636924915355], "acc": 0.2538338976299542, "acc_ci": 0.0, "acc_runs": [0.2538338976299542], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004353212146198103, "centered_acc": 0.007501825665538068, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007501825665538068], "ci": 0.0}, "piqa": {"acc_norm": 0.5228509249183896, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5228509249183896], "acc": 0.5495103373231773, "acc_ci": 0.0, "acc_runs": [0.5495103373231773], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011653634832401168, "centered_acc": 0.04570184983677916, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04570184983677916], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2601010101010101, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2601010101010101], "acc": 0.24621212121212122, "acc_ci": 0.0, "acc_runs": [0.24621212121212122], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009001718541079735, "centered_acc": 0.013468013468013481, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013468013468013481], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2696245733788396, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2696245733788396], "acc": 0.21331058020477817, "acc_ci": 0.0, "acc_runs": [0.21331058020477817], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012968040686869096, "centered_acc": 0.026166097838452806, "centered_acc_ci": 0.0, "centered_acc_runs": [0.026166097838452806], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.188, "acc_ci": 0.0, "acc_runs": [0.188], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48382004735595896, "acc_ci": 0.0, "acc_runs": [0.48382004735595896], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014045126130978676, "centered_acc": -0.03235990528808208, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.03235990528808208], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 186823097.99509394, "perplexity_ci": 0.0, "perplexity_runs": [186823097.99509394], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.042, "acc_ci": 0.0, "acc_runs": [0.042], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2952004796078988, "mean_acc_norm_ci": 0.0, "mean_acc": 0.293497444549969, "mean_acc_ci": 0.0, "mean_centered_acc": -0.008031169807247922, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2952004796078988], "mean_acc_runs": [0.293497444549969], "mean_centered_acc_runs": [-0.008031169807247922], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6048", "checkpoint": "checkpoint-6048", "global_step": 6048, "tokens_seen": 6341787648, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25114519020115517, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25114519020115517], "acc": 0.25323640709022105, "acc_ci": 0.0, "acc_runs": [0.25323640709022105], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004327855588466433, "centered_acc": 0.0015269202682068883, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0015269202682068883], "ci": 0.0}, "piqa": {"acc_norm": 0.5315560391730142, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5315560391730142], "acc": 0.5369967355821545, "acc_ci": 0.0, "acc_runs": [0.5369967355821545], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011642567710878958, "centered_acc": 0.06311207834602839, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06311207834602839], "ci": 0.0}, "arc_easy": {"acc_norm": 0.273989898989899, "acc_norm_ci": 0.0, "acc_norm_runs": [0.273989898989899], "acc": 0.2537878787878788, "acc_ci": 0.0, "acc_runs": [0.2537878787878788], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00915180590154416, "centered_acc": 0.03198653198653201, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03198653198653201], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25341296928327645, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25341296928327645], "acc": 0.20563139931740615, "acc_ci": 0.0, "acc_runs": [0.20563139931740615], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012710896778378677, "centered_acc": 0.004550625711035261, "centered_acc_ci": 0.0, "centered_acc_runs": [0.004550625711035261], "ci": 0.0}, "openbookqa": {"acc_norm": 0.29, "acc_norm_ci": 0.0, "acc_norm_runs": [0.29], "acc": 0.192, "acc_ci": 0.0, "acc_runs": [0.192], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020313179231745138, "centered_acc": 0.05333333333333331, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05333333333333331], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4909234411996843, "acc_ci": 0.0, "acc_runs": [0.4909234411996843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050170094497657, "centered_acc": -0.01815311760063143, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.01815311760063143], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 161489972.21028882, "perplexity_ci": 0.0, "perplexity_runs": [161489972.21028882], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.042, "acc_ci": 0.0, "acc_runs": [0.042], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2971840162745575, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2942863323751045, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0018224486379364812, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2971840162745575], "mean_acc_runs": [0.2942863323751045], "mean_centered_acc_runs": [-0.0018224486379364812], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6336", "checkpoint": "checkpoint-6336", "global_step": 6336, "tokens_seen": 6643777536, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2517426807408883, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2517426807408883], "acc": 0.25492929695279826, "acc_ci": 0.0, "acc_runs": [0.25492929695279826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004331271717773648, "centered_acc": 0.0023235743211843887, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0023235743211843887], "ci": 0.0}, "piqa": {"acc_norm": 0.5255712731229597, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5255712731229597], "acc": 0.544069640914037, "acc_ci": 0.0, "acc_runs": [0.544069640914037], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165055784457337, "centered_acc": 0.05114254624591941, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05114254624591941], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2647306397306397, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2647306397306397], "acc": 0.24663299663299662, "acc_ci": 0.0, "acc_runs": [0.24663299663299662], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009053021086174166, "centered_acc": 0.019640852974186274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019640852974186274], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2593856655290102, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2593856655290102], "acc": 0.20563139931740615, "acc_ci": 0.0, "acc_runs": [0.20563139931740615], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012808273573927224, "centered_acc": 0.012514220705346949, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012514220705346949], "ci": 0.0}, "openbookqa": {"acc_norm": 0.284, "acc_norm_ci": 0.0, "acc_norm_runs": [0.284], "acc": 0.174, "acc_ci": 0.0, "acc_runs": [0.174], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020186703693570774, "centered_acc": 0.0453333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0453333333333333], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5043409629044988, "acc_ci": 0.0, "acc_runs": [0.5043409629044988], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051956064076884, "centered_acc": 0.008681925808997626, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008681925808997626], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.61, "acc_ci": 0.0, "acc_runs": [0.61], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04902071300001973, "centered_acc": 0.21999999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.21999999999999997], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 159404815.95343396, "perplexity_ci": 0.0, "perplexity_runs": [159404815.95343396], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.187, "acc_norm_ci": 0.0, "acc_norm_runs": [0.187], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012336254828074166, "centered_acc": -0.084, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.084], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2954050431872497, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29628917584954373, "mean_acc_ci": 0.0, "mean_centered_acc": 0.0035790418818187264, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2954050431872497], "mean_acc_runs": [0.29628917584954373], "mean_centered_acc_runs": [0.0035790418818187264], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6624", "checkpoint": "checkpoint-6624", "global_step": 6624, "tokens_seen": 6945767424, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25473013343955386, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25473013343955386], "acc": 0.2528380800637323, "acc_ci": 0.0, "acc_runs": [0.2528380800637323], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043481894593367845, "centered_acc": 0.006306844586071818, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006306844586071818], "ci": 0.0}, "piqa": {"acc_norm": 0.5206746463547334, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5206746463547334], "acc": 0.5326441784548422, "acc_ci": 0.0, "acc_runs": [0.5326441784548422], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011655846995729838, "centered_acc": 0.041349292709466745, "centered_acc_ci": 0.0, "centered_acc_runs": [0.041349292709466745], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2676767676767677, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2676767676767677], "acc": 0.2441077441077441, "acc_ci": 0.0, "acc_runs": [0.2441077441077441], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009085000147099275, "centered_acc": 0.023569023569023573, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023569023569023573], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2645051194539249, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2645051194539249], "acc": 0.20477815699658702, "acc_ci": 0.0, "acc_runs": [0.20477815699658702], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012889272949313479, "centered_acc": 0.019340159271899877, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019340159271899877], "ci": 0.0}, "openbookqa": {"acc_norm": 0.278, "acc_norm_ci": 0.0, "acc_norm_runs": [0.278], "acc": 0.178, "acc_ci": 0.0, "acc_runs": [0.178], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020055833888070865, "centered_acc": 0.037333333333333364, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037333333333333364], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48855564325177586, "acc_ci": 0.0, "acc_runs": [0.48855564325177586], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014048804199859391, "centered_acc": -0.022888713496448276, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.022888713496448276], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.62, "acc_ci": 0.0, "acc_runs": [0.62], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04878317312145634, "centered_acc": 0.24, "centered_acc_ci": 0.0, "centered_acc_runs": [0.24], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 324660306.4564743, "perplexity_ci": 0.0, "perplexity_runs": [324660306.4564743], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.039, "acc_ci": 0.0, "acc_runs": [0.039], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29476444448749667, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29382112646483816, "mean_acc_ci": 0.0, "mean_centered_acc": 0.0013612811319349317, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29476444448749667], "mean_acc_runs": [0.29382112646483816], "mean_centered_acc_runs": [0.0013612811319349317], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-6912", "checkpoint": "checkpoint-6912", "global_step": 6912, "tokens_seen": 7247757312, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2546305516829317, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2546305516829317], "acc": 0.2550288787094204, "acc_ci": 0.0, "acc_runs": [0.2550288787094204], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004347629889041179, "centered_acc": 0.006174068910575592, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006174068910575592], "ci": 0.0}, "piqa": {"acc_norm": 0.514145810663765, "acc_norm_ci": 0.0, "acc_norm_runs": [0.514145810663765], "acc": 0.5429815016322089, "acc_ci": 0.0, "acc_runs": [0.5429815016322089], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011661154475524689, "centered_acc": 0.028291621327529937, "centered_acc_ci": 0.0, "centered_acc_runs": [0.028291621327529937], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2617845117845118, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2617845117845118], "acc": 0.24873737373737373, "acc_ci": 0.0, "acc_runs": [0.24873737373737373], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009020523527210056, "centered_acc": 0.015712682379349047, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015712682379349047], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26109215017064846, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26109215017064846], "acc": 0.20563139931740615, "acc_ci": 0.0, "acc_runs": [0.20563139931740615], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012835523909473708, "centered_acc": 0.014789533560864617, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014789533560864617], "ci": 0.0}, "openbookqa": {"acc_norm": 0.288, "acc_norm_ci": 0.0, "acc_norm_runs": [0.288], "acc": 0.176, "acc_ci": 0.0, "acc_runs": [0.176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020271503835075255, "centered_acc": 0.05066666666666664, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05066666666666664], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5090765588003157, "acc_ci": 0.0, "acc_runs": [0.5090765588003157], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050170094497657, "centered_acc": 0.01815311760063132, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01815311760063132], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 360058537.6461883, "perplexity_ci": 0.0, "perplexity_runs": [360058537.6461883], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.177, "acc_norm_ci": 0.0, "acc_norm_runs": [0.177], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012075463420375087, "centered_acc": -0.09733333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09733333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2927755040503095, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2950743173970425, "mean_acc_ci": 0.0, "mean_centered_acc": -0.002996746593368403, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2927755040503095], "mean_acc_runs": [0.2950743173970425], "mean_centered_acc_runs": [-0.002996746593368403], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-7200", "checkpoint": "checkpoint-7200", "global_step": 7200, "tokens_seen": 7549747200, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2523401712806214, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2523401712806214], "acc": 0.25323640709022105, "acc_ci": 0.0, "acc_runs": [0.25323640709022105], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004334676952703991, "centered_acc": 0.0031202283741618895, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0031202283741618895], "ci": 0.0}, "piqa": {"acc_norm": 0.514145810663765, "acc_norm_ci": 0.0, "acc_norm_runs": [0.514145810663765], "acc": 0.5342763873775843, "acc_ci": 0.0, "acc_runs": [0.5342763873775843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011661154475524689, "centered_acc": 0.028291621327529937, "centered_acc_ci": 0.0, "centered_acc_runs": [0.028291621327529937], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26851851851851855, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26851851851851855], "acc": 0.2474747474747475, "acc_ci": 0.0, "acc_runs": [0.2474747474747475], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009094042554994908, "centered_acc": 0.024691358024691395, "centered_acc_ci": 0.0, "centered_acc_runs": [0.024691358024691395], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26023890784982934, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26023890784982934], "acc": 0.21331058020477817, "acc_ci": 0.0, "acc_runs": [0.21331058020477817], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012821930225112672, "centered_acc": 0.013651877133105783, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013651877133105783], "ci": 0.0}, "openbookqa": {"acc_norm": 0.296, "acc_norm_ci": 0.0, "acc_norm_runs": [0.296], "acc": 0.176, "acc_ci": 0.0, "acc_runs": [0.176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020435342091896222, "centered_acc": 0.061333333333333316, "centered_acc_ci": 0.0, "centered_acc_runs": [0.061333333333333316], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4909234411996843, "acc_ci": 0.0, "acc_runs": [0.4909234411996843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050170094497657, "centered_acc": -0.01815311760063143, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.01815311760063143], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.62, "acc_ci": 0.0, "acc_runs": [0.62], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04878317312145634, "centered_acc": 0.24, "centered_acc_ci": 0.0, "centered_acc_runs": [0.24], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 536351121.9164391, "perplexity_ci": 0.0, "perplexity_runs": [536351121.9164391], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.042, "acc_ci": 0.0, "acc_runs": [0.042], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29570723471878907, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29555090251207156, "mean_acc_ci": 0.0, "mean_centered_acc": 0.00224187675625091, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29570723471878907], "mean_acc_runs": [0.29555090251207156], "mean_centered_acc_runs": [0.00224187675625091], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-7488", "checkpoint": "checkpoint-7488", "global_step": 7488, "tokens_seen": 7851737088, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2544313881696873, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2544313881696873], "acc": 0.2524397530372436, "acc_ci": 0.0, "acc_runs": [0.2524397530372436], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00434650985067984, "centered_acc": 0.005908517559583067, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005908517559583067], "ci": 0.0}, "piqa": {"acc_norm": 0.5272034820457019, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5272034820457019], "acc": 0.5402611534276387, "acc_ci": 0.0, "acc_runs": [0.5402611534276387], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01164854526242911, "centered_acc": 0.054406964091403776, "centered_acc_ci": 0.0, "centered_acc_runs": [0.054406964091403776], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2647306397306397, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2647306397306397], "acc": 0.24915824915824916, "acc_ci": 0.0, "acc_runs": [0.24915824915824916], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009053021086174166, "centered_acc": 0.019640852974186274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019640852974186274], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2627986348122867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2627986348122867], "acc": 0.20392491467576793, "acc_ci": 0.0, "acc_runs": [0.20392491467576793], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012862523175351397, "centered_acc": 0.017064846416382284, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017064846416382284], "ci": 0.0}, "openbookqa": {"acc_norm": 0.292, "acc_norm_ci": 0.0, "acc_norm_runs": [0.292], "acc": 0.18, "acc_ci": 0.0, "acc_runs": [0.18], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020354375480530096, "centered_acc": 0.05599999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05599999999999997], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48224151539068666, "acc_ci": 0.0, "acc_runs": [0.48224151539068666], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014043619596174988, "centered_acc": -0.03551696921862668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.03551696921862668], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.63, "acc_ci": 0.0, "acc_runs": [0.63], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.048523658709390974, "centered_acc": 0.26, "centered_acc_ci": 0.0, "centered_acc_runs": [0.26], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 575840534.3786378, "perplexity_ci": 0.0, "perplexity_runs": [575840534.3786378], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.041, "acc_ci": 0.0, "acc_runs": [0.041], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29702735745971925, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29573130474632864, "mean_acc_ci": 0.0, "mean_centered_acc": 0.004675459485592147, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29702735745971925], "mean_acc_runs": [0.29573130474632864], "mean_centered_acc_runs": [0.004675459485592147], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-7776", "checkpoint": "checkpoint-7776", "global_step": 7776, "tokens_seen": 8153726976, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2535351523600876, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2535351523600876], "acc": 0.2535351523600876, "acc_ci": 0.0, "acc_runs": [0.2535351523600876], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004341454841892553, "centered_acc": 0.004713536480116816, "centered_acc_ci": 0.0, "centered_acc_runs": [0.004713536480116816], "ci": 0.0}, "piqa": {"acc_norm": 0.515233949945593, "acc_norm_ci": 0.0, "acc_norm_runs": [0.515233949945593], "acc": 0.5380848748639826, "acc_ci": 0.0, "acc_runs": [0.5380848748639826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011660408257153443, "centered_acc": 0.030467899891186034, "centered_acc_ci": 0.0, "centered_acc_runs": [0.030467899891186034], "ci": 0.0}, "arc_easy": {"acc_norm": 0.25841750841750843, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25841750841750843], "acc": 0.24621212121212122, "acc_ci": 0.0, "acc_runs": [0.24621212121212122], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008982741341291388, "centered_acc": 0.011223344556677914, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011223344556677914], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2627986348122867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2627986348122867], "acc": 0.20051194539249148, "acc_ci": 0.0, "acc_runs": [0.20051194539249148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012862523175351397, "centered_acc": 0.017064846416382284, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017064846416382284], "ci": 0.0}, "openbookqa": {"acc_norm": 0.294, "acc_norm_ci": 0.0, "acc_norm_runs": [0.294], "acc": 0.178, "acc_ci": 0.0, "acc_runs": [0.178], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02039509548493655, "centered_acc": 0.058666666666666645, "centered_acc_ci": 0.0, "centered_acc_runs": [0.058666666666666645], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5114443567482242, "acc_ci": 0.0, "acc_runs": [0.5114443567482242], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014048804199859391, "centered_acc": 0.022888713496448387, "centered_acc_ci": 0.0, "centered_acc_runs": [0.022888713496448387], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.63, "acc_ci": 0.0, "acc_runs": [0.63], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.048523658709390974, "centered_acc": 0.26, "centered_acc_ci": 0.0, "centered_acc_runs": [0.26], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 1420866287.5098183, "perplexity_ci": 0.0, "perplexity_runs": [1420866287.5098183], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.185, "acc_norm_ci": 0.0, "acc_norm_runs": [0.185], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012285191326386655, "centered_acc": -0.08666666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08666666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29483087425591264, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2982075912350607, "mean_acc_ci": 0.0, "mean_centered_acc": 0.008325918265356898, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29483087425591264], "mean_acc_runs": [0.2982075912350607], "mean_centered_acc_runs": [0.008325918265356898], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8064", "checkpoint": "checkpoint-8064", "global_step": 8064, "tokens_seen": 8455716864, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2546305516829317, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2546305516829317], "acc": 0.25363473411670984, "acc_ci": 0.0, "acc_runs": [0.25363473411670984], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004347629889041179, "centered_acc": 0.006174068910575592, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006174068910575592], "ci": 0.0}, "piqa": {"acc_norm": 0.5195865070729053, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5195865070729053], "acc": 0.5408052230685527, "acc_ci": 0.0, "acc_runs": [0.5408052230685527], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165686997928841, "centered_acc": 0.03917301414581065, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03917301414581065], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26851851851851855, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26851851851851855], "acc": 0.24621212121212122, "acc_ci": 0.0, "acc_runs": [0.24621212121212122], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009094042554994908, "centered_acc": 0.024691358024691395, "centered_acc_ci": 0.0, "centered_acc_runs": [0.024691358024691395], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2627986348122867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2627986348122867], "acc": 0.20819112627986347, "acc_ci": 0.0, "acc_runs": [0.20819112627986347], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012862523175351397, "centered_acc": 0.017064846416382284, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017064846416382284], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.184, "acc_ci": 0.0, "acc_runs": [0.184], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48539857932123126, "acc_ci": 0.0, "acc_runs": [0.48539857932123126], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014046492383275955, "centered_acc": -0.02920284135753748, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.02920284135753748], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.66, "acc_ci": 0.0, "acc_runs": [0.66], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04760952285695234, "centered_acc": 0.32000000000000006, "centered_acc_ci": 0.0, "centered_acc_runs": [0.32000000000000006], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 1333346339.8321455, "perplexity_ci": 0.0, "perplexity_runs": [1333346339.8321455], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.182, "acc_norm_ci": 0.0, "acc_norm_runs": [0.182], "acc": 0.041, "acc_ci": 0.0, "acc_runs": [0.041], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0122075806376622, "centered_acc": -0.09066666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09066666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2955890353477737, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29975292457721786, "mean_acc_ci": 0.0, "mean_centered_acc": 0.010200967002295157, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2955890353477737], "mean_acc_runs": [0.29975292457721786], "mean_centered_acc_runs": [0.010200967002295157], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8352", "checkpoint": "checkpoint-8352", "global_step": 8352, "tokens_seen": 8757706752, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25423222465644296, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25423222465644296], "acc": 0.25403306114319857, "acc_ci": 0.0, "acc_runs": [0.25403306114319857], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00434538861452022, "centered_acc": 0.005642966208590616, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005642966208590616], "ci": 0.0}, "piqa": {"acc_norm": 0.5277475516866159, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5277475516866159], "acc": 0.5402611534276387, "acc_ci": 0.0, "acc_runs": [0.5402611534276387], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011647846656062001, "centered_acc": 0.055495103373231824, "centered_acc_ci": 0.0, "centered_acc_runs": [0.055495103373231824], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2596801346801347, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2596801346801347], "acc": 0.25252525252525254, "acc_ci": 0.0, "acc_runs": [0.25252525252525254], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008996990428562134, "centered_acc": 0.01290684624017957, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01290684624017957], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26791808873720135, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26791808873720135], "acc": 0.21416382252559726, "acc_ci": 0.0, "acc_runs": [0.21416382252559726], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012942030195136482, "centered_acc": 0.023890784982935138, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023890784982935138], "ci": 0.0}, "openbookqa": {"acc_norm": 0.288, "acc_norm_ci": 0.0, "acc_norm_runs": [0.288], "acc": 0.184, "acc_ci": 0.0, "acc_runs": [0.184], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020271503835075255, "centered_acc": 0.05066666666666664, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05066666666666664], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48303078137332284, "acc_ci": 0.0, "acc_runs": [0.48303078137332284], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014044390401613118, "centered_acc": -0.033938437253354325, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.033938437253354325], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.62, "acc_ci": 0.0, "acc_runs": [0.62], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04878317312145634, "centered_acc": 0.24, "centered_acc_ci": 0.0, "centered_acc_runs": [0.24], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 2343983809.724925, "perplexity_ci": 0.0, "perplexity_runs": [2343983809.724925], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2964296666267325, "mean_acc_norm_ci": 0.0, "mean_acc": 0.296930153276871, "mean_acc_ci": 0.0, "mean_centered_acc": 0.0021376504184055633, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2964296666267325], "mean_acc_runs": [0.296930153276871], "mean_centered_acc_runs": [0.0021376504184055633], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8640", "checkpoint": "checkpoint-8640", "global_step": 8640, "tokens_seen": 9059696640, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2517426807408883, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2517426807408883], "acc": 0.2504481179047998, "acc_ci": 0.0, "acc_runs": [0.2504481179047998], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004331271717773648, "centered_acc": 0.0023235743211843887, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0023235743211843887], "ci": 0.0}, "piqa": {"acc_norm": 0.5130576713819369, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5130576713819369], "acc": 0.5364526659412405, "acc_ci": 0.0, "acc_runs": [0.5364526659412405], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011661845375886444, "centered_acc": 0.02611534276387384, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02611534276387384], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2718855218855219, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2718855218855219], "acc": 0.25462962962962965, "acc_ci": 0.0, "acc_runs": [0.25462962962962965], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009129795867310605, "centered_acc": 0.02918069584736253, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02918069584736253], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2627986348122867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2627986348122867], "acc": 0.20477815699658702, "acc_ci": 0.0, "acc_runs": [0.20477815699658702], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012862523175351397, "centered_acc": 0.017064846416382284, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017064846416382284], "ci": 0.0}, "openbookqa": {"acc_norm": 0.292, "acc_norm_ci": 0.0, "acc_norm_runs": [0.292], "acc": 0.184, "acc_ci": 0.0, "acc_runs": [0.184], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020354375480530096, "centered_acc": 0.05599999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05599999999999997], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4940805051302289, "acc_ci": 0.0, "acc_runs": [0.4940805051302289], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051500838485925, "centered_acc": -0.011838989739542227, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.011838989739542227], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 2731795470.477959, "perplexity_ci": 0.0, "perplexity_runs": [2731795470.477959], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.185, "acc_norm_ci": 0.0, "acc_norm_runs": [0.185], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012285191326386655, "centered_acc": -0.08666666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08666666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29608075147010565, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29396765373761863, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0034718081677783666, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29608075147010565], "mean_acc_runs": [0.29396765373761863], "mean_centered_acc_runs": [-0.0034718081677783666], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-8928", "checkpoint": "checkpoint-8928", "global_step": 8928, "tokens_seen": 9361686528, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25253933479386576, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25253933479386576], "acc": 0.2539334793865764, "acc_ci": 0.0, "acc_runs": [0.2539334793865764], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004335809614480365, "centered_acc": 0.00338577972515434, "centered_acc_ci": 0.0, "centered_acc_runs": [0.00338577972515434], "ci": 0.0}, "piqa": {"acc_norm": 0.5179542981501633, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5179542981501633], "acc": 0.5402611534276387, "acc_ci": 0.0, "acc_runs": [0.5402611534276387], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011658300623287121, "centered_acc": 0.0359085963003265, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0359085963003265], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2622053872053872, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2622053872053872], "acc": 0.24957912457912457, "acc_ci": 0.0, "acc_runs": [0.24957912457912457], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009025197991724937, "centered_acc": 0.01627384960718296, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01627384960718296], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26023890784982934, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26023890784982934], "acc": 0.2150170648464164, "acc_ci": 0.0, "acc_runs": [0.2150170648464164], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012821930225112672, "centered_acc": 0.013651877133105783, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013651877133105783], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.198, "acc_ci": 0.0, "acc_runs": [0.198], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.47513812154696133, "acc_ci": 0.0, "acc_runs": [0.47513812154696133], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01403510288362781, "centered_acc": -0.04972375690607733, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.04972375690607733], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.64, "acc_ci": 0.0, "acc_runs": [0.64], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.048241815132442176, "centered_acc": 0.28, "centered_acc_ci": 0.0, "centered_acc_runs": [0.28], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 2380239310.07631, "perplexity_ci": 0.0, "perplexity_runs": [2380239310.07631], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.04, "acc_ci": 0.0, "acc_runs": [0.04], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2936563213332076, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2990216405560418, "mean_acc_ci": 0.0, "mean_centered_acc": 0.0016375484526399498, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2936563213332076], "mean_acc_runs": [0.2990216405560418], "mean_centered_acc_runs": [0.0016375484526399498], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-9216", "checkpoint": "checkpoint-9216", "global_step": 9216, "tokens_seen": 9663676416, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.252141007767377, "acc_norm_ci": 0.0, "acc_norm_runs": [0.252141007767377], "acc": 0.2513443537143995, "acc_ci": 0.0, "acc_runs": [0.2513443537143995], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004333543083293202, "centered_acc": 0.0028546770231693643, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0028546770231693643], "ci": 0.0}, "piqa": {"acc_norm": 0.5212187159956474, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5212187159956474], "acc": 0.5348204570184983, "acc_ci": 0.0, "acc_runs": [0.5348204570184983], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011655314732288771, "centered_acc": 0.042437431991294794, "centered_acc_ci": 0.0, "centered_acc_runs": [0.042437431991294794], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26557239057239057, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26557239057239057], "acc": 0.2478956228956229, "acc_ci": 0.0, "acc_runs": [0.2478956228956229], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00906221062697191, "centered_acc": 0.020763187429854096, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020763187429854096], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2525597269624573, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2525597269624573], "acc": 0.20819112627986347, "acc_ci": 0.0, "acc_runs": [0.20819112627986347], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012696728980207687, "centered_acc": 0.003412969283276427, "centered_acc_ci": 0.0, "centered_acc_runs": [0.003412969283276427], "ci": 0.0}, "openbookqa": {"acc_norm": 0.278, "acc_norm_ci": 0.0, "acc_norm_runs": [0.278], "acc": 0.176, "acc_ci": 0.0, "acc_runs": [0.176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020055833888070865, "centered_acc": 0.037333333333333364, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037333333333333364], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4877663772691397, "acc_ci": 0.0, "acc_runs": [0.4877663772691397], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01404827882040573, "centered_acc": -0.024467245461720633, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.024467245461720633], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.61, "acc_ci": 0.0, "acc_runs": [0.61], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04902071300001973, "centered_acc": 0.21999999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.21999999999999997], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 3063566637.8900723, "perplexity_ci": 0.0, "perplexity_runs": [3063566637.8900723], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.18, "acc_norm_ci": 0.0, "acc_norm_runs": [0.18], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012155153135512024, "centered_acc": -0.09333333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09333333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2915819735496454, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29373053989512243, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0038248951318583684, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2915819735496454], "mean_acc_runs": [0.29373053989512243], "mean_centered_acc_runs": [-0.0038248951318583684], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-9504", "checkpoint": "checkpoint-9504", "global_step": 9504, "tokens_seen": 9965666304, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25114519020115517, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25114519020115517], "acc": 0.25184226249751046, "acc_ci": 0.0, "acc_runs": [0.25184226249751046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004327855588466433, "centered_acc": 0.0015269202682068883, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0015269202682068883], "ci": 0.0}, "piqa": {"acc_norm": 0.529379760609358, "acc_norm_ci": 0.0, "acc_norm_runs": [0.529379760609358], "acc": 0.5391730141458106, "acc_ci": 0.0, "acc_runs": [0.5391730141458106], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01164566756505079, "centered_acc": 0.05875952121871597, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05875952121871597], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2676767676767677, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2676767676767677], "acc": 0.25547138047138046, "acc_ci": 0.0, "acc_runs": [0.25547138047138046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009085000147099275, "centered_acc": 0.023569023569023573, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023569023569023573], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2721843003412969, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2721843003412969], "acc": 0.20051194539249148, "acc_ci": 0.0, "acc_runs": [0.20051194539249148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01300660040642361, "centered_acc": 0.029579067121729235, "centered_acc_ci": 0.0, "centered_acc_runs": [0.029579067121729235], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.18, "acc_ci": 0.0, "acc_runs": [0.18], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.49329123914759276, "acc_ci": 0.0, "acc_runs": [0.49329123914759276], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051220692330247, "centered_acc": -0.013417521704814472, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.013417521704814472], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.63, "acc_ci": 0.0, "acc_runs": [0.63], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.048523658709390974, "centered_acc": 0.26, "centered_acc_ci": 0.0, "centered_acc_runs": [0.26], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 3305910352.2299867, "perplexity_ci": 0.0, "perplexity_runs": [3305910352.2299867], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29756433647142966, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2973577303428486, "mean_acc_ci": 0.0, "mean_centered_acc": 0.007769474150399461, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29756433647142966], "mean_acc_runs": [0.2973577303428486], "mean_centered_acc_runs": [0.007769474150399461], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-9792", "checkpoint": "checkpoint-9792", "global_step": 9792, "tokens_seen": 10267656192, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2533359888468433, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2533359888468433], "acc": 0.2524397530372436, "acc_ci": 0.0, "acc_runs": [0.2524397530372436], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004340328204135388, "centered_acc": 0.0044479851291243655, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0044479851291243655], "ci": 0.0}, "piqa": {"acc_norm": 0.5315560391730142, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5315560391730142], "acc": 0.5310119695321001, "acc_ci": 0.0, "acc_runs": [0.5310119695321001], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011642567710878958, "centered_acc": 0.06311207834602839, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06311207834602839], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26515151515151514, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26515151515151514], "acc": 0.24621212121212122, "acc_ci": 0.0, "acc_runs": [0.24621212121212122], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009057621139172679, "centered_acc": 0.020202020202020183, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020202020202020183], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26706484641638223, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26706484641638223], "acc": 0.1962457337883959, "acc_ci": 0.0, "acc_runs": [0.1962457337883959], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01292893319649643, "centered_acc": 0.022753128555176305, "centered_acc_ci": 0.0, "centered_acc_runs": [0.022753128555176305], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.192, "acc_ci": 0.0, "acc_runs": [0.192], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48539857932123126, "acc_ci": 0.0, "acc_runs": [0.48539857932123126], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014046492383275955, "centered_acc": -0.02920284135753748, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.02920284135753748], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.66, "acc_ci": 0.0, "acc_runs": [0.66], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04760952285695234, "centered_acc": 0.32000000000000006, "centered_acc_ci": 0.0, "centered_acc_runs": [0.32000000000000006], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 3783817234.3022566, "perplexity_ci": 0.0, "perplexity_runs": [3783817234.3022566], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.184, "acc_norm_ci": 0.0, "acc_norm_runs": [0.184], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01225945734093864, "centered_acc": -0.08800000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08800000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2978513982646258, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29875956186647923, "mean_acc_ci": 0.0, "mean_centered_acc": 0.013098588269134714, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2978513982646258], "mean_acc_runs": [0.29875956186647923], "mean_centered_acc_runs": [0.013098588269134714], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-10080", "checkpoint": "checkpoint-10080", "global_step": 10080, "tokens_seen": 10569646080, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2538338976299542, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2538338976299542], "acc": 0.2533359888468433, "acc_ci": 0.0, "acc_runs": [0.2533359888468433], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004343142545094425, "centered_acc": 0.005111863506605567, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005111863506605567], "ci": 0.0}, "piqa": {"acc_norm": 0.5250272034820457, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5250272034820457], "acc": 0.5397170837867247, "acc_ci": 0.0, "acc_runs": [0.5397170837867247], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011651200967623562, "centered_acc": 0.05005440696409136, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05005440696409136], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2689393939393939, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2689393939393939], "acc": 0.24831649831649832, "acc_ci": 0.0, "acc_runs": [0.24831649831649832], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009098548093009251, "centered_acc": 0.02525252525252523, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02525252525252523], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.257679180887372, "acc_norm_ci": 0.0, "acc_norm_runs": [0.257679180887372], "acc": 0.20051194539249148, "acc_ci": 0.0, "acc_runs": [0.20051194539249148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012780770562768334, "centered_acc": 0.010238907849829356, "centered_acc_ci": 0.0, "centered_acc_runs": [0.010238907849829356], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.182, "acc_ci": 0.0, "acc_runs": [0.182], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4988161010260458, "acc_ci": 0.0, "acc_runs": [0.4988161010260458], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052446290528871, "centered_acc": -0.0023677979479084232, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0023677979479084232], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 5418371311.711865, "perplexity_ci": 0.0, "perplexity_runs": [5418371311.711865], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2934132793231276, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2926985079142304, "mean_acc_ci": 0.0, "mean_centered_acc": -0.006126130091939594, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2934132793231276], "mean_acc_runs": [0.2926985079142304], "mean_centered_acc_runs": [-0.006126130091939594], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask15-fulle-lm", "run": "mask15-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask15-fulle-lm/checkpoint-10368", "checkpoint": "checkpoint-10368", "global_step": 10368, "tokens_seen": 10871635968, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25184226249751046, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25184226249751046], "acc": 0.2533359888468433, "acc_ci": 0.0, "acc_runs": [0.2533359888468433], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004331840012787794, "centered_acc": 0.0024563499966806144, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0024563499966806144], "ci": 0.0}, "piqa": {"acc_norm": 0.529923830250272, "acc_norm_ci": 0.0, "acc_norm_runs": [0.529923830250272], "acc": 0.5386289445048966, "acc_ci": 0.0, "acc_runs": [0.5386289445048966], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011644913435420134, "centered_acc": 0.05984766050054402, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05984766050054402], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26430976430976433, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26430976430976433], "acc": 0.2537878787878788, "acc_ci": 0.0, "acc_runs": [0.2537878787878788], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009048410451862887, "centered_acc": 0.019079685746352437, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019079685746352437], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2525597269624573, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2525597269624573], "acc": 0.197098976109215, "acc_ci": 0.0, "acc_runs": [0.197098976109215], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012696728980207687, "centered_acc": 0.003412969283276427, "centered_acc_ci": 0.0, "centered_acc_runs": [0.003412969283276427], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.178, "acc_ci": 0.0, "acc_runs": [0.178], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4996053670086819, "acc_ci": 0.0, "acc_runs": [0.4996053670086819], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405248130604961, "centered_acc": -0.0007892659826361781, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0007892659826361781], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.64, "acc_ci": 0.0, "acc_runs": [0.64], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.048241815132442176, "centered_acc": 0.28, "centered_acc_ci": 0.0, "centered_acc_runs": [0.28], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 9574311667.190289, "perplexity_ci": 0.0, "perplexity_runs": [9574311667.190289], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2946059306700007, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29857446170312163, "mean_acc_ci": 0.0, "mean_centered_acc": 0.008805443306476068, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2946059306700007], "mean_acc_runs": [0.29857446170312163], "mean_centered_acc_runs": [0.008805443306476068], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-1152.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-1152.json new file mode 100644 index 0000000000000000000000000000000000000000..3c2090d7cd75aa544ccb45e14ea5c91924f3be7a --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-1152.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-1152", + "checkpoint": "checkpoint-1152", + "global_step": 1152, + "eval_loss": 4.284478664398193, + "metrics": { + "eval_loss": 4.284478664398193, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.892, + "eval_samples_per_second": 18.8, + "eval_steps_per_second": 2.356, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-1440.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-1440.json new file mode 100644 index 0000000000000000000000000000000000000000..d8f7801afe972d4e0932784ba3d0ad0947201a06 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-1440.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-1440", + "checkpoint": "checkpoint-1440", + "global_step": 1440, + "eval_loss": 4.102666854858398, + "metrics": { + "eval_loss": 4.102666854858398, + "eval_model_preparation_time": 0.0042, + "eval_runtime": 129.9887, + "eval_samples_per_second": 18.786, + "eval_steps_per_second": 2.354, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-1728.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-1728.json new file mode 100644 index 0000000000000000000000000000000000000000..4e4f0a9e8e8f6942cc0f9bd614649d125fc5eb43 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-1728.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-1728", + "checkpoint": "checkpoint-1728", + "global_step": 1728, + "eval_loss": 4.01711893081665, + "metrics": { + "eval_loss": 4.01711893081665, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.8601, + "eval_samples_per_second": 18.805, + "eval_steps_per_second": 2.356, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-2016.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-2016.json new file mode 100644 index 0000000000000000000000000000000000000000..af9394bc547c851505e353a6711fdf6a7abdd937 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-2016.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2016", + "checkpoint": "checkpoint-2016", + "global_step": 2016, + "eval_loss": 3.963214874267578, + "metrics": { + "eval_loss": 3.963214874267578, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.8817, + "eval_samples_per_second": 18.802, + "eval_steps_per_second": 2.356, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-2304.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-2304.json new file mode 100644 index 0000000000000000000000000000000000000000..0e0dadd50e746f78abbe069249640dde8cafa835 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-2304.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2304", + "checkpoint": "checkpoint-2304", + "global_step": 2304, + "eval_loss": 3.9317541122436523, + "metrics": { + "eval_loss": 3.9317541122436523, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.962, + "eval_samples_per_second": 18.79, + "eval_steps_per_second": 2.355, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-2592.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-2592.json new file mode 100644 index 0000000000000000000000000000000000000000..3603ccf74a0cda32246a09d432344ef510a57f91 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-2592.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2592", + "checkpoint": "checkpoint-2592", + "global_step": 2592, + "eval_loss": 3.9291980266571045, + "metrics": { + "eval_loss": 3.9291980266571045, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 130.0113, + "eval_samples_per_second": 18.783, + "eval_steps_per_second": 2.354, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-288.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-288.json new file mode 100644 index 0000000000000000000000000000000000000000..c4f217572f7436d968fb38c8d1a9a96736cbdb41 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-288.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-288", + "checkpoint": "checkpoint-288", + "global_step": 288, + "eval_loss": 6.716034889221191, + "metrics": { + "eval_loss": 6.716034889221191, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.1607, + "eval_samples_per_second": 18.907, + "eval_steps_per_second": 2.369, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-2880.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-2880.json new file mode 100644 index 0000000000000000000000000000000000000000..a32dac746793d6e4afdccbc834d937ab3f6a3762 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-2880.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2880", + "checkpoint": "checkpoint-2880", + "global_step": 2880, + "eval_loss": 3.9233133792877197, + "metrics": { + "eval_loss": 3.9233133792877197, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.9926, + "eval_samples_per_second": 18.786, + "eval_steps_per_second": 2.354, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-3168.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-3168.json new file mode 100644 index 0000000000000000000000000000000000000000..a0f0518daaade48ffb7fb710ae0838d5caca44e5 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-3168.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-3168", + "checkpoint": "checkpoint-3168", + "global_step": 3168, + "eval_loss": 3.9287912845611572, + "metrics": { + "eval_loss": 3.9287912845611572, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.9712, + "eval_samples_per_second": 18.789, + "eval_steps_per_second": 2.354, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-3456.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-3456.json new file mode 100644 index 0000000000000000000000000000000000000000..d2bbc6b54233772c03c12dd9c8ca710cc6a69913 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-3456.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-3456", + "checkpoint": "checkpoint-3456", + "global_step": 3456, + "eval_loss": 3.9334733486175537, + "metrics": { + "eval_loss": 3.9334733486175537, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.6496, + "eval_samples_per_second": 18.835, + "eval_steps_per_second": 2.36, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-3744.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-3744.json new file mode 100644 index 0000000000000000000000000000000000000000..ac0ec2d87c2a285f924c015f68e54d669bf699d8 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-3744.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-3744", + "checkpoint": "checkpoint-3744", + "global_step": 3744, + "eval_loss": 3.947754383087158, + "metrics": { + "eval_loss": 3.947754383087158, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.34, + "eval_samples_per_second": 18.88, + "eval_steps_per_second": 2.366, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-4032.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-4032.json new file mode 100644 index 0000000000000000000000000000000000000000..36063b1040bf24149e5b1c9fb3aaea200c507aa1 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-4032.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4032", + "checkpoint": "checkpoint-4032", + "global_step": 4032, + "eval_loss": 3.9591052532196045, + "metrics": { + "eval_loss": 3.9591052532196045, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.0689, + "eval_samples_per_second": 18.92, + "eval_steps_per_second": 2.371, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-4320.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-4320.json new file mode 100644 index 0000000000000000000000000000000000000000..d461b8c890c81a1816038461bbee5da97c2eb24e --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-4320.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4320", + "checkpoint": "checkpoint-4320", + "global_step": 4320, + "eval_loss": 3.97983455657959, + "metrics": { + "eval_loss": 3.97983455657959, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.1071, + "eval_samples_per_second": 18.915, + "eval_steps_per_second": 2.37, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-4608.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-4608.json new file mode 100644 index 0000000000000000000000000000000000000000..a530a8a5eb630b032bbe07c6fce2194bb5a7d752 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-4608.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4608", + "checkpoint": "checkpoint-4608", + "global_step": 4608, + "eval_loss": 3.9967243671417236, + "metrics": { + "eval_loss": 3.9967243671417236, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.5864, + "eval_samples_per_second": 18.845, + "eval_steps_per_second": 2.361, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-4896.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-4896.json new file mode 100644 index 0000000000000000000000000000000000000000..567c9e02161f3e3e0467edb9461be1d55e1885c3 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-4896.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4896", + "checkpoint": "checkpoint-4896", + "global_step": 4896, + "eval_loss": 4.008735656738281, + "metrics": { + "eval_loss": 4.008735656738281, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.6687, + "eval_samples_per_second": 18.833, + "eval_steps_per_second": 2.36, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-5184.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-5184.json new file mode 100644 index 0000000000000000000000000000000000000000..100b248cd3f35b3ce933276a8221c7773a4c3f37 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-5184.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-5184", + "checkpoint": "checkpoint-5184", + "global_step": 5184, + "eval_loss": 4.030752658843994, + "metrics": { + "eval_loss": 4.030752658843994, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.7888, + "eval_samples_per_second": 18.815, + "eval_steps_per_second": 2.358, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-5472.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-5472.json new file mode 100644 index 0000000000000000000000000000000000000000..c7fd14934461a93308053c32e8072478c121d272 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-5472.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-5472", + "checkpoint": "checkpoint-5472", + "global_step": 5472, + "eval_loss": 4.042012691497803, + "metrics": { + "eval_loss": 4.042012691497803, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.8586, + "eval_samples_per_second": 18.805, + "eval_steps_per_second": 2.356, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-576.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-576.json new file mode 100644 index 0000000000000000000000000000000000000000..42126e640292fdb7b2623ff33a1264688c54bad0 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-576.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-576", + "checkpoint": "checkpoint-576", + "global_step": 576, + "eval_loss": 5.710719108581543, + "metrics": { + "eval_loss": 5.710719108581543, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.6921, + "eval_samples_per_second": 18.829, + "eval_steps_per_second": 2.359, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-5760.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-5760.json new file mode 100644 index 0000000000000000000000000000000000000000..9ecea32c407811ff44364283c02559b109cf712f --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-5760.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-5760", + "checkpoint": "checkpoint-5760", + "global_step": 5760, + "eval_loss": 4.072123050689697, + "metrics": { + "eval_loss": 4.072123050689697, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.8226, + "eval_samples_per_second": 18.81, + "eval_steps_per_second": 2.357, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-6048.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-6048.json new file mode 100644 index 0000000000000000000000000000000000000000..fbcc849a1a79097fe641b53dcd77a28071e6acc6 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-6048.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6048", + "checkpoint": "checkpoint-6048", + "global_step": 6048, + "eval_loss": 4.091675758361816, + "metrics": { + "eval_loss": 4.091675758361816, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.8757, + "eval_samples_per_second": 18.803, + "eval_steps_per_second": 2.356, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-6336.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-6336.json new file mode 100644 index 0000000000000000000000000000000000000000..a0c1ebe150b108863067582cfb220d5d59416787 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-6336.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6336", + "checkpoint": "checkpoint-6336", + "global_step": 6336, + "eval_loss": 4.101736068725586, + "metrics": { + "eval_loss": 4.101736068725586, + "eval_model_preparation_time": 0.0033, + "eval_runtime": 129.9359, + "eval_samples_per_second": 18.794, + "eval_steps_per_second": 2.355, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-6624.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-6624.json new file mode 100644 index 0000000000000000000000000000000000000000..49d02f329566f8755dceae7447049f9f50a4cd1a --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-6624.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6624", + "checkpoint": "checkpoint-6624", + "global_step": 6624, + "eval_loss": 4.129087924957275, + "metrics": { + "eval_loss": 4.129087924957275, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.9152, + "eval_samples_per_second": 18.797, + "eval_steps_per_second": 2.355, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-6912.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-6912.json new file mode 100644 index 0000000000000000000000000000000000000000..8141b1f0649ed5fe4f8326103978b0d18344a086 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-6912.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6912", + "checkpoint": "checkpoint-6912", + "global_step": 6912, + "eval_loss": 4.151818752288818, + "metrics": { + "eval_loss": 4.151818752288818, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.8337, + "eval_samples_per_second": 18.809, + "eval_steps_per_second": 2.357, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-72.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-72.json new file mode 100644 index 0000000000000000000000000000000000000000..9b40fe9afa5e1cefab7c7823e6d9c51ec672c3e2 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-72.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-72", + "checkpoint": "checkpoint-72", + "global_step": 72, + "eval_loss": 8.227867126464844, + "metrics": { + "eval_loss": 8.227867126464844, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 127.0065, + "eval_samples_per_second": 19.227, + "eval_steps_per_second": 2.409, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-7200.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-7200.json new file mode 100644 index 0000000000000000000000000000000000000000..22430f7c66749bd7ddaea0785ce359b7faf044d5 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-7200.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-7200", + "checkpoint": "checkpoint-7200", + "global_step": 7200, + "eval_loss": 4.161838531494141, + "metrics": { + "eval_loss": 4.161838531494141, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 130.0247, + "eval_samples_per_second": 18.781, + "eval_steps_per_second": 2.353, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-7488.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-7488.json new file mode 100644 index 0000000000000000000000000000000000000000..73c7f0b0beb3810deed65eca7862b332a2c8134c --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-7488.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-7488", + "checkpoint": "checkpoint-7488", + "global_step": 7488, + "eval_loss": 4.176450729370117, + "metrics": { + "eval_loss": 4.176450729370117, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 130.0744, + "eval_samples_per_second": 18.774, + "eval_steps_per_second": 2.353, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-7776.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-7776.json new file mode 100644 index 0000000000000000000000000000000000000000..f358e24c96ea048b310d08f055a36cf2a316132b --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-7776.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-7776", + "checkpoint": "checkpoint-7776", + "global_step": 7776, + "eval_loss": 4.197800636291504, + "metrics": { + "eval_loss": 4.197800636291504, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 130.0869, + "eval_samples_per_second": 18.772, + "eval_steps_per_second": 2.352, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-8064.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-8064.json new file mode 100644 index 0000000000000000000000000000000000000000..361127da2c51001dfb04cf55558fe1ffc2d93652 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-8064.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-8064", + "checkpoint": "checkpoint-8064", + "global_step": 8064, + "eval_loss": 4.205382347106934, + "metrics": { + "eval_loss": 4.205382347106934, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.8667, + "eval_samples_per_second": 18.804, + "eval_steps_per_second": 2.356, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-8352.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-8352.json new file mode 100644 index 0000000000000000000000000000000000000000..0c07c0e2c62640b9b8f69767c1da3513619e6690 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-8352.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-8352", + "checkpoint": "checkpoint-8352", + "global_step": 8352, + "eval_loss": 4.2360711097717285, + "metrics": { + "eval_loss": 4.2360711097717285, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.838, + "eval_samples_per_second": 18.808, + "eval_steps_per_second": 2.357, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-864.json b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-864.json new file mode 100644 index 0000000000000000000000000000000000000000..59ae3e3c311b0c4def512a1636089adc106b4bf4 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/checkpoint-864.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-864", + "checkpoint": "checkpoint-864", + "global_step": 864, + "eval_loss": 4.695986747741699, + "metrics": { + "eval_loss": 4.695986747741699, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.9428, + "eval_samples_per_second": 18.793, + "eval_steps_per_second": 2.355, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/mask30-fulle-lm/l2r_n1_eval/summary.jsonl b/results/mask30-fulle-lm/l2r_n1_eval/summary.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..4b6fd5ef105a8887b661fd2674d9d93b9270f8f2 --- /dev/null +++ b/results/mask30-fulle-lm/l2r_n1_eval/summary.jsonl @@ -0,0 +1,30 @@ +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-72", "checkpoint": "checkpoint-72", "global_step": 72, "eval_loss": 8.227867126464844, "metrics": {"eval_loss": 8.227867126464844, "eval_model_preparation_time": 0.0036, "eval_runtime": 127.0065, "eval_samples_per_second": 19.227, "eval_steps_per_second": 2.409, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-288", "checkpoint": "checkpoint-288", "global_step": 288, "eval_loss": 6.716034889221191, "metrics": {"eval_loss": 6.716034889221191, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.1607, "eval_samples_per_second": 18.907, "eval_steps_per_second": 2.369, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-576", "checkpoint": "checkpoint-576", "global_step": 576, "eval_loss": 5.710719108581543, "metrics": {"eval_loss": 5.710719108581543, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.6921, "eval_samples_per_second": 18.829, "eval_steps_per_second": 2.359, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-864", "checkpoint": "checkpoint-864", "global_step": 864, "eval_loss": 4.695986747741699, "metrics": {"eval_loss": 4.695986747741699, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.9428, "eval_samples_per_second": 18.793, "eval_steps_per_second": 2.355, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-1152", "checkpoint": "checkpoint-1152", "global_step": 1152, "eval_loss": 4.284478664398193, "metrics": {"eval_loss": 4.284478664398193, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.892, "eval_samples_per_second": 18.8, "eval_steps_per_second": 2.356, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-1440", "checkpoint": "checkpoint-1440", "global_step": 1440, "eval_loss": 4.102666854858398, "metrics": {"eval_loss": 4.102666854858398, "eval_model_preparation_time": 0.0042, "eval_runtime": 129.9887, "eval_samples_per_second": 18.786, "eval_steps_per_second": 2.354, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-1728", "checkpoint": "checkpoint-1728", "global_step": 1728, "eval_loss": 4.01711893081665, "metrics": {"eval_loss": 4.01711893081665, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.8601, "eval_samples_per_second": 18.805, "eval_steps_per_second": 2.356, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2016", "checkpoint": "checkpoint-2016", "global_step": 2016, "eval_loss": 3.963214874267578, "metrics": {"eval_loss": 3.963214874267578, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.8817, "eval_samples_per_second": 18.802, "eval_steps_per_second": 2.356, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2304", "checkpoint": "checkpoint-2304", "global_step": 2304, "eval_loss": 3.9317541122436523, "metrics": {"eval_loss": 3.9317541122436523, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.962, "eval_samples_per_second": 18.79, "eval_steps_per_second": 2.355, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2592", "checkpoint": "checkpoint-2592", "global_step": 2592, "eval_loss": 3.9291980266571045, "metrics": {"eval_loss": 3.9291980266571045, "eval_model_preparation_time": 0.0033, "eval_runtime": 130.0113, "eval_samples_per_second": 18.783, "eval_steps_per_second": 2.354, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2880", "checkpoint": "checkpoint-2880", "global_step": 2880, "eval_loss": 3.9233133792877197, "metrics": {"eval_loss": 3.9233133792877197, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.9926, "eval_samples_per_second": 18.786, "eval_steps_per_second": 2.354, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-3168", "checkpoint": "checkpoint-3168", "global_step": 3168, "eval_loss": 3.9287912845611572, "metrics": {"eval_loss": 3.9287912845611572, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.9712, "eval_samples_per_second": 18.789, "eval_steps_per_second": 2.354, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-3456", "checkpoint": "checkpoint-3456", "global_step": 3456, "eval_loss": 3.9334733486175537, "metrics": {"eval_loss": 3.9334733486175537, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.6496, "eval_samples_per_second": 18.835, "eval_steps_per_second": 2.36, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-3744", "checkpoint": "checkpoint-3744", "global_step": 3744, "eval_loss": 3.947754383087158, "metrics": {"eval_loss": 3.947754383087158, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.34, "eval_samples_per_second": 18.88, "eval_steps_per_second": 2.366, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4032", "checkpoint": "checkpoint-4032", "global_step": 4032, "eval_loss": 3.9591052532196045, "metrics": {"eval_loss": 3.9591052532196045, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.0689, "eval_samples_per_second": 18.92, "eval_steps_per_second": 2.371, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4320", "checkpoint": "checkpoint-4320", "global_step": 4320, "eval_loss": 3.97983455657959, "metrics": {"eval_loss": 3.97983455657959, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.1071, "eval_samples_per_second": 18.915, "eval_steps_per_second": 2.37, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4608", "checkpoint": "checkpoint-4608", "global_step": 4608, "eval_loss": 3.9967243671417236, "metrics": {"eval_loss": 3.9967243671417236, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.5864, "eval_samples_per_second": 18.845, "eval_steps_per_second": 2.361, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4896", "checkpoint": "checkpoint-4896", "global_step": 4896, "eval_loss": 4.008735656738281, "metrics": {"eval_loss": 4.008735656738281, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.6687, "eval_samples_per_second": 18.833, "eval_steps_per_second": 2.36, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-5184", "checkpoint": "checkpoint-5184", "global_step": 5184, "eval_loss": 4.030752658843994, "metrics": {"eval_loss": 4.030752658843994, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.7888, "eval_samples_per_second": 18.815, "eval_steps_per_second": 2.358, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-5472", "checkpoint": "checkpoint-5472", "global_step": 5472, "eval_loss": 4.042012691497803, "metrics": {"eval_loss": 4.042012691497803, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.8586, "eval_samples_per_second": 18.805, "eval_steps_per_second": 2.356, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-5760", "checkpoint": "checkpoint-5760", "global_step": 5760, "eval_loss": 4.072123050689697, "metrics": {"eval_loss": 4.072123050689697, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.8226, "eval_samples_per_second": 18.81, "eval_steps_per_second": 2.357, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6048", "checkpoint": "checkpoint-6048", "global_step": 6048, "eval_loss": 4.091675758361816, "metrics": {"eval_loss": 4.091675758361816, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.8757, "eval_samples_per_second": 18.803, "eval_steps_per_second": 2.356, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6336", "checkpoint": "checkpoint-6336", "global_step": 6336, "eval_loss": 4.101736068725586, "metrics": {"eval_loss": 4.101736068725586, "eval_model_preparation_time": 0.0033, "eval_runtime": 129.9359, "eval_samples_per_second": 18.794, "eval_steps_per_second": 2.355, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6624", "checkpoint": "checkpoint-6624", "global_step": 6624, "eval_loss": 4.129087924957275, "metrics": {"eval_loss": 4.129087924957275, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.9152, "eval_samples_per_second": 18.797, "eval_steps_per_second": 2.355, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6912", "checkpoint": "checkpoint-6912", "global_step": 6912, "eval_loss": 4.151818752288818, "metrics": {"eval_loss": 4.151818752288818, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.8337, "eval_samples_per_second": 18.809, "eval_steps_per_second": 2.357, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-7200", "checkpoint": "checkpoint-7200", "global_step": 7200, "eval_loss": 4.161838531494141, "metrics": {"eval_loss": 4.161838531494141, "eval_model_preparation_time": 0.0036, "eval_runtime": 130.0247, "eval_samples_per_second": 18.781, "eval_steps_per_second": 2.353, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-7488", "checkpoint": "checkpoint-7488", "global_step": 7488, "eval_loss": 4.176450729370117, "metrics": {"eval_loss": 4.176450729370117, "eval_model_preparation_time": 0.0035, "eval_runtime": 130.0744, "eval_samples_per_second": 18.774, "eval_steps_per_second": 2.353, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-7776", "checkpoint": "checkpoint-7776", "global_step": 7776, "eval_loss": 4.197800636291504, "metrics": {"eval_loss": 4.197800636291504, "eval_model_preparation_time": 0.0034, "eval_runtime": 130.0869, "eval_samples_per_second": 18.772, "eval_steps_per_second": 2.352, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-8064", "checkpoint": "checkpoint-8064", "global_step": 8064, "eval_loss": 4.205382347106934, "metrics": {"eval_loss": 4.205382347106934, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.8667, "eval_samples_per_second": 18.804, "eval_steps_per_second": 2.356, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-8352", "checkpoint": "checkpoint-8352", "global_step": 8352, "eval_loss": 4.2360711097717285, "metrics": {"eval_loss": 4.2360711097717285, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.838, "eval_samples_per_second": 18.808, "eval_steps_per_second": 2.357, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-1152.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-1152.json new file mode 100644 index 0000000000000000000000000000000000000000..877b0b2ed258e06b8c7a9e5299f191f62946641a --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-1152.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-1152", + "checkpoint": "checkpoint-1152", + "global_step": 1152, + "tokens_seen": 1207959552, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2593108942441745, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2593108942441745 + ], + "acc": 0.2560246962756423, + "acc_ci": 0.0, + "acc_runs": [ + 0.2560246962756423 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043736082125607105, + "centered_acc": 0.012414525658899297, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012414525658899297 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5179542981501633, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5179542981501633 + ], + "acc": 0.5544069640914037, + "acc_ci": 0.0, + "acc_runs": [ + 0.5544069640914037 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011658300623287121, + "centered_acc": 0.0359085963003265, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0359085963003265 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.27104377104377103, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27104377104377103 + ], + "acc": 0.26515151515151514, + "acc_ci": 0.0, + "acc_runs": [ + 0.26515151515151514 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009120919741760646, + "centered_acc": 0.02805836139169471, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02805836139169471 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26535836177474403, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26535836177474403 + ], + "acc": 0.2235494880546075, + "acc_ci": 0.0, + "acc_runs": [ + 0.2235494880546075 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012902554762313832, + "centered_acc": 0.020477815699658713, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020477815699658713 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.298, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.298 + ], + "acc": 0.17, + "acc_ci": 0.0, + "acc_runs": [ + 0.17 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020475118092988954, + "centered_acc": 0.06399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06399999999999999 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5011838989739542, + "acc_ci": 0.0, + "acc_runs": [ + 0.5011838989739542 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052446290528871, + "centered_acc": 0.0023677979479084232, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0023677979479084232 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 2982265.0370979873, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 2982265.0370979873 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.18, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.18 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012155153135512024, + "centered_acc": -0.09333333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09333333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2986112208688088, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2973604024320823, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0015034791986050173, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2986112208688088 + ], + "mean_acc_runs": [ + 0.2973604024320823 + ], + "mean_centered_acc_runs": [ + -0.0015034791986050173 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-1440.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-1440.json new file mode 100644 index 0000000000000000000000000000000000000000..12570d6dc2b10b2775d3ca0b70a58ab7ac254a21 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-1440.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-1440", + "checkpoint": "checkpoint-1440", + "global_step": 1440, + "tokens_seen": 1509949440, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26180043815972914, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26180043815972914 + ], + "acc": 0.25731925911173076, + "acc_ci": 0.0, + "acc_runs": [ + 0.25731925911173076 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004387161203088004, + "centered_acc": 0.015733917546305525, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015733917546305525 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5282916213275299, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5282916213275299 + ], + "acc": 0.5571273122959739, + "acc_ci": 0.0, + "acc_runs": [ + 0.5571273122959739 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0116471341727495, + "centered_acc": 0.05658324265505987, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05658324265505987 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2748316498316498, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2748316498316498 + ], + "acc": 0.2668350168350168, + "acc_ci": 0.0, + "acc_runs": [ + 0.2668350168350168 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009160538115254864, + "centered_acc": 0.03310886644219976, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03310886644219976 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.23720136518771331, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.23720136518771331 + ], + "acc": 0.21160409556313994, + "acc_ci": 0.0, + "acc_runs": [ + 0.21160409556313994 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01243039982926071, + "centered_acc": -0.017064846416382246, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.017064846416382246 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.288, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.288 + ], + "acc": 0.16, + "acc_ci": 0.0, + "acc_runs": [ + 0.16 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020271503835075255, + "centered_acc": 0.05066666666666664, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05066666666666664 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4996053670086819, + "acc_ci": 0.0, + "acc_runs": [ + 0.4996053670086819 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405248130604961, + "centered_acc": -0.0007892659826361781, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0007892659826361781 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 5011021.250640561, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 5011021.250640561 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.051, + "acc_ci": 0.0, + "acc_runs": [ + 0.051 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29552084575110377, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2951778512588244, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.006057758763857714, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29552084575110377 + ], + "mean_acc_runs": [ + 0.2951778512588244 + ], + "mean_centered_acc_runs": [ + -0.006057758763857714 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-1728.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-1728.json new file mode 100644 index 0000000000000000000000000000000000000000..b5604864672339ea61186f0aae70219692224ebb --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-1728.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-1728", + "checkpoint": "checkpoint-1728", + "global_step": 1728, + "tokens_seen": 1811939328, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25721967735510853, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25721967735510853 + ], + "acc": 0.2546305516829317, + "acc_ci": 0.0, + "acc_runs": [ + 0.2546305516829317 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004362081806560604, + "centered_acc": 0.009626236473478045, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009626236473478045 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5337323177366703, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5337323177366703 + ], + "acc": 0.5489662676822633, + "acc_ci": 0.0, + "acc_runs": [ + 0.5489662676822633 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01163924552241405, + "centered_acc": 0.06746463547334058, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06746463547334058 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2777777777777778, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2777777777777778 + ], + "acc": 0.2638888888888889, + "acc_ci": 0.0, + "acc_runs": [ + 0.2638888888888889 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009190779909650122, + "centered_acc": 0.037037037037037056, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037037037037037056 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25341296928327645, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25341296928327645 + ], + "acc": 0.21245733788395904, + "acc_ci": 0.0, + "acc_runs": [ + 0.21245733788395904 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012710896778378677, + "centered_acc": 0.004550625711035261, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.004550625711035261 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.292, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.292 + ], + "acc": 0.164, + "acc_ci": 0.0, + "acc_runs": [ + 0.164 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020354375480530096, + "centered_acc": 0.05599999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05599999999999997 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5122336227308603, + "acc_ci": 0.0, + "acc_runs": [ + 0.5122336227308603 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014048278820405729, + "centered_acc": 0.02446724546172052, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02446724546172052 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.56, + "acc_ci": 0.0, + "acc_runs": [ + 0.56 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049888765156985884, + "centered_acc": 0.1200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1200000000000001 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 4497233.641430045, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 4497233.641430045 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.189, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.189 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01238678458811779, + "centered_acc": -0.08133333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08133333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.30052379035880555, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29394641306426034, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0006236255143690161, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.30052379035880555 + ], + "mean_acc_runs": [ + 0.29394641306426034 + ], + "mean_centered_acc_runs": [ + -0.0006236255143690161 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-2016.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-2016.json new file mode 100644 index 0000000000000000000000000000000000000000..57c660c58c0c776fd0e61262ac242671cc109771 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-2016.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2016", + "checkpoint": "checkpoint-2016", + "global_step": 2016, + "tokens_seen": 2113929216, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2593108942441745, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2593108942441745 + ], + "acc": 0.255327623979287, + "acc_ci": 0.0, + "acc_runs": [ + 0.255327623979287 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043736082125607105, + "centered_acc": 0.012414525658899297, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012414525658899297 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5326441784548422, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5326441784548422 + ], + "acc": 0.5495103373231773, + "acc_ci": 0.0, + "acc_runs": [ + 0.5495103373231773 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011640934420031744, + "centered_acc": 0.06528835690968449, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06528835690968449 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.27146464646464646, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27146464646464646 + ], + "acc": 0.25673400673400676, + "acc_ci": 0.0, + "acc_runs": [ + 0.25673400673400676 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00912536297036068, + "centered_acc": 0.028619528619528618, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.028619528619528618 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2627986348122867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2627986348122867 + ], + "acc": 0.2175767918088737, + "acc_ci": 0.0, + "acc_runs": [ + 0.2175767918088737 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012862523175351397, + "centered_acc": 0.017064846416382284, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017064846416382284 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.298, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.298 + ], + "acc": 0.16, + "acc_ci": 0.0, + "acc_runs": [ + 0.16 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020475118092988954, + "centered_acc": 0.06399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06399999999999999 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5067087608524072, + "acc_ci": 0.0, + "acc_runs": [ + 0.5067087608524072 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051220692330247, + "centered_acc": 0.013417521704814472, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013417521704814472 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.49, + "acc_ci": 0.0, + "acc_runs": [ + 0.49 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05024183937956913, + "centered_acc": -0.020000000000000018, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.020000000000000018 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 6360126.962215008, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 6360126.962215008 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.30120305899599165, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2859144982471452, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.016883736719624842, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.30120305899599165 + ], + "mean_acc_runs": [ + 0.2859144982471452 + ], + "mean_centered_acc_runs": [ + -0.016883736719624842 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-2304.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-2304.json new file mode 100644 index 0000000000000000000000000000000000000000..9e1eaffe7daffd4ee69800e03d6597d39cdfbb01 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-2304.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2304", + "checkpoint": "checkpoint-2304", + "global_step": 2304, + "tokens_seen": 2415919104, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2597092212706632, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2597092212706632 + ], + "acc": 0.2560246962756423, + "acc_ci": 0.0, + "acc_runs": [ + 0.2560246962756423 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004375788991216637, + "centered_acc": 0.012945628360884273, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012945628360884273 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5184983677910773, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5184983677910773 + ], + "acc": 0.5538628944504896, + "acc_ci": 0.0, + "acc_runs": [ + 0.5538628944504896 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165783758381818, + "centered_acc": 0.03699673558215455, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03699673558215455 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.27735690235690236, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27735690235690236 + ], + "acc": 0.25925925925925924, + "acc_ci": 0.0, + "acc_runs": [ + 0.25925925925925924 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009186490105111987, + "centered_acc": 0.03647586980920314, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03647586980920314 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.27047781569965873, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27047781569965873 + ], + "acc": 0.21416382252559726, + "acc_ci": 0.0, + "acc_runs": [ + 0.21416382252559726 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012980954547659553, + "centered_acc": 0.02730375426621164, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02730375426621164 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.500394632991318, + "acc_ci": 0.0, + "acc_runs": [ + 0.500394632991318 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405248130604961, + "centered_acc": 0.0007892659826360671, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0007892659826360671 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5, + "acc_ci": 0.0, + "acc_runs": [ + 0.5 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.050251890762960605, + "centered_acc": 0.0, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 5577220.558768843, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 5577220.558768843 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.186, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.186 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012310790208412926, + "centered_acc": -0.08533333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08533333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2990070511863836, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2862992767276007, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.01906449879090849, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2990070511863836 + ], + "mean_acc_runs": [ + 0.2862992767276007 + ], + "mean_centered_acc_runs": [ + -0.01906449879090849 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-2592.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-2592.json new file mode 100644 index 0000000000000000000000000000000000000000..c730fd958c4a43a74f68232d7641de912bf93f0e --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-2592.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2592", + "checkpoint": "checkpoint-2592", + "global_step": 2592, + "tokens_seen": 2717908992, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2613025293766182, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2613025293766182 + ], + "acc": 0.25473013343955386, + "acc_ci": 0.0, + "acc_runs": [ + 0.25473013343955386 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043844652190712236, + "centered_acc": 0.015070039168824248, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015070039168824248 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5228509249183896, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5228509249183896 + ], + "acc": 0.5462459194776932, + "acc_ci": 0.0, + "acc_runs": [ + 0.5462459194776932 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011653634832401168, + "centered_acc": 0.04570184983677916, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04570184983677916 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2748316498316498, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2748316498316498 + ], + "acc": 0.2537878787878788, + "acc_ci": 0.0, + "acc_runs": [ + 0.2537878787878788 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009160538115254864, + "centered_acc": 0.03310886644219976, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03310886644219976 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2627986348122867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2627986348122867 + ], + "acc": 0.2098976109215017, + "acc_ci": 0.0, + "acc_runs": [ + 0.2098976109215017 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012862523175351397, + "centered_acc": 0.017064846416382284, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017064846416382284 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.292, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.292 + ], + "acc": 0.168, + "acc_ci": 0.0, + "acc_runs": [ + 0.168 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020354375480530096, + "centered_acc": 0.05599999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05599999999999997 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4972375690607735, + "acc_ci": 0.0, + "acc_runs": [ + 0.4972375690607735 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052271211616401, + "centered_acc": -0.0055248618784530246, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0055248618784530246 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.52, + "acc_ci": 0.0, + "acc_runs": [ + 0.52 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05021167315686783, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 7382693.344635282, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 7382693.344635282 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.191 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012436787112179482, + "centered_acc": -0.07866666666666666, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07866666666666666 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.3007972898231574, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2873186573461101, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.013407889237059288, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.3007972898231574 + ], + "mean_acc_runs": [ + 0.2873186573461101 + ], + "mean_centered_acc_runs": [ + -0.013407889237059288 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-288.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-288.json new file mode 100644 index 0000000000000000000000000000000000000000..31a616f148cc6a7f8116cc2070888f9f06286f89 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-288.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-288", + "checkpoint": "checkpoint-288", + "global_step": 288, + "tokens_seen": 301989888, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25801633140808605, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25801633140808605 + ], + "acc": 0.2562238597888867, + "acc_ci": 0.0, + "acc_runs": [ + 0.2562238597888867 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004366488167386302, + "centered_acc": 0.01068844187744807, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01068844187744807 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5070729053318824, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5070729053318824 + ], + "acc": 0.528835690968444, + "acc_ci": 0.0, + "acc_runs": [ + 0.528835690968444 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011664656918146136, + "centered_acc": 0.014145810663764857, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014145810663764857 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.25252525252525254, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25252525252525254 + ], + "acc": 0.25042087542087543, + "acc_ci": 0.0, + "acc_runs": [ + 0.25042087542087543 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008914948991495836, + "centered_acc": 0.0033670033670033885, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0033670033670033885 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26023890784982934, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26023890784982934 + ], + "acc": 0.2030716723549488, + "acc_ci": 0.0, + "acc_runs": [ + 0.2030716723549488 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012821930225112672, + "centered_acc": 0.013651877133105783, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013651877133105783 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.3, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.3 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020514426225627987, + "centered_acc": 0.06666666666666665, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06666666666666665 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4909234411996843, + "acc_ci": 0.0, + "acc_runs": [ + 0.4909234411996843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050170094497657, + "centered_acc": -0.01815311760063143, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.01815311760063143 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 184794.37275025138, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 184794.37275025138 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.177, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.177 + ], + "acc": 0.048, + "acc_ci": 0.0, + "acc_runs": [ + 0.048 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012075463420375087, + "centered_acc": -0.09733333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09733333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29247556618584175, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28537630015065396, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.018932414186508366, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29247556618584175 + ], + "mean_acc_runs": [ + 0.28537630015065396 + ], + "mean_centered_acc_runs": [ + -0.018932414186508366 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-2880.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-2880.json new file mode 100644 index 0000000000000000000000000000000000000000..061c43d5cfdd49a3efe590d33407b794affb3540 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-2880.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2880", + "checkpoint": "checkpoint-2880", + "global_step": 2880, + "tokens_seen": 3019898880, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25781716789484166, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25781716789484166 + ], + "acc": 0.2561242780322645, + "acc_ci": 0.0, + "acc_runs": [ + 0.2561242780322645 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043653883515635365, + "centered_acc": 0.010422890526455545, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.010422890526455545 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5190424374319913, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5190424374319913 + ], + "acc": 0.544613710554951, + "acc_ci": 0.0, + "acc_runs": [ + 0.544613710554951 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011657360703051524, + "centered_acc": 0.0380848748639826, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0380848748639826 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.265993265993266, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.265993265993266 + ], + "acc": 0.24621212121212122, + "acc_ci": 0.0, + "acc_runs": [ + 0.24621212121212122 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009066789565615746, + "centered_acc": 0.021324354657688005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021324354657688005 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26109215017064846, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26109215017064846 + ], + "acc": 0.2098976109215017, + "acc_ci": 0.0, + "acc_runs": [ + 0.2098976109215017 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012835523909473708, + "centered_acc": 0.014789533560864617, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014789533560864617 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.27, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27 + ], + "acc": 0.162, + "acc_ci": 0.0, + "acc_runs": [ + 0.162 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019874354831287473, + "centered_acc": 0.02666666666666669, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02666666666666669 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4940805051302289, + "acc_ci": 0.0, + "acc_runs": [ + 0.4940805051302289 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051500838485925, + "centered_acc": -0.011838989739542227, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.011838989739542227 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 6018945.46504603, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 6018945.46504603 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.187, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.187 + ], + "acc": 0.042, + "acc_ci": 0.0, + "acc_runs": [ + 0.042 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012336254828074166, + "centered_acc": -0.084, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.084 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2934908369151246, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2873215687624767, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.016441749546276006, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2934908369151246 + ], + "mean_acc_runs": [ + 0.2873215687624767 + ], + "mean_centered_acc_runs": [ + -0.016441749546276006 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-3168.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-3168.json new file mode 100644 index 0000000000000000000000000000000000000000..ca8abf1b3db93bc71dffa01bb672f1a400731c70 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-3168.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-3168", + "checkpoint": "checkpoint-3168", + "global_step": 3168, + "tokens_seen": 3321888768, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2588129854610635, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2588129854610635 + ], + "acc": 0.25712009559848636, + "acc_ci": 0.0, + "acc_runs": [ + 0.25712009559848636 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004370875625258676, + "centered_acc": 0.011750647281418022, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011750647281418022 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5266594124047879, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5266594124047879 + ], + "acc": 0.5462459194776932, + "acc_ci": 0.0, + "acc_runs": [ + 0.5462459194776932 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01164922999434743, + "centered_acc": 0.05331882480957573, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05331882480957573 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2718855218855219, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2718855218855219 + ], + "acc": 0.25252525252525254, + "acc_ci": 0.0, + "acc_runs": [ + 0.25252525252525254 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009129795867310605, + "centered_acc": 0.02918069584736253, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02918069584736253 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2508532423208191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2508532423208191 + ], + "acc": 0.21928327645051193, + "acc_ci": 0.0, + "acc_runs": [ + 0.21928327645051193 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012668198621315347, + "centered_acc": 0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.162, + "acc_ci": 0.0, + "acc_runs": [ + 0.162 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5059194948697711, + "acc_ci": 0.0, + "acc_runs": [ + 0.5059194948697711 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051500838485923, + "centered_acc": 0.011838989739542116, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011838989739542116 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 11618776.604118865, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 11618776.604118865 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.191 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012436787112179482, + "centered_acc": -0.07866666666666666, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07866666666666666 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29653519367869874, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.290557556240704, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.010540547668178744, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29653519367869874 + ], + "mean_acc_runs": [ + 0.290557556240704 + ], + "mean_centered_acc_runs": [ + -0.010540547668178744 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-3456.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-3456.json new file mode 100644 index 0000000000000000000000000000000000000000..1d3e254dff61c85f1909837091eb836b8d067e49 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-3456.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-3456", + "checkpoint": "checkpoint-3456", + "global_step": 3456, + "tokens_seen": 3623878656, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25871340370444135, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25871340370444135 + ], + "acc": 0.25562636924915355, + "acc_ci": 0.0, + "acc_runs": [ + 0.25562636924915355 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004370328224831643, + "centered_acc": 0.011617871605921795, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011617871605921795 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5261153427638737, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5261153427638737 + ], + "acc": 0.5408052230685527, + "acc_ci": 0.0, + "acc_runs": [ + 0.5408052230685527 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01164990085426341, + "centered_acc": 0.052230685527747456, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.052230685527747456 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2638888888888889, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2638888888888889 + ], + "acc": 0.25462962962962965, + "acc_ci": 0.0, + "acc_runs": [ + 0.25462962962962965 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009043789220055392, + "centered_acc": 0.018518518518518528, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018518518518518528 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.257679180887372, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.257679180887372 + ], + "acc": 0.21245733788395904, + "acc_ci": 0.0, + "acc_runs": [ + 0.21245733788395904 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012780770562768334, + "centered_acc": 0.010238907849829356, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.010238907849829356 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.282 + ], + "acc": 0.166, + "acc_ci": 0.0, + "acc_runs": [ + 0.166 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020143572847290726, + "centered_acc": 0.04266666666666663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04266666666666663 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5146014206787688, + "acc_ci": 0.0, + "acc_runs": [ + 0.5146014206787688 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014046492383275955, + "centered_acc": 0.02920284135753759, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02920284135753759 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 6122193.9436408905, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 6122193.9436408905 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.19, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.19 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012411851354816256, + "centered_acc": -0.08, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29639946937409595, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29164074422837644, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.006549953880708658, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29639946937409595 + ], + "mean_acc_runs": [ + 0.29164074422837644 + ], + "mean_centered_acc_runs": [ + -0.006549953880708658 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-3744.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-3744.json new file mode 100644 index 0000000000000000000000000000000000000000..b282c4cc755503366b59baee7b2d8ba192aca0d3 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-3744.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-3744", + "checkpoint": "checkpoint-3744", + "global_step": 3744, + "tokens_seen": 3925868544, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26020713005377416, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26020713005377416 + ], + "acc": 0.25791674965146383, + "acc_ci": 0.0, + "acc_runs": [ + 0.25791674965146383 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004378508362084079, + "centered_acc": 0.013609506738365548, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013609506738365548 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5244831338411317, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5244831338411317 + ], + "acc": 0.5429815016322089, + "acc_ci": 0.0, + "acc_runs": [ + 0.5429815016322089 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011651830225710194, + "centered_acc": 0.04896626768226331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04896626768226331 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2601010101010101, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2601010101010101 + ], + "acc": 0.25084175084175087, + "acc_ci": 0.0, + "acc_runs": [ + 0.25084175084175087 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009001718541079735, + "centered_acc": 0.013468013468013481, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013468013468013481 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25341296928327645, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25341296928327645 + ], + "acc": 0.22013651877133106, + "acc_ci": 0.0, + "acc_runs": [ + 0.22013651877133106 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012710896778378677, + "centered_acc": 0.004550625711035261, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.004550625711035261 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.178, + "acc_ci": 0.0, + "acc_runs": [ + 0.178 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5146014206787688, + "acc_ci": 0.0, + "acc_runs": [ + 0.5146014206787688 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014046492383275955, + "centered_acc": 0.02920284135753759, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02920284135753759 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.51, + "acc_ci": 0.0, + "acc_runs": [ + 0.51 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05024183937956913, + "centered_acc": 0.020000000000000018, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020000000000000018 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 6756597.327496335, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 6756597.327496335 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.194, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.194 + ], + "acc": 0.051, + "acc_ci": 0.0, + "acc_runs": [ + 0.051 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012510816141264348, + "centered_acc": -0.07466666666666666, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07466666666666666 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2963673738798654, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29037654033492244, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.015588276462450097, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2963673738798654 + ], + "mean_acc_runs": [ + 0.29037654033492244 + ], + "mean_centered_acc_runs": [ + -0.015588276462450097 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-4032.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-4032.json new file mode 100644 index 0000000000000000000000000000000000000000..80774bec0d929aa98b24f02f217a62e316da3e1b --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-4032.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4032", + "checkpoint": "checkpoint-4032", + "global_step": 4032, + "tokens_seen": 4227858432, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25761800438159727, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25761800438159727 + ], + "acc": 0.2581159131647082, + "acc_ci": 0.0, + "acc_runs": [ + 0.2581159131647082 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004364287353415334, + "centered_acc": 0.01015733917546302, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01015733917546302 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5163220892274211, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5163220892274211 + ], + "acc": 0.5413492927094669, + "acc_ci": 0.0, + "acc_runs": [ + 0.5413492927094669 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011659606710151597, + "centered_acc": 0.03264417845484213, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03264417845484213 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26346801346801346, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26346801346801346 + ], + "acc": 0.2563131313131313, + "acc_ci": 0.0, + "acc_runs": [ + 0.2563131313131313 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009039157374497906, + "centered_acc": 0.017957351290684615, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017957351290684615 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2645051194539249, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2645051194539249 + ], + "acc": 0.21331058020477817, + "acc_ci": 0.0, + "acc_runs": [ + 0.21331058020477817 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012889272949313479, + "centered_acc": 0.019340159271899877, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019340159271899877 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.278, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.278 + ], + "acc": 0.166, + "acc_ci": 0.0, + "acc_runs": [ + 0.166 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020055833888070865, + "centered_acc": 0.037333333333333364, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037333333333333364 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5146014206787688, + "acc_ci": 0.0, + "acc_runs": [ + 0.5146014206787688 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014046492383275955, + "centered_acc": 0.02920284135753759, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02920284135753759 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5, + "acc_ci": 0.0, + "acc_runs": [ + 0.5 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.050251890762960605, + "centered_acc": 0.0, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 5537756.627233293, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 5537756.627233293 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.192, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.192 + ], + "acc": 0.048, + "acc_ci": 0.0, + "acc_runs": [ + 0.048 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01246159264666003, + "centered_acc": -0.07733333333333332, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.07733333333333332 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2953188710884928, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2876171861556178, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.01934702298913023, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2953188710884928 + ], + "mean_acc_runs": [ + 0.2876171861556178 + ], + "mean_centered_acc_runs": [ + -0.01934702298913023 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-4320.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-4320.json new file mode 100644 index 0000000000000000000000000000000000000000..a7a28578b5b1952119cdb7c8b7097119f00709ef --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-4320.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4320", + "checkpoint": "checkpoint-4320", + "global_step": 4320, + "tokens_seen": 4529848320, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26160127464648475, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26160127464648475 + ], + "acc": 0.25672176857199763, + "acc_ci": 0.0, + "acc_runs": [ + 0.25672176857199763 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004386083683839967, + "centered_acc": 0.015468366195313, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015468366195313 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5190424374319913, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5190424374319913 + ], + "acc": 0.5408052230685527, + "acc_ci": 0.0, + "acc_runs": [ + 0.5408052230685527 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011657360703051524, + "centered_acc": 0.0380848748639826, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0380848748639826 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2676767676767677, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2676767676767677 + ], + "acc": 0.2516835016835017, + "acc_ci": 0.0, + "acc_runs": [ + 0.2516835016835017 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009085000147099275, + "centered_acc": 0.023569023569023573, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023569023569023573 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26621160409556316, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26621160409556316 + ], + "acc": 0.2158703071672355, + "acc_ci": 0.0, + "acc_runs": [ + 0.2158703071672355 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012915774781523113, + "centered_acc": 0.021615472127417545, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021615472127417545 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.27, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27 + ], + "acc": 0.166, + "acc_ci": 0.0, + "acc_runs": [ + 0.166 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019874354831287473, + "centered_acc": 0.02666666666666669, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02666666666666669 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5209155485398579, + "acc_ci": 0.0, + "acc_runs": [ + 0.5209155485398579 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014040185494212997, + "centered_acc": 0.04183109707971577, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04183109707971577 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.53, + "acc_ci": 0.0, + "acc_runs": [ + 0.53 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05016135580465919, + "centered_acc": 0.06000000000000005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06000000000000005 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 10874466.034167549, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 10874466.034167549 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.19, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.19 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012411851354816256, + "centered_acc": -0.08, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2957553473084678, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2905283810804845, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.01068773066116446, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2957553473084678 + ], + "mean_acc_runs": [ + 0.2905283810804845 + ], + "mean_centered_acc_runs": [ + -0.01068773066116446 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-4608.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-4608.json new file mode 100644 index 0000000000000000000000000000000000000000..070536b05937f70fa25c2aa1040fd180f49e5499 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-4608.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4608", + "checkpoint": "checkpoint-4608", + "global_step": 4608, + "tokens_seen": 4831838208, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.261700856403107, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.261700856403107 + ], + "acc": 0.2582154949213304, + "acc_ci": 0.0, + "acc_runs": [ + 0.2582154949213304 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004386622589118801, + "centered_acc": 0.015601141870809299, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015601141870809299 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5070729053318824, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5070729053318824 + ], + "acc": 0.5277475516866159, + "acc_ci": 0.0, + "acc_runs": [ + 0.5277475516866159 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011664656918146136, + "centered_acc": 0.014145810663764857, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014145810663764857 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2601010101010101, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2601010101010101 + ], + "acc": 0.24957912457912457, + "acc_ci": 0.0, + "acc_runs": [ + 0.24957912457912457 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009001718541079735, + "centered_acc": 0.013468013468013481, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013468013468013481 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26791808873720135, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26791808873720135 + ], + "acc": 0.21843003412969283, + "acc_ci": 0.0, + "acc_runs": [ + 0.21843003412969283 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012942030195136482, + "centered_acc": 0.023890784982935138, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023890784982935138 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.272, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.272 + ], + "acc": 0.164, + "acc_ci": 0.0, + "acc_runs": [ + 0.164 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019920483209566107, + "centered_acc": 0.02933333333333336, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02933333333333336 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5067087608524072, + "acc_ci": 0.0, + "acc_runs": [ + 0.5067087608524072 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051220692330247, + "centered_acc": 0.013417521704814472, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013417521704814472 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.56, + "acc_ci": 0.0, + "acc_runs": [ + 0.56 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049888765156985884, + "centered_acc": 0.1200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1200000000000001 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 9808288.695865558, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 9808288.695865558 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.189, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.189 + ], + "acc": 0.048, + "acc_ci": 0.0, + "acc_runs": [ + 0.048 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01238678458811779, + "centered_acc": -0.08133333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08133333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2929654767622002, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29109684279428716, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.010544644862473553, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2929654767622002 + ], + "mean_acc_runs": [ + 0.29109684279428716 + ], + "mean_centered_acc_runs": [ + -0.010544644862473553 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-4896.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-4896.json new file mode 100644 index 0000000000000000000000000000000000000000..0e8397fadb472ea48340052efda0133a555c28ac --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-4896.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4896", + "checkpoint": "checkpoint-4896", + "global_step": 4896, + "tokens_seen": 5133828096, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26010754829715194, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26010754829715194 + ], + "acc": 0.2590121489743079, + "acc_ci": 0.0, + "acc_runs": [ + 0.2590121489743079 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004377965074211591, + "centered_acc": 0.013476731062869248, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013476731062869248 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5277475516866159, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5277475516866159 + ], + "acc": 0.5331882480957563, + "acc_ci": 0.0, + "acc_runs": [ + 0.5331882480957563 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011647846656062001, + "centered_acc": 0.055495103373231824, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.055495103373231824 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26304713804713803, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26304713804713803 + ], + "acc": 0.25462962962962965, + "acc_ci": 0.0, + "acc_runs": [ + 0.25462962962962965 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009034514898865944, + "centered_acc": 0.017396184062850706, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017396184062850706 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26023890784982934, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26023890784982934 + ], + "acc": 0.2175767918088737, + "acc_ci": 0.0, + "acc_runs": [ + 0.2175767918088737 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012821930225112672, + "centered_acc": 0.013651877133105783, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013651877133105783 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.278, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.278 + ], + "acc": 0.162, + "acc_ci": 0.0, + "acc_runs": [ + 0.162 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020055833888070865, + "centered_acc": 0.037333333333333364, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037333333333333364 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5122336227308603, + "acc_ci": 0.0, + "acc_runs": [ + 0.5122336227308603 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014048278820405729, + "centered_acc": 0.02446724546172052, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02446724546172052 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.52, + "acc_ci": 0.0, + "acc_runs": [ + 0.52 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05021167315686783, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 13003877.95079824, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 13003877.95079824 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.19, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.19 + ], + "acc": 0.051, + "acc_ci": 0.0, + "acc_runs": [ + 0.051 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012411851354816256, + "centered_acc": -0.08, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29652352431345586, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2887927903013129, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.013511622447276432, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29652352431345586 + ], + "mean_acc_runs": [ + 0.2887927903013129 + ], + "mean_centered_acc_runs": [ + -0.013511622447276432 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-5184.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-5184.json new file mode 100644 index 0000000000000000000000000000000000000000..4b1fcf914872beeb9d9b93aa1636d4da8eb46c96 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-5184.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-5184", + "checkpoint": "checkpoint-5184", + "global_step": 5184, + "tokens_seen": 5435817984, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25562636924915355, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25562636924915355 + ], + "acc": 0.2604062935670185, + "acc_ci": 0.0, + "acc_runs": [ + 0.2604062935670185 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004353212146198103, + "centered_acc": 0.007501825665538068, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007501825665538068 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5136017410228509, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5136017410228509 + ], + "acc": 0.530467899891186, + "acc_ci": 0.0, + "acc_runs": [ + 0.530467899891186 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011661506839823792, + "centered_acc": 0.027203482045701888, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.027203482045701888 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26557239057239057, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26557239057239057 + ], + "acc": 0.2588383838383838, + "acc_ci": 0.0, + "acc_runs": [ + 0.2588383838383838 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00906221062697191, + "centered_acc": 0.020763187429854096, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020763187429854096 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25853242320819114, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25853242320819114 + ], + "acc": 0.2158703071672355, + "acc_ci": 0.0, + "acc_runs": [ + 0.2158703071672355 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012794553754288777, + "centered_acc": 0.011376564277588189, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011376564277588189 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.266, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.266 + ], + "acc": 0.16, + "acc_ci": 0.0, + "acc_runs": [ + 0.16 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019780559675655396, + "centered_acc": 0.021333333333333353, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021333333333333353 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5153906866614049, + "acc_ci": 0.0, + "acc_runs": [ + 0.5153906866614049 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014045826789783585, + "centered_acc": 0.030781373322809724, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.030781373322809724 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5, + "acc_ci": 0.0, + "acc_runs": [ + 0.5 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.050251890762960605, + "centered_acc": 0.0, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 16251763.125292294, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 16251763.125292294 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29038882067543104, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2866455094610553, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.023755404856789706, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29038882067543104 + ], + "mean_acc_runs": [ + 0.2866455094610553 + ], + "mean_centered_acc_runs": [ + -0.023755404856789706 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-5472.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-5472.json new file mode 100644 index 0000000000000000000000000000000000000000..eb4a1550f636ef102fbf9598f89cd2cb853ea138 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-5472.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-5472", + "checkpoint": "checkpoint-5472", + "global_step": 5472, + "tokens_seen": 5737807872, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2597092212706632, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2597092212706632 + ], + "acc": 0.2584146584345748, + "acc_ci": 0.0, + "acc_runs": [ + 0.2584146584345748 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004375788991216637, + "centered_acc": 0.012945628360884273, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012945628360884273 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5179542981501633, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5179542981501633 + ], + "acc": 0.5217627856365615, + "acc_ci": 0.0, + "acc_runs": [ + 0.5217627856365615 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011658300623287121, + "centered_acc": 0.0359085963003265, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0359085963003265 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2676767676767677, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2676767676767677 + ], + "acc": 0.24915824915824916, + "acc_ci": 0.0, + "acc_runs": [ + 0.24915824915824916 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009085000147099275, + "centered_acc": 0.023569023569023573, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023569023569023573 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.27047781569965873, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27047781569965873 + ], + "acc": 0.21331058020477817, + "acc_ci": 0.0, + "acc_runs": [ + 0.21331058020477817 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012980954547659553, + "centered_acc": 0.02730375426621164, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02730375426621164 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.266, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.266 + ], + "acc": 0.164, + "acc_ci": 0.0, + "acc_runs": [ + 0.164 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019780559675655396, + "centered_acc": 0.021333333333333353, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021333333333333353 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5138121546961326, + "acc_ci": 0.0, + "acc_runs": [ + 0.5138121546961326 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014047122916440271, + "centered_acc": 0.027624309392265234, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.027624309392265234 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.53, + "acc_ci": 0.0, + "acc_runs": [ + 0.53 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05016135580465919, + "centered_acc": 0.06000000000000005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06000000000000005 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 17905072.61922125, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 17905072.61922125 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2938030171328755, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2875939951615621, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.014082270136728303, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2938030171328755 + ], + "mean_acc_runs": [ + 0.2875939951615621 + ], + "mean_centered_acc_runs": [ + -0.014082270136728303 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-576.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-576.json new file mode 100644 index 0000000000000000000000000000000000000000..e10902c7408142d79d0437d9300adfc37e08b7f3 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-576.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-576", + "checkpoint": "checkpoint-576", + "global_step": 576, + "tokens_seen": 603979776, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25801633140808605, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25801633140808605 + ], + "acc": 0.2566221868153754, + "acc_ci": 0.0, + "acc_runs": [ + 0.2566221868153754 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004366488167386302, + "centered_acc": 0.01068844187744807, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01068844187744807 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5228509249183896, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5228509249183896 + ], + "acc": 0.5326441784548422, + "acc_ci": 0.0, + "acc_runs": [ + 0.5326441784548422 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011653634832401168, + "centered_acc": 0.04570184983677916, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04570184983677916 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26430976430976433, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26430976430976433 + ], + "acc": 0.2474747474747475, + "acc_ci": 0.0, + "acc_runs": [ + 0.2474747474747475 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009048410451862887, + "centered_acc": 0.019079685746352437, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019079685746352437 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2687713310580205, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2687713310580205 + ], + "acc": 0.21843003412969283, + "acc_ci": 0.0, + "acc_runs": [ + 0.21843003412969283 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012955065963710852, + "centered_acc": 0.025028441410693974, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.025028441410693974 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.298, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.298 + ], + "acc": 0.152, + "acc_ci": 0.0, + "acc_runs": [ + 0.152 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020475118092988954, + "centered_acc": 0.06399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06399999999999999 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.47908445146014206, + "acc_ci": 0.0, + "acc_runs": [ + 0.47908445146014206 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014040185494212999, + "centered_acc": -0.041831097079715884, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.041831097079715884 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 1167376.1007467627, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 1167376.1007467627 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.177, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.177 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012075463420375087, + "centered_acc": -0.09733333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09733333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29815805861571004, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28885430601085005, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.006454565332708338, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29815805861571004 + ], + "mean_acc_runs": [ + 0.28885430601085005 + ], + "mean_centered_acc_runs": [ + -0.006454565332708338 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-5760.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-5760.json new file mode 100644 index 0000000000000000000000000000000000000000..eda7fe521d440722fc55109345f01f9b10e950f9 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-5760.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-5760", + "checkpoint": "checkpoint-5760", + "global_step": 5760, + "tokens_seen": 6039797760, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2584146584345748, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2584146584345748 + ], + "acc": 0.2559251145190201, + "acc_ci": 0.0, + "acc_runs": [ + 0.2559251145190201 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043686842556263684, + "centered_acc": 0.011219544579433046, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011219544579433046 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.499455930359086, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.499455930359086 + ], + "acc": 0.5261153427638737, + "acc_ci": 0.0, + "acc_runs": [ + 0.5261153427638737 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665817258899291, + "centered_acc": -0.0010881392818280489, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0010881392818280489 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26262626262626265, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26262626262626265 + ], + "acc": 0.25126262626262624, + "acc_ci": 0.0, + "acc_runs": [ + 0.25126262626262624 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009029861776763957, + "centered_acc": 0.01683501683501687, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01683501683501687 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2636518771331058, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2636518771331058 + ], + "acc": 0.2158703071672355, + "acc_ci": 0.0, + "acc_runs": [ + 0.2158703071672355 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012875929151296968, + "centered_acc": 0.018202502844141044, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018202502844141044 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.268, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.268 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019827714859587477, + "centered_acc": 0.02400000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02400000000000002 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5122336227308603, + "acc_ci": 0.0, + "acc_runs": [ + 0.5122336227308603 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014048278820405729, + "centered_acc": 0.02446724546172052, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02446724546172052 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5, + "acc_ci": 0.0, + "acc_runs": [ + 0.5 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.050251890762960605, + "centered_acc": 0.0, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 15540285.037326762, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 15540285.037326762 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.182, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.182 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0122075806376622, + "centered_acc": -0.09066666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09066666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2890247880921715, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2842694475217316, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.026717285853420286, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2890247880921715 + ], + "mean_acc_runs": [ + 0.2842694475217316 + ], + "mean_centered_acc_runs": [ + -0.026717285853420286 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-6048.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-6048.json new file mode 100644 index 0000000000000000000000000000000000000000..b6d3ec44b8e4284fb3c7dba0d8e37dfc9dabb7e4 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-6048.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6048", + "checkpoint": "checkpoint-6048", + "global_step": 6048, + "tokens_seen": 6341787648, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2606054570802629, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2606054570802629 + ], + "acc": 0.2566221868153754, + "acc_ci": 0.0, + "acc_runs": [ + 0.2566221868153754 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004380678585341082, + "centered_acc": 0.014140609440350524, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014140609440350524 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.499455930359086, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.499455930359086 + ], + "acc": 0.5331882480957563, + "acc_ci": 0.0, + "acc_runs": [ + 0.5331882480957563 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665817258899291, + "centered_acc": -0.0010881392818280489, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0010881392818280489 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.27146464646464646, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.27146464646464646 + ], + "acc": 0.25547138047138046, + "acc_ci": 0.0, + "acc_runs": [ + 0.25547138047138046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00912536297036068, + "centered_acc": 0.028619528619528618, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.028619528619528618 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2636518771331058, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2636518771331058 + ], + "acc": 0.2167235494880546, + "acc_ci": 0.0, + "acc_runs": [ + 0.2167235494880546 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012875929151296968, + "centered_acc": 0.018202502844141044, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018202502844141044 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.17, + "acc_ci": 0.0, + "acc_runs": [ + 0.17 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5067087608524072, + "acc_ci": 0.0, + "acc_runs": [ + 0.5067087608524072 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051220692330247, + "centered_acc": 0.013417521704814472, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013417521704814472 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.51, + "acc_ci": 0.0, + "acc_runs": [ + 0.51 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05024183937956913, + "centered_acc": 0.020000000000000018, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020000000000000018 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 11857136.932694603, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 11857136.932694603 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.185, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.185 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012285191326386655, + "centered_acc": -0.08666666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08666666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2923629851728502, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2873195649208299, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.022755524421362153, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2923629851728502 + ], + "mean_acc_runs": [ + 0.2873195649208299 + ], + "mean_centered_acc_runs": [ + -0.022755524421362153 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-6336.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-6336.json new file mode 100644 index 0000000000000000000000000000000000000000..d90a8db107b88f279e3d6a075b59038440a4a135 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-6336.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6336", + "checkpoint": "checkpoint-6336", + "global_step": 6336, + "tokens_seen": 6643777536, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2636924915355507, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2636924915355507 + ], + "acc": 0.2577175861382195, + "acc_ci": 0.0, + "acc_runs": [ + 0.2577175861382195 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004397339661695592, + "centered_acc": 0.01825665538073425, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01825665538073425 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5048966267682263, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5048966267682263 + ], + "acc": 0.5261153427638737, + "acc_ci": 0.0, + "acc_runs": [ + 0.5261153427638737 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665264730078137, + "centered_acc": 0.009793253536452662, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009793253536452662 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.265993265993266, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.265993265993266 + ], + "acc": 0.2558922558922559, + "acc_ci": 0.0, + "acc_runs": [ + 0.2558922558922559 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009066789565615746, + "centered_acc": 0.021324354657688005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021324354657688005 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2738907849829352, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2738907849829352 + ], + "acc": 0.21416382252559726, + "acc_ci": 0.0, + "acc_runs": [ + 0.21416382252559726 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.013032004972989623, + "centered_acc": 0.031854379977246904, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.031854379977246904 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.164, + "acc_ci": 0.0, + "acc_runs": [ + 0.164 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.510655090765588, + "acc_ci": 0.0, + "acc_runs": [ + 0.510655090765588 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014049294536290372, + "centered_acc": 0.02131018153117603, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02131018153117603 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.49, + "acc_ci": 0.0, + "acc_runs": [ + 0.49 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05024183937956913, + "centered_acc": -0.020000000000000018, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.020000000000000018 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 20496188.555832285, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 20496188.555832285 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.184, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.184 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01225945734093864, + "centered_acc": -0.08800000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08800000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29441219487999637, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28438315598592345, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.024098472374366835, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29441219487999637 + ], + "mean_acc_runs": [ + 0.28438315598592345 + ], + "mean_centered_acc_runs": [ + -0.024098472374366835 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-6624.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-6624.json new file mode 100644 index 0000000000000000000000000000000000000000..e904dde7b413658b3b0c23b797e2f1555493b41b --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-6624.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6624", + "checkpoint": "checkpoint-6624", + "global_step": 6624, + "tokens_seen": 6945767424, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2599083847839076, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2599083847839076 + ], + "acc": 0.2577175861382195, + "acc_ci": 0.0, + "acc_runs": [ + 0.2577175861382195 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004376877619233649, + "centered_acc": 0.013211179711876797, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013211179711876797 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5136017410228509, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5136017410228509 + ], + "acc": 0.5315560391730142, + "acc_ci": 0.0, + "acc_runs": [ + 0.5315560391730142 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011661506839823792, + "centered_acc": 0.027203482045701888, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.027203482045701888 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26851851851851855, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26851851851851855 + ], + "acc": 0.2622053872053872, + "acc_ci": 0.0, + "acc_runs": [ + 0.2622053872053872 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009094042554994908, + "centered_acc": 0.024691358024691395, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.024691358024691395 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2636518771331058, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2636518771331058 + ], + "acc": 0.21843003412969283, + "acc_ci": 0.0, + "acc_runs": [ + 0.21843003412969283 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012875929151296968, + "centered_acc": 0.018202502844141044, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.018202502844141044 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.16, + "acc_ci": 0.0, + "acc_runs": [ + 0.16 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5114443567482242, + "acc_ci": 0.0, + "acc_runs": [ + 0.5114443567482242 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014048804199859391, + "centered_acc": 0.022888713496448387, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.022888713496448387 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.47, + "acc_ci": 0.0, + "acc_runs": [ + 0.47 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05016135580465919, + "centered_acc": -0.06000000000000005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.06000000000000005 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 29291163.031463638, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 29291163.031463638 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.184, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.184 + ], + "acc": 0.05, + "acc_ci": 0.0, + "acc_runs": [ + 0.05 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01225945734093864, + "centered_acc": -0.08800000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08800000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29394675357639716, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2839834926879862, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.02813642670330443, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29394675357639716 + ], + "mean_acc_runs": [ + 0.2839834926879862 + ], + "mean_centered_acc_runs": [ + -0.02813642670330443 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-6912.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-6912.json new file mode 100644 index 0000000000000000000000000000000000000000..632c9bd64b9b57f7d45f0d6fe84a43036f7f891b --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-6912.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6912", + "checkpoint": "checkpoint-6912", + "global_step": 6912, + "tokens_seen": 7247757312, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26090420235012945, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26090420235012945 + ], + "acc": 0.25951005775741887, + "acc_ci": 0.0, + "acc_runs": [ + 0.25951005775741887 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004382303181183701, + "centered_acc": 0.014538936466839273, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014538936466839273 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5021762785636561, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5021762785636561 + ], + "acc": 0.5261153427638737, + "acc_ci": 0.0, + "acc_runs": [ + 0.5261153427638737 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665713661738719, + "centered_acc": 0.0043525571273121955, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0043525571273121955 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2697811447811448, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2697811447811448 + ], + "acc": 0.2516835016835017, + "acc_ci": 0.0, + "acc_runs": [ + 0.2516835016835017 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009107527914671116, + "centered_acc": 0.02637485970819305, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02637485970819305 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2619453924914676, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2619453924914676 + ], + "acc": 0.21075085324232082, + "acc_ci": 0.0, + "acc_runs": [ + 0.21075085324232082 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012849054826857987, + "centered_acc": 0.015927189988623452, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015927189988623452 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.26, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26 + ], + "acc": 0.168, + "acc_ci": 0.0, + "acc_runs": [ + 0.168 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019635965529725526, + "centered_acc": 0.013333333333333345, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013333333333333345 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.510655090765588, + "acc_ci": 0.0, + "acc_runs": [ + 0.510655090765588 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014049294536290372, + "centered_acc": 0.02131018153117603, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02131018153117603 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48, + "acc_ci": 0.0, + "acc_runs": [ + 0.48 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05021167315686783, + "centered_acc": -0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.040000000000000036 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 30090393.269364145, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 30090393.269364145 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.182, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.182 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0122075806376622, + "centered_acc": -0.09066666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09066666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.28946783636439966, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28310023079864033, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.03091718721819875, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.28946783636439966 + ], + "mean_acc_runs": [ + 0.28310023079864033 + ], + "mean_centered_acc_runs": [ + -0.03091718721819875 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-72.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-72.json new file mode 100644 index 0000000000000000000000000000000000000000..8a777dd200fe45b71e91f910e89f57cbfec347bc --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-72.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-72", + "checkpoint": "checkpoint-72", + "global_step": 72, + "tokens_seen": 75497472, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2672774347739494, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2672774347739494 + ], + "acc": 0.2582154949213304, + "acc_ci": 0.0, + "acc_runs": [ + 0.2582154949213304 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004416339450436367, + "centered_acc": 0.02303657969859918, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02303657969859918 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5179542981501633, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5179542981501633 + ], + "acc": 0.5277475516866159, + "acc_ci": 0.0, + "acc_runs": [ + 0.5277475516866159 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011658300623287121, + "centered_acc": 0.0359085963003265, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0359085963003265 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.255050505050505, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.255050505050505 + ], + "acc": 0.24705387205387205, + "acc_ci": 0.0, + "acc_runs": [ + 0.24705387205387205 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008944265906130788, + "centered_acc": 0.006734006734006703, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006734006734006703 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2593856655290102, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2593856655290102 + ], + "acc": 0.1945392491467577, + "acc_ci": 0.0, + "acc_runs": [ + 0.1945392491467577 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012808273573927224, + "centered_acc": 0.012514220705346949, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012514220705346949 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.278, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.278 + ], + "acc": 0.15, + "acc_ci": 0.0, + "acc_runs": [ + 0.15 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020055833888070865, + "centered_acc": 0.037333333333333364, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.037333333333333364 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5122336227308603, + "acc_ci": 0.0, + "acc_runs": [ + 0.5122336227308603 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014048278820405729, + "centered_acc": 0.02446724546172052, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02446724546172052 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.54, + "acc_ci": 0.0, + "acc_runs": [ + 0.54 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05009082659620331, + "centered_acc": 0.08000000000000007, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.08000000000000007 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 10481.449047254906, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 10481.449047254906 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.173, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.173 + ], + "acc": 0.039, + "acc_ci": 0.0, + "acc_runs": [ + 0.039 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011967214137559927, + "centered_acc": -0.10266666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10266666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29177798391727133, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2847077252313137, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.014010862320659198, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29177798391727133 + ], + "mean_acc_runs": [ + 0.2847077252313137 + ], + "mean_centered_acc_runs": [ + -0.014010862320659198 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-7200.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-7200.json new file mode 100644 index 0000000000000000000000000000000000000000..adcfee488ecb59ae884ec30ff6fb050e22871b7d --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-7200.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-7200", + "checkpoint": "checkpoint-7200", + "global_step": 7200, + "tokens_seen": 7549747200, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2629954192391954, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2629954192391954 + ], + "acc": 0.25891256721768574, + "acc_ci": 0.0, + "acc_runs": [ + 0.25891256721768574 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004393601887506574, + "centered_acc": 0.017327225652260525, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017327225652260525 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5070729053318824, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5070729053318824 + ], + "acc": 0.5310119695321001, + "acc_ci": 0.0, + "acc_runs": [ + 0.5310119695321001 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011664656918146136, + "centered_acc": 0.014145810663764857, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014145810663764857 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26851851851851855, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26851851851851855 + ], + "acc": 0.26052188552188554, + "acc_ci": 0.0, + "acc_runs": [ + 0.26052188552188554 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009094042554994908, + "centered_acc": 0.024691358024691395, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.024691358024691395 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26791808873720135, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26791808873720135 + ], + "acc": 0.2167235494880546, + "acc_ci": 0.0, + "acc_runs": [ + 0.2167235494880546 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012942030195136482, + "centered_acc": 0.023890784982935138, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023890784982935138 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.266, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.266 + ], + "acc": 0.164, + "acc_ci": 0.0, + "acc_runs": [ + 0.164 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019780559675655396, + "centered_acc": 0.021333333333333353, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021333333333333353 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5059194948697711, + "acc_ci": 0.0, + "acc_runs": [ + 0.5059194948697711 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051500838485923, + "centered_acc": 0.011838989739542116, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.011838989739542116 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5, + "acc_ci": 0.0, + "acc_runs": [ + 0.5 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.050251890762960605, + "centered_acc": 0.0, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 31246624.518439956, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 31246624.518439956 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.186, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.186 + ], + "acc": 0.049, + "acc_ci": 0.0, + "acc_runs": [ + 0.049 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012310790208412926, + "centered_acc": -0.08533333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08533333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2930841553044663, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2864570990114822, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.02394787859882281, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2930841553044663 + ], + "mean_acc_runs": [ + 0.2864570990114822 + ], + "mean_centered_acc_runs": [ + -0.02394787859882281 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-7488.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-7488.json new file mode 100644 index 0000000000000000000000000000000000000000..6913d619fb531e28dd12ff3ec40b5b467bbf5c6e --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-7488.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-7488", + "checkpoint": "checkpoint-7488", + "global_step": 7488, + "tokens_seen": 7851737088, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25960963951404104, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25960963951404104 + ], + "acc": 0.25851424019119695, + "acc_ci": 0.0, + "acc_runs": [ + 0.25851424019119695 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004375244237044889, + "centered_acc": 0.012812852685388046, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012812852685388046 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5021762785636561, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5021762785636561 + ], + "acc": 0.5223068552774756, + "acc_ci": 0.0, + "acc_runs": [ + 0.5223068552774756 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665713661738719, + "centered_acc": 0.0043525571273121955, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0043525571273121955 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2718855218855219, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2718855218855219 + ], + "acc": 0.25252525252525254, + "acc_ci": 0.0, + "acc_runs": [ + 0.25252525252525254 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009129795867310605, + "centered_acc": 0.02918069584736253, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02918069584736253 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26706484641638223, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26706484641638223 + ], + "acc": 0.20819112627986347, + "acc_ci": 0.0, + "acc_runs": [ + 0.20819112627986347 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01292893319649643, + "centered_acc": 0.022753128555176305, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.022753128555176305 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.268, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.268 + ], + "acc": 0.162, + "acc_ci": 0.0, + "acc_runs": [ + 0.162 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019827714859587477, + "centered_acc": 0.02400000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02400000000000002 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4925019731649566, + "acc_ci": 0.0, + "acc_runs": [ + 0.4925019731649566 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050905521228646, + "centered_acc": -0.014996053670086829, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.014996053670086829 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.51, + "acc_ci": 0.0, + "acc_runs": [ + 0.51 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05024183937956913, + "centered_acc": 0.020000000000000018, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020000000000000018 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 28540135.2715266, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 28540135.2715266 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.188, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.188 + ], + "acc": 0.05, + "acc_ci": 0.0, + "acc_runs": [ + 0.05 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01236158601510376, + "centered_acc": -0.08266666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08266666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29278938106326685, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.283452097092407, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.02533206250823486, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29278938106326685 + ], + "mean_acc_runs": [ + 0.283452097092407 + ], + "mean_centered_acc_runs": [ + -0.02533206250823486 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-7776.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-7776.json new file mode 100644 index 0000000000000000000000000000000000000000..2a3af3370dd16c3c831020301e17cef7cfbf8d61 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-7776.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-7776", + "checkpoint": "checkpoint-7776", + "global_step": 7776, + "tokens_seen": 8153726976, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2593108942441745, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2593108942441745 + ], + "acc": 0.25712009559848636, + "acc_ci": 0.0, + "acc_runs": [ + 0.25712009559848636 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043736082125607105, + "centered_acc": 0.012414525658899297, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012414525658899297 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5076169749727966, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5076169749727966 + ], + "acc": 0.5282916213275299, + "acc_ci": 0.0, + "acc_runs": [ + 0.5282916213275299 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011664470424044771, + "centered_acc": 0.015233949945593128, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015233949945593128 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2706228956228956, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2706228956228956 + ], + "acc": 0.2529461279461279, + "acc_ci": 0.0, + "acc_runs": [ + 0.2529461279461279 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009116466166404035, + "centered_acc": 0.0274971941638608, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0274971941638608 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26109215017064846, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26109215017064846 + ], + "acc": 0.21245733788395904, + "acc_ci": 0.0, + "acc_runs": [ + 0.21245733788395904 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012835523909473708, + "centered_acc": 0.014789533560864617, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014789533560864617 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.276, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.276 + ], + "acc": 0.158, + "acc_ci": 0.0, + "acc_runs": [ + 0.158 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.02001121929807355, + "centered_acc": 0.0346666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0346666666666667 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4909234411996843, + "acc_ci": 0.0, + "acc_runs": [ + 0.4909234411996843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050170094497657, + "centered_acc": -0.01815311760063143, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.01815311760063143 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.52, + "acc_ci": 0.0, + "acc_runs": [ + 0.52 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05021167315686783, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 20094626.55806883, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 20094626.55806883 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.182, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.182 + ], + "acc": 0.048, + "acc_ci": 0.0, + "acc_runs": [ + 0.048 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0122075806376622, + "centered_acc": -0.09066666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09066666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29277381916841916, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28462201474411125, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.02307144341377921, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29277381916841916 + ], + "mean_acc_runs": [ + 0.28462201474411125 + ], + "mean_centered_acc_runs": [ + -0.02307144341377921 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-8064.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-8064.json new file mode 100644 index 0000000000000000000000000000000000000000..3a3795d091b3de56db8f30767f244e44a038d0d6 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-8064.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-8064", + "checkpoint": "checkpoint-8064", + "global_step": 8064, + "tokens_seen": 8455716864, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2591117307309301, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2591117307309301 + ], + "acc": 0.2577175861382195, + "acc_ci": 0.0, + "acc_runs": [ + 0.2577175861382195 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004372516060164899, + "centered_acc": 0.012148974307906771, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012148974307906771 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5027203482045702, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5027203482045702 + ], + "acc": 0.5217627856365615, + "acc_ci": 0.0, + "acc_runs": [ + 0.5217627856365615 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665651503000776, + "centered_acc": 0.005440696409140466, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005440696409140466 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2702020202020202, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2702020202020202 + ], + "acc": 0.25084175084175087, + "acc_ci": 0.0, + "acc_runs": [ + 0.25084175084175087 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009112002229119622, + "centered_acc": 0.026936026936026963, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.026936026936026963 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26023890784982934, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26023890784982934 + ], + "acc": 0.20733788395904437, + "acc_ci": 0.0, + "acc_runs": [ + 0.20733788395904437 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012821930225112672, + "centered_acc": 0.013651877133105783, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013651877133105783 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.268, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.268 + ], + "acc": 0.166, + "acc_ci": 0.0, + "acc_runs": [ + 0.166 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019827714859587477, + "centered_acc": 0.02400000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02400000000000002 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48382004735595896, + "acc_ci": 0.0, + "acc_runs": [ + 0.48382004735595896 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014045126130978676, + "centered_acc": -0.03235990528808208, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.03235990528808208 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.49, + "acc_ci": 0.0, + "acc_runs": [ + 0.49 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05024183937956913, + "centered_acc": -0.020000000000000018, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.020000000000000018 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 27210636.409949224, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 27210636.409949224 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.187, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.187 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012336254828074166, + "centered_acc": -0.084, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.084 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.291212167831225, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.280296157741686, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.03306748966161127, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.291212167831225 + ], + "mean_acc_runs": [ + 0.280296157741686 + ], + "mean_centered_acc_runs": [ + -0.03306748966161127 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-8352.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-8352.json new file mode 100644 index 0000000000000000000000000000000000000000..3488403722470e08b79e7a3d2d7062854b9375d8 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-8352.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-8352", + "checkpoint": "checkpoint-8352", + "global_step": 8352, + "tokens_seen": 8757706752, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26010754829715194, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26010754829715194 + ], + "acc": 0.25721967735510853, + "acc_ci": 0.0, + "acc_runs": [ + 0.25721967735510853 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004377965074211591, + "centered_acc": 0.013476731062869248, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013476731062869248 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5179542981501633, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5179542981501633 + ], + "acc": 0.5266594124047879, + "acc_ci": 0.0, + "acc_runs": [ + 0.5266594124047879 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011658300623287121, + "centered_acc": 0.0359085963003265, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0359085963003265 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2702020202020202, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2702020202020202 + ], + "acc": 0.2516835016835017, + "acc_ci": 0.0, + "acc_runs": [ + 0.2516835016835017 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009112002229119622, + "centered_acc": 0.026936026936026963, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.026936026936026963 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25341296928327645, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25341296928327645 + ], + "acc": 0.20648464163822525, + "acc_ci": 0.0, + "acc_runs": [ + 0.20648464163822525 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012710896778378677, + "centered_acc": 0.004550625711035261, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.004550625711035261 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.262, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.262 + ], + "acc": 0.162, + "acc_ci": 0.0, + "acc_runs": [ + 0.162 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019684688820194637, + "centered_acc": 0.016000000000000014, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.016000000000000014 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48697711128650356, + "acc_ci": 0.0, + "acc_runs": [ + 0.48697711128650356 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014047718393997596, + "centered_acc": -0.026045777426992878, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.026045777426992878 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5, + "acc_ci": 0.0, + "acc_runs": [ + 0.5 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.050251890762960605, + "centered_acc": 0.0, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 29409355.486137535, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 29409355.486137535 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.188, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.188 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01236158601510376, + "centered_acc": -0.08266666666666667, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08266666666666667 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29194613932210195, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28165058678534516, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.028362837837333437, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29194613932210195 + ], + "mean_acc_runs": [ + 0.28165058678534516 + ], + "mean_centered_acc_runs": [ + -0.028362837837333437 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/checkpoint-864.json b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-864.json new file mode 100644 index 0000000000000000000000000000000000000000..69c2d1157a19b42f464e5005316a6fbb010f0819 --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/checkpoint-864.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", + "run": "mask30-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-864", + "checkpoint": "checkpoint-864", + "global_step": 864, + "tokens_seen": 905969664, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2599083847839076, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2599083847839076 + ], + "acc": 0.25492929695279826, + "acc_ci": 0.0, + "acc_runs": [ + 0.25492929695279826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004376877619233649, + "centered_acc": 0.013211179711876797, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013211179711876797 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5184983677910773, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5184983677910773 + ], + "acc": 0.5505984766050055, + "acc_ci": 0.0, + "acc_runs": [ + 0.5505984766050055 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165783758381818, + "centered_acc": 0.03699673558215455, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03699673558215455 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2647306397306397, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2647306397306397 + ], + "acc": 0.2601010101010101, + "acc_ci": 0.0, + "acc_runs": [ + 0.2601010101010101 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009053021086174166, + "centered_acc": 0.019640852974186274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019640852974186274 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2508532423208191, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2508532423208191 + ], + "acc": 0.2090443686006826, + "acc_ci": 0.0, + "acc_runs": [ + 0.2090443686006826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012668198621315347, + "centered_acc": 0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.298, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.298 + ], + "acc": 0.154, + "acc_ci": 0.0, + "acc_runs": [ + 0.154 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020475118092988954, + "centered_acc": 0.06399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06399999999999999 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4877663772691397, + "acc_ci": 0.0, + "acc_runs": [ + 0.4877663772691397 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01404827882040573, + "centered_acc": -0.024467245461720633, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.024467245461720633 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 2060784.008710211, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 2060784.008710211 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2954984391044073, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28897269913023366, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.013878433024260388, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2954984391044073 + ], + "mean_acc_runs": [ + 0.28897269913023366 + ], + "mean_centered_acc_runs": [ + -0.013878433024260388 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/mask30-fulle-lm/zero_shot_eval/summary.jsonl b/results/mask30-fulle-lm/zero_shot_eval/summary.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..b3773c9c2f83f92a136a9ddcbbcd7b1354b231ee --- /dev/null +++ b/results/mask30-fulle-lm/zero_shot_eval/summary.jsonl @@ -0,0 +1,30 @@ +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-72", "checkpoint": "checkpoint-72", "global_step": 72, "tokens_seen": 75497472, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2672774347739494, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2672774347739494], "acc": 0.2582154949213304, "acc_ci": 0.0, "acc_runs": [0.2582154949213304], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004416339450436367, "centered_acc": 0.02303657969859918, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02303657969859918], "ci": 0.0}, "piqa": {"acc_norm": 0.5179542981501633, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5179542981501633], "acc": 0.5277475516866159, "acc_ci": 0.0, "acc_runs": [0.5277475516866159], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011658300623287121, "centered_acc": 0.0359085963003265, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0359085963003265], "ci": 0.0}, "arc_easy": {"acc_norm": 0.255050505050505, "acc_norm_ci": 0.0, "acc_norm_runs": [0.255050505050505], "acc": 0.24705387205387205, "acc_ci": 0.0, "acc_runs": [0.24705387205387205], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008944265906130788, "centered_acc": 0.006734006734006703, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006734006734006703], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2593856655290102, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2593856655290102], "acc": 0.1945392491467577, "acc_ci": 0.0, "acc_runs": [0.1945392491467577], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012808273573927224, "centered_acc": 0.012514220705346949, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012514220705346949], "ci": 0.0}, "openbookqa": {"acc_norm": 0.278, "acc_norm_ci": 0.0, "acc_norm_runs": [0.278], "acc": 0.15, "acc_ci": 0.0, "acc_runs": [0.15], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020055833888070865, "centered_acc": 0.037333333333333364, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037333333333333364], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5122336227308603, "acc_ci": 0.0, "acc_runs": [0.5122336227308603], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014048278820405729, "centered_acc": 0.02446724546172052, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02446724546172052], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 10481.449047254906, "perplexity_ci": 0.0, "perplexity_runs": [10481.449047254906], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.173, "acc_norm_ci": 0.0, "acc_norm_runs": [0.173], "acc": 0.039, "acc_ci": 0.0, "acc_runs": [0.039], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011967214137559927, "centered_acc": -0.10266666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10266666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29177798391727133, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2847077252313137, "mean_acc_ci": 0.0, "mean_centered_acc": -0.014010862320659198, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29177798391727133], "mean_acc_runs": [0.2847077252313137], "mean_centered_acc_runs": [-0.014010862320659198], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-288", "checkpoint": "checkpoint-288", "global_step": 288, "tokens_seen": 301989888, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25801633140808605, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25801633140808605], "acc": 0.2562238597888867, "acc_ci": 0.0, "acc_runs": [0.2562238597888867], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004366488167386302, "centered_acc": 0.01068844187744807, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01068844187744807], "ci": 0.0}, "piqa": {"acc_norm": 0.5070729053318824, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5070729053318824], "acc": 0.528835690968444, "acc_ci": 0.0, "acc_runs": [0.528835690968444], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011664656918146136, "centered_acc": 0.014145810663764857, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014145810663764857], "ci": 0.0}, "arc_easy": {"acc_norm": 0.25252525252525254, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25252525252525254], "acc": 0.25042087542087543, "acc_ci": 0.0, "acc_runs": [0.25042087542087543], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008914948991495836, "centered_acc": 0.0033670033670033885, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0033670033670033885], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26023890784982934, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26023890784982934], "acc": 0.2030716723549488, "acc_ci": 0.0, "acc_runs": [0.2030716723549488], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012821930225112672, "centered_acc": 0.013651877133105783, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013651877133105783], "ci": 0.0}, "openbookqa": {"acc_norm": 0.3, "acc_norm_ci": 0.0, "acc_norm_runs": [0.3], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020514426225627987, "centered_acc": 0.06666666666666665, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06666666666666665], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4909234411996843, "acc_ci": 0.0, "acc_runs": [0.4909234411996843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050170094497657, "centered_acc": -0.01815311760063143, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.01815311760063143], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 184794.37275025138, "perplexity_ci": 0.0, "perplexity_runs": [184794.37275025138], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.177, "acc_norm_ci": 0.0, "acc_norm_runs": [0.177], "acc": 0.048, "acc_ci": 0.0, "acc_runs": [0.048], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012075463420375087, "centered_acc": -0.09733333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09733333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29247556618584175, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28537630015065396, "mean_acc_ci": 0.0, "mean_centered_acc": -0.018932414186508366, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29247556618584175], "mean_acc_runs": [0.28537630015065396], "mean_centered_acc_runs": [-0.018932414186508366], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-576", "checkpoint": "checkpoint-576", "global_step": 576, "tokens_seen": 603979776, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25801633140808605, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25801633140808605], "acc": 0.2566221868153754, "acc_ci": 0.0, "acc_runs": [0.2566221868153754], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004366488167386302, "centered_acc": 0.01068844187744807, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01068844187744807], "ci": 0.0}, "piqa": {"acc_norm": 0.5228509249183896, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5228509249183896], "acc": 0.5326441784548422, "acc_ci": 0.0, "acc_runs": [0.5326441784548422], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011653634832401168, "centered_acc": 0.04570184983677916, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04570184983677916], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26430976430976433, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26430976430976433], "acc": 0.2474747474747475, "acc_ci": 0.0, "acc_runs": [0.2474747474747475], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009048410451862887, "centered_acc": 0.019079685746352437, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019079685746352437], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2687713310580205, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2687713310580205], "acc": 0.21843003412969283, "acc_ci": 0.0, "acc_runs": [0.21843003412969283], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012955065963710852, "centered_acc": 0.025028441410693974, "centered_acc_ci": 0.0, "centered_acc_runs": [0.025028441410693974], "ci": 0.0}, "openbookqa": {"acc_norm": 0.298, "acc_norm_ci": 0.0, "acc_norm_runs": [0.298], "acc": 0.152, "acc_ci": 0.0, "acc_runs": [0.152], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020475118092988954, "centered_acc": 0.06399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06399999999999999], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.47908445146014206, "acc_ci": 0.0, "acc_runs": [0.47908445146014206], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014040185494212999, "centered_acc": -0.041831097079715884, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.041831097079715884], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 1167376.1007467627, "perplexity_ci": 0.0, "perplexity_runs": [1167376.1007467627], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.177, "acc_norm_ci": 0.0, "acc_norm_runs": [0.177], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012075463420375087, "centered_acc": -0.09733333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09733333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29815805861571004, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28885430601085005, "mean_acc_ci": 0.0, "mean_centered_acc": -0.006454565332708338, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29815805861571004], "mean_acc_runs": [0.28885430601085005], "mean_centered_acc_runs": [-0.006454565332708338], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-864", "checkpoint": "checkpoint-864", "global_step": 864, "tokens_seen": 905969664, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2599083847839076, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2599083847839076], "acc": 0.25492929695279826, "acc_ci": 0.0, "acc_runs": [0.25492929695279826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004376877619233649, "centered_acc": 0.013211179711876797, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013211179711876797], "ci": 0.0}, "piqa": {"acc_norm": 0.5184983677910773, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5184983677910773], "acc": 0.5505984766050055, "acc_ci": 0.0, "acc_runs": [0.5505984766050055], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165783758381818, "centered_acc": 0.03699673558215455, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03699673558215455], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2647306397306397, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2647306397306397], "acc": 0.2601010101010101, "acc_ci": 0.0, "acc_runs": [0.2601010101010101], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009053021086174166, "centered_acc": 0.019640852974186274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019640852974186274], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2508532423208191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2508532423208191], "acc": 0.2090443686006826, "acc_ci": 0.0, "acc_runs": [0.2090443686006826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012668198621315347, "centered_acc": 0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.298, "acc_norm_ci": 0.0, "acc_norm_runs": [0.298], "acc": 0.154, "acc_ci": 0.0, "acc_runs": [0.154], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020475118092988954, "centered_acc": 0.06399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06399999999999999], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4877663772691397, "acc_ci": 0.0, "acc_runs": [0.4877663772691397], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01404827882040573, "centered_acc": -0.024467245461720633, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.024467245461720633], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 2060784.008710211, "perplexity_ci": 0.0, "perplexity_runs": [2060784.008710211], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2954984391044073, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28897269913023366, "mean_acc_ci": 0.0, "mean_centered_acc": -0.013878433024260388, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2954984391044073], "mean_acc_runs": [0.28897269913023366], "mean_centered_acc_runs": [-0.013878433024260388], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-1152", "checkpoint": "checkpoint-1152", "global_step": 1152, "tokens_seen": 1207959552, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2593108942441745, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2593108942441745], "acc": 0.2560246962756423, "acc_ci": 0.0, "acc_runs": [0.2560246962756423], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043736082125607105, "centered_acc": 0.012414525658899297, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012414525658899297], "ci": 0.0}, "piqa": {"acc_norm": 0.5179542981501633, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5179542981501633], "acc": 0.5544069640914037, "acc_ci": 0.0, "acc_runs": [0.5544069640914037], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011658300623287121, "centered_acc": 0.0359085963003265, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0359085963003265], "ci": 0.0}, "arc_easy": {"acc_norm": 0.27104377104377103, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27104377104377103], "acc": 0.26515151515151514, "acc_ci": 0.0, "acc_runs": [0.26515151515151514], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009120919741760646, "centered_acc": 0.02805836139169471, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02805836139169471], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26535836177474403, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26535836177474403], "acc": 0.2235494880546075, "acc_ci": 0.0, "acc_runs": [0.2235494880546075], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012902554762313832, "centered_acc": 0.020477815699658713, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020477815699658713], "ci": 0.0}, "openbookqa": {"acc_norm": 0.298, "acc_norm_ci": 0.0, "acc_norm_runs": [0.298], "acc": 0.17, "acc_ci": 0.0, "acc_runs": [0.17], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020475118092988954, "centered_acc": 0.06399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06399999999999999], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5011838989739542, "acc_ci": 0.0, "acc_runs": [0.5011838989739542], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052446290528871, "centered_acc": 0.0023677979479084232, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0023677979479084232], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 2982265.0370979873, "perplexity_ci": 0.0, "perplexity_runs": [2982265.0370979873], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.18, "acc_norm_ci": 0.0, "acc_norm_runs": [0.18], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012155153135512024, "centered_acc": -0.09333333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09333333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2986112208688088, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2973604024320823, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0015034791986050173, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2986112208688088], "mean_acc_runs": [0.2973604024320823], "mean_centered_acc_runs": [-0.0015034791986050173], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-1440", "checkpoint": "checkpoint-1440", "global_step": 1440, "tokens_seen": 1509949440, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26180043815972914, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26180043815972914], "acc": 0.25731925911173076, "acc_ci": 0.0, "acc_runs": [0.25731925911173076], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004387161203088004, "centered_acc": 0.015733917546305525, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015733917546305525], "ci": 0.0}, "piqa": {"acc_norm": 0.5282916213275299, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5282916213275299], "acc": 0.5571273122959739, "acc_ci": 0.0, "acc_runs": [0.5571273122959739], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0116471341727495, "centered_acc": 0.05658324265505987, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05658324265505987], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2748316498316498, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2748316498316498], "acc": 0.2668350168350168, "acc_ci": 0.0, "acc_runs": [0.2668350168350168], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009160538115254864, "centered_acc": 0.03310886644219976, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03310886644219976], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.23720136518771331, "acc_norm_ci": 0.0, "acc_norm_runs": [0.23720136518771331], "acc": 0.21160409556313994, "acc_ci": 0.0, "acc_runs": [0.21160409556313994], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01243039982926071, "centered_acc": -0.017064846416382246, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.017064846416382246], "ci": 0.0}, "openbookqa": {"acc_norm": 0.288, "acc_norm_ci": 0.0, "acc_norm_runs": [0.288], "acc": 0.16, "acc_ci": 0.0, "acc_runs": [0.16], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020271503835075255, "centered_acc": 0.05066666666666664, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05066666666666664], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4996053670086819, "acc_ci": 0.0, "acc_runs": [0.4996053670086819], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405248130604961, "centered_acc": -0.0007892659826361781, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0007892659826361781], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 5011021.250640561, "perplexity_ci": 0.0, "perplexity_runs": [5011021.250640561], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.051, "acc_ci": 0.0, "acc_runs": [0.051], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29552084575110377, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2951778512588244, "mean_acc_ci": 0.0, "mean_centered_acc": -0.006057758763857714, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29552084575110377], "mean_acc_runs": [0.2951778512588244], "mean_centered_acc_runs": [-0.006057758763857714], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-1728", "checkpoint": "checkpoint-1728", "global_step": 1728, "tokens_seen": 1811939328, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25721967735510853, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25721967735510853], "acc": 0.2546305516829317, "acc_ci": 0.0, "acc_runs": [0.2546305516829317], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004362081806560604, "centered_acc": 0.009626236473478045, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009626236473478045], "ci": 0.0}, "piqa": {"acc_norm": 0.5337323177366703, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5337323177366703], "acc": 0.5489662676822633, "acc_ci": 0.0, "acc_runs": [0.5489662676822633], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01163924552241405, "centered_acc": 0.06746463547334058, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06746463547334058], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2777777777777778, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2777777777777778], "acc": 0.2638888888888889, "acc_ci": 0.0, "acc_runs": [0.2638888888888889], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009190779909650122, "centered_acc": 0.037037037037037056, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037037037037037056], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25341296928327645, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25341296928327645], "acc": 0.21245733788395904, "acc_ci": 0.0, "acc_runs": [0.21245733788395904], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012710896778378677, "centered_acc": 0.004550625711035261, "centered_acc_ci": 0.0, "centered_acc_runs": [0.004550625711035261], "ci": 0.0}, "openbookqa": {"acc_norm": 0.292, "acc_norm_ci": 0.0, "acc_norm_runs": [0.292], "acc": 0.164, "acc_ci": 0.0, "acc_runs": [0.164], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020354375480530096, "centered_acc": 0.05599999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05599999999999997], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5122336227308603, "acc_ci": 0.0, "acc_runs": [0.5122336227308603], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014048278820405729, "centered_acc": 0.02446724546172052, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02446724546172052], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.56, "acc_ci": 0.0, "acc_runs": [0.56], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049888765156985884, "centered_acc": 0.1200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1200000000000001], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 4497233.641430045, "perplexity_ci": 0.0, "perplexity_runs": [4497233.641430045], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.189, "acc_norm_ci": 0.0, "acc_norm_runs": [0.189], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01238678458811779, "centered_acc": -0.08133333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08133333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.30052379035880555, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29394641306426034, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0006236255143690161, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.30052379035880555], "mean_acc_runs": [0.29394641306426034], "mean_centered_acc_runs": [-0.0006236255143690161], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2016", "checkpoint": "checkpoint-2016", "global_step": 2016, "tokens_seen": 2113929216, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2593108942441745, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2593108942441745], "acc": 0.255327623979287, "acc_ci": 0.0, "acc_runs": [0.255327623979287], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043736082125607105, "centered_acc": 0.012414525658899297, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012414525658899297], "ci": 0.0}, "piqa": {"acc_norm": 0.5326441784548422, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5326441784548422], "acc": 0.5495103373231773, "acc_ci": 0.0, "acc_runs": [0.5495103373231773], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011640934420031744, "centered_acc": 0.06528835690968449, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06528835690968449], "ci": 0.0}, "arc_easy": {"acc_norm": 0.27146464646464646, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27146464646464646], "acc": 0.25673400673400676, "acc_ci": 0.0, "acc_runs": [0.25673400673400676], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00912536297036068, "centered_acc": 0.028619528619528618, "centered_acc_ci": 0.0, "centered_acc_runs": [0.028619528619528618], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2627986348122867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2627986348122867], "acc": 0.2175767918088737, "acc_ci": 0.0, "acc_runs": [0.2175767918088737], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012862523175351397, "centered_acc": 0.017064846416382284, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017064846416382284], "ci": 0.0}, "openbookqa": {"acc_norm": 0.298, "acc_norm_ci": 0.0, "acc_norm_runs": [0.298], "acc": 0.16, "acc_ci": 0.0, "acc_runs": [0.16], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020475118092988954, "centered_acc": 0.06399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06399999999999999], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5067087608524072, "acc_ci": 0.0, "acc_runs": [0.5067087608524072], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051220692330247, "centered_acc": 0.013417521704814472, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013417521704814472], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.49, "acc_ci": 0.0, "acc_runs": [0.49], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05024183937956913, "centered_acc": -0.020000000000000018, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.020000000000000018], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 6360126.962215008, "perplexity_ci": 0.0, "perplexity_runs": [6360126.962215008], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.30120305899599165, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2859144982471452, "mean_acc_ci": 0.0, "mean_centered_acc": -0.016883736719624842, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.30120305899599165], "mean_acc_runs": [0.2859144982471452], "mean_centered_acc_runs": [-0.016883736719624842], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2304", "checkpoint": "checkpoint-2304", "global_step": 2304, "tokens_seen": 2415919104, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2597092212706632, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2597092212706632], "acc": 0.2560246962756423, "acc_ci": 0.0, "acc_runs": [0.2560246962756423], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004375788991216637, "centered_acc": 0.012945628360884273, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012945628360884273], "ci": 0.0}, "piqa": {"acc_norm": 0.5184983677910773, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5184983677910773], "acc": 0.5538628944504896, "acc_ci": 0.0, "acc_runs": [0.5538628944504896], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165783758381818, "centered_acc": 0.03699673558215455, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03699673558215455], "ci": 0.0}, "arc_easy": {"acc_norm": 0.27735690235690236, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27735690235690236], "acc": 0.25925925925925924, "acc_ci": 0.0, "acc_runs": [0.25925925925925924], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009186490105111987, "centered_acc": 0.03647586980920314, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03647586980920314], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.27047781569965873, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27047781569965873], "acc": 0.21416382252559726, "acc_ci": 0.0, "acc_runs": [0.21416382252559726], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012980954547659553, "centered_acc": 0.02730375426621164, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02730375426621164], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.500394632991318, "acc_ci": 0.0, "acc_runs": [0.500394632991318], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405248130604961, "centered_acc": 0.0007892659826360671, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0007892659826360671], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5, "acc_ci": 0.0, "acc_runs": [0.5], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.050251890762960605, "centered_acc": 0.0, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 5577220.558768843, "perplexity_ci": 0.0, "perplexity_runs": [5577220.558768843], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.186, "acc_norm_ci": 0.0, "acc_norm_runs": [0.186], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012310790208412926, "centered_acc": -0.08533333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08533333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2990070511863836, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2862992767276007, "mean_acc_ci": 0.0, "mean_centered_acc": -0.01906449879090849, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2990070511863836], "mean_acc_runs": [0.2862992767276007], "mean_centered_acc_runs": [-0.01906449879090849], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2592", "checkpoint": "checkpoint-2592", "global_step": 2592, "tokens_seen": 2717908992, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2613025293766182, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2613025293766182], "acc": 0.25473013343955386, "acc_ci": 0.0, "acc_runs": [0.25473013343955386], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043844652190712236, "centered_acc": 0.015070039168824248, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015070039168824248], "ci": 0.0}, "piqa": {"acc_norm": 0.5228509249183896, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5228509249183896], "acc": 0.5462459194776932, "acc_ci": 0.0, "acc_runs": [0.5462459194776932], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011653634832401168, "centered_acc": 0.04570184983677916, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04570184983677916], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2748316498316498, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2748316498316498], "acc": 0.2537878787878788, "acc_ci": 0.0, "acc_runs": [0.2537878787878788], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009160538115254864, "centered_acc": 0.03310886644219976, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03310886644219976], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2627986348122867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2627986348122867], "acc": 0.2098976109215017, "acc_ci": 0.0, "acc_runs": [0.2098976109215017], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012862523175351397, "centered_acc": 0.017064846416382284, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017064846416382284], "ci": 0.0}, "openbookqa": {"acc_norm": 0.292, "acc_norm_ci": 0.0, "acc_norm_runs": [0.292], "acc": 0.168, "acc_ci": 0.0, "acc_runs": [0.168], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020354375480530096, "centered_acc": 0.05599999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05599999999999997], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4972375690607735, "acc_ci": 0.0, "acc_runs": [0.4972375690607735], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052271211616401, "centered_acc": -0.0055248618784530246, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0055248618784530246], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.52, "acc_ci": 0.0, "acc_runs": [0.52], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05021167315686783, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 7382693.344635282, "perplexity_ci": 0.0, "perplexity_runs": [7382693.344635282], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.191], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012436787112179482, "centered_acc": -0.07866666666666666, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07866666666666666], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.3007972898231574, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2873186573461101, "mean_acc_ci": 0.0, "mean_centered_acc": -0.013407889237059288, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.3007972898231574], "mean_acc_runs": [0.2873186573461101], "mean_centered_acc_runs": [-0.013407889237059288], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-2880", "checkpoint": "checkpoint-2880", "global_step": 2880, "tokens_seen": 3019898880, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25781716789484166, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25781716789484166], "acc": 0.2561242780322645, "acc_ci": 0.0, "acc_runs": [0.2561242780322645], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043653883515635365, "centered_acc": 0.010422890526455545, "centered_acc_ci": 0.0, "centered_acc_runs": [0.010422890526455545], "ci": 0.0}, "piqa": {"acc_norm": 0.5190424374319913, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5190424374319913], "acc": 0.544613710554951, "acc_ci": 0.0, "acc_runs": [0.544613710554951], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011657360703051524, "centered_acc": 0.0380848748639826, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0380848748639826], "ci": 0.0}, "arc_easy": {"acc_norm": 0.265993265993266, "acc_norm_ci": 0.0, "acc_norm_runs": [0.265993265993266], "acc": 0.24621212121212122, "acc_ci": 0.0, "acc_runs": [0.24621212121212122], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009066789565615746, "centered_acc": 0.021324354657688005, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021324354657688005], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26109215017064846, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26109215017064846], "acc": 0.2098976109215017, "acc_ci": 0.0, "acc_runs": [0.2098976109215017], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012835523909473708, "centered_acc": 0.014789533560864617, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014789533560864617], "ci": 0.0}, "openbookqa": {"acc_norm": 0.27, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27], "acc": 0.162, "acc_ci": 0.0, "acc_runs": [0.162], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019874354831287473, "centered_acc": 0.02666666666666669, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02666666666666669], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4940805051302289, "acc_ci": 0.0, "acc_runs": [0.4940805051302289], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051500838485925, "centered_acc": -0.011838989739542227, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.011838989739542227], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 6018945.46504603, "perplexity_ci": 0.0, "perplexity_runs": [6018945.46504603], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.187, "acc_norm_ci": 0.0, "acc_norm_runs": [0.187], "acc": 0.042, "acc_ci": 0.0, "acc_runs": [0.042], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012336254828074166, "centered_acc": -0.084, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.084], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2934908369151246, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2873215687624767, "mean_acc_ci": 0.0, "mean_centered_acc": -0.016441749546276006, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2934908369151246], "mean_acc_runs": [0.2873215687624767], "mean_centered_acc_runs": [-0.016441749546276006], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-3168", "checkpoint": "checkpoint-3168", "global_step": 3168, "tokens_seen": 3321888768, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2588129854610635, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2588129854610635], "acc": 0.25712009559848636, "acc_ci": 0.0, "acc_runs": [0.25712009559848636], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004370875625258676, "centered_acc": 0.011750647281418022, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011750647281418022], "ci": 0.0}, "piqa": {"acc_norm": 0.5266594124047879, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5266594124047879], "acc": 0.5462459194776932, "acc_ci": 0.0, "acc_runs": [0.5462459194776932], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01164922999434743, "centered_acc": 0.05331882480957573, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05331882480957573], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2718855218855219, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2718855218855219], "acc": 0.25252525252525254, "acc_ci": 0.0, "acc_runs": [0.25252525252525254], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009129795867310605, "centered_acc": 0.02918069584736253, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02918069584736253], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2508532423208191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2508532423208191], "acc": 0.21928327645051193, "acc_ci": 0.0, "acc_runs": [0.21928327645051193], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012668198621315347, "centered_acc": 0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.162, "acc_ci": 0.0, "acc_runs": [0.162], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5059194948697711, "acc_ci": 0.0, "acc_runs": [0.5059194948697711], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051500838485923, "centered_acc": 0.011838989739542116, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011838989739542116], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.54, "acc_ci": 0.0, "acc_runs": [0.54], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05009082659620331, "centered_acc": 0.08000000000000007, "centered_acc_ci": 0.0, "centered_acc_runs": [0.08000000000000007], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 11618776.604118865, "perplexity_ci": 0.0, "perplexity_runs": [11618776.604118865], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.191, "acc_norm_ci": 0.0, "acc_norm_runs": [0.191], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012436787112179482, "centered_acc": -0.07866666666666666, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07866666666666666], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29653519367869874, "mean_acc_norm_ci": 0.0, "mean_acc": 0.290557556240704, "mean_acc_ci": 0.0, "mean_centered_acc": -0.010540547668178744, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29653519367869874], "mean_acc_runs": [0.290557556240704], "mean_centered_acc_runs": [-0.010540547668178744], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-3456", "checkpoint": "checkpoint-3456", "global_step": 3456, "tokens_seen": 3623878656, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25871340370444135, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25871340370444135], "acc": 0.25562636924915355, "acc_ci": 0.0, "acc_runs": [0.25562636924915355], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004370328224831643, "centered_acc": 0.011617871605921795, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011617871605921795], "ci": 0.0}, "piqa": {"acc_norm": 0.5261153427638737, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5261153427638737], "acc": 0.5408052230685527, "acc_ci": 0.0, "acc_runs": [0.5408052230685527], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01164990085426341, "centered_acc": 0.052230685527747456, "centered_acc_ci": 0.0, "centered_acc_runs": [0.052230685527747456], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2638888888888889, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2638888888888889], "acc": 0.25462962962962965, "acc_ci": 0.0, "acc_runs": [0.25462962962962965], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009043789220055392, "centered_acc": 0.018518518518518528, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018518518518518528], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.257679180887372, "acc_norm_ci": 0.0, "acc_norm_runs": [0.257679180887372], "acc": 0.21245733788395904, "acc_ci": 0.0, "acc_runs": [0.21245733788395904], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012780770562768334, "centered_acc": 0.010238907849829356, "centered_acc_ci": 0.0, "centered_acc_runs": [0.010238907849829356], "ci": 0.0}, "openbookqa": {"acc_norm": 0.282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.282], "acc": 0.166, "acc_ci": 0.0, "acc_runs": [0.166], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020143572847290726, "centered_acc": 0.04266666666666663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04266666666666663], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5146014206787688, "acc_ci": 0.0, "acc_runs": [0.5146014206787688], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014046492383275955, "centered_acc": 0.02920284135753759, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02920284135753759], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 6122193.9436408905, "perplexity_ci": 0.0, "perplexity_runs": [6122193.9436408905], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.19, "acc_norm_ci": 0.0, "acc_norm_runs": [0.19], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012411851354816256, "centered_acc": -0.08, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29639946937409595, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29164074422837644, "mean_acc_ci": 0.0, "mean_centered_acc": -0.006549953880708658, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29639946937409595], "mean_acc_runs": [0.29164074422837644], "mean_centered_acc_runs": [-0.006549953880708658], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-3744", "checkpoint": "checkpoint-3744", "global_step": 3744, "tokens_seen": 3925868544, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26020713005377416, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26020713005377416], "acc": 0.25791674965146383, "acc_ci": 0.0, "acc_runs": [0.25791674965146383], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004378508362084079, "centered_acc": 0.013609506738365548, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013609506738365548], "ci": 0.0}, "piqa": {"acc_norm": 0.5244831338411317, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5244831338411317], "acc": 0.5429815016322089, "acc_ci": 0.0, "acc_runs": [0.5429815016322089], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011651830225710194, "centered_acc": 0.04896626768226331, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04896626768226331], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2601010101010101, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2601010101010101], "acc": 0.25084175084175087, "acc_ci": 0.0, "acc_runs": [0.25084175084175087], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009001718541079735, "centered_acc": 0.013468013468013481, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013468013468013481], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25341296928327645, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25341296928327645], "acc": 0.22013651877133106, "acc_ci": 0.0, "acc_runs": [0.22013651877133106], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012710896778378677, "centered_acc": 0.004550625711035261, "centered_acc_ci": 0.0, "centered_acc_runs": [0.004550625711035261], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.178, "acc_ci": 0.0, "acc_runs": [0.178], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5146014206787688, "acc_ci": 0.0, "acc_runs": [0.5146014206787688], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014046492383275955, "centered_acc": 0.02920284135753759, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02920284135753759], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.51, "acc_ci": 0.0, "acc_runs": [0.51], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05024183937956913, "centered_acc": 0.020000000000000018, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020000000000000018], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 6756597.327496335, "perplexity_ci": 0.0, "perplexity_runs": [6756597.327496335], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.194, "acc_norm_ci": 0.0, "acc_norm_runs": [0.194], "acc": 0.051, "acc_ci": 0.0, "acc_runs": [0.051], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012510816141264348, "centered_acc": -0.07466666666666666, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07466666666666666], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2963673738798654, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29037654033492244, "mean_acc_ci": 0.0, "mean_centered_acc": -0.015588276462450097, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2963673738798654], "mean_acc_runs": [0.29037654033492244], "mean_centered_acc_runs": [-0.015588276462450097], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4032", "checkpoint": "checkpoint-4032", "global_step": 4032, "tokens_seen": 4227858432, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25761800438159727, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25761800438159727], "acc": 0.2581159131647082, "acc_ci": 0.0, "acc_runs": [0.2581159131647082], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004364287353415334, "centered_acc": 0.01015733917546302, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01015733917546302], "ci": 0.0}, "piqa": {"acc_norm": 0.5163220892274211, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5163220892274211], "acc": 0.5413492927094669, "acc_ci": 0.0, "acc_runs": [0.5413492927094669], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011659606710151597, "centered_acc": 0.03264417845484213, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03264417845484213], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26346801346801346, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26346801346801346], "acc": 0.2563131313131313, "acc_ci": 0.0, "acc_runs": [0.2563131313131313], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009039157374497906, "centered_acc": 0.017957351290684615, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017957351290684615], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2645051194539249, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2645051194539249], "acc": 0.21331058020477817, "acc_ci": 0.0, "acc_runs": [0.21331058020477817], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012889272949313479, "centered_acc": 0.019340159271899877, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019340159271899877], "ci": 0.0}, "openbookqa": {"acc_norm": 0.278, "acc_norm_ci": 0.0, "acc_norm_runs": [0.278], "acc": 0.166, "acc_ci": 0.0, "acc_runs": [0.166], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020055833888070865, "centered_acc": 0.037333333333333364, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037333333333333364], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5146014206787688, "acc_ci": 0.0, "acc_runs": [0.5146014206787688], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014046492383275955, "centered_acc": 0.02920284135753759, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02920284135753759], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5, "acc_ci": 0.0, "acc_runs": [0.5], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.050251890762960605, "centered_acc": 0.0, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 5537756.627233293, "perplexity_ci": 0.0, "perplexity_runs": [5537756.627233293], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.192, "acc_norm_ci": 0.0, "acc_norm_runs": [0.192], "acc": 0.048, "acc_ci": 0.0, "acc_runs": [0.048], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01246159264666003, "centered_acc": -0.07733333333333332, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.07733333333333332], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2953188710884928, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2876171861556178, "mean_acc_ci": 0.0, "mean_centered_acc": -0.01934702298913023, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2953188710884928], "mean_acc_runs": [0.2876171861556178], "mean_centered_acc_runs": [-0.01934702298913023], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4320", "checkpoint": "checkpoint-4320", "global_step": 4320, "tokens_seen": 4529848320, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26160127464648475, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26160127464648475], "acc": 0.25672176857199763, "acc_ci": 0.0, "acc_runs": [0.25672176857199763], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004386083683839967, "centered_acc": 0.015468366195313, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015468366195313], "ci": 0.0}, "piqa": {"acc_norm": 0.5190424374319913, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5190424374319913], "acc": 0.5408052230685527, "acc_ci": 0.0, "acc_runs": [0.5408052230685527], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011657360703051524, "centered_acc": 0.0380848748639826, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0380848748639826], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2676767676767677, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2676767676767677], "acc": 0.2516835016835017, "acc_ci": 0.0, "acc_runs": [0.2516835016835017], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009085000147099275, "centered_acc": 0.023569023569023573, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023569023569023573], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26621160409556316, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26621160409556316], "acc": 0.2158703071672355, "acc_ci": 0.0, "acc_runs": [0.2158703071672355], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012915774781523113, "centered_acc": 0.021615472127417545, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021615472127417545], "ci": 0.0}, "openbookqa": {"acc_norm": 0.27, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27], "acc": 0.166, "acc_ci": 0.0, "acc_runs": [0.166], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019874354831287473, "centered_acc": 0.02666666666666669, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02666666666666669], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5209155485398579, "acc_ci": 0.0, "acc_runs": [0.5209155485398579], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014040185494212997, "centered_acc": 0.04183109707971577, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04183109707971577], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.53, "acc_ci": 0.0, "acc_runs": [0.53], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05016135580465919, "centered_acc": 0.06000000000000005, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06000000000000005], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 10874466.034167549, "perplexity_ci": 0.0, "perplexity_runs": [10874466.034167549], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.19, "acc_norm_ci": 0.0, "acc_norm_runs": [0.19], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012411851354816256, "centered_acc": -0.08, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2957553473084678, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2905283810804845, "mean_acc_ci": 0.0, "mean_centered_acc": -0.01068773066116446, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2957553473084678], "mean_acc_runs": [0.2905283810804845], "mean_centered_acc_runs": [-0.01068773066116446], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4608", "checkpoint": "checkpoint-4608", "global_step": 4608, "tokens_seen": 4831838208, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.261700856403107, "acc_norm_ci": 0.0, "acc_norm_runs": [0.261700856403107], "acc": 0.2582154949213304, "acc_ci": 0.0, "acc_runs": [0.2582154949213304], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004386622589118801, "centered_acc": 0.015601141870809299, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015601141870809299], "ci": 0.0}, "piqa": {"acc_norm": 0.5070729053318824, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5070729053318824], "acc": 0.5277475516866159, "acc_ci": 0.0, "acc_runs": [0.5277475516866159], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011664656918146136, "centered_acc": 0.014145810663764857, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014145810663764857], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2601010101010101, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2601010101010101], "acc": 0.24957912457912457, "acc_ci": 0.0, "acc_runs": [0.24957912457912457], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009001718541079735, "centered_acc": 0.013468013468013481, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013468013468013481], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26791808873720135, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26791808873720135], "acc": 0.21843003412969283, "acc_ci": 0.0, "acc_runs": [0.21843003412969283], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012942030195136482, "centered_acc": 0.023890784982935138, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023890784982935138], "ci": 0.0}, "openbookqa": {"acc_norm": 0.272, "acc_norm_ci": 0.0, "acc_norm_runs": [0.272], "acc": 0.164, "acc_ci": 0.0, "acc_runs": [0.164], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019920483209566107, "centered_acc": 0.02933333333333336, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02933333333333336], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5067087608524072, "acc_ci": 0.0, "acc_runs": [0.5067087608524072], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051220692330247, "centered_acc": 0.013417521704814472, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013417521704814472], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.56, "acc_ci": 0.0, "acc_runs": [0.56], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049888765156985884, "centered_acc": 0.1200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1200000000000001], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 9808288.695865558, "perplexity_ci": 0.0, "perplexity_runs": [9808288.695865558], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.189, "acc_norm_ci": 0.0, "acc_norm_runs": [0.189], "acc": 0.048, "acc_ci": 0.0, "acc_runs": [0.048], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01238678458811779, "centered_acc": -0.08133333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08133333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2929654767622002, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29109684279428716, "mean_acc_ci": 0.0, "mean_centered_acc": -0.010544644862473553, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2929654767622002], "mean_acc_runs": [0.29109684279428716], "mean_centered_acc_runs": [-0.010544644862473553], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-4896", "checkpoint": "checkpoint-4896", "global_step": 4896, "tokens_seen": 5133828096, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26010754829715194, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26010754829715194], "acc": 0.2590121489743079, "acc_ci": 0.0, "acc_runs": [0.2590121489743079], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004377965074211591, "centered_acc": 0.013476731062869248, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013476731062869248], "ci": 0.0}, "piqa": {"acc_norm": 0.5277475516866159, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5277475516866159], "acc": 0.5331882480957563, "acc_ci": 0.0, "acc_runs": [0.5331882480957563], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011647846656062001, "centered_acc": 0.055495103373231824, "centered_acc_ci": 0.0, "centered_acc_runs": [0.055495103373231824], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26304713804713803, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26304713804713803], "acc": 0.25462962962962965, "acc_ci": 0.0, "acc_runs": [0.25462962962962965], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009034514898865944, "centered_acc": 0.017396184062850706, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017396184062850706], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26023890784982934, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26023890784982934], "acc": 0.2175767918088737, "acc_ci": 0.0, "acc_runs": [0.2175767918088737], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012821930225112672, "centered_acc": 0.013651877133105783, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013651877133105783], "ci": 0.0}, "openbookqa": {"acc_norm": 0.278, "acc_norm_ci": 0.0, "acc_norm_runs": [0.278], "acc": 0.162, "acc_ci": 0.0, "acc_runs": [0.162], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020055833888070865, "centered_acc": 0.037333333333333364, "centered_acc_ci": 0.0, "centered_acc_runs": [0.037333333333333364], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5122336227308603, "acc_ci": 0.0, "acc_runs": [0.5122336227308603], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014048278820405729, "centered_acc": 0.02446724546172052, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02446724546172052], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.52, "acc_ci": 0.0, "acc_runs": [0.52], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05021167315686783, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 13003877.95079824, "perplexity_ci": 0.0, "perplexity_runs": [13003877.95079824], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.19, "acc_norm_ci": 0.0, "acc_norm_runs": [0.19], "acc": 0.051, "acc_ci": 0.0, "acc_runs": [0.051], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012411851354816256, "centered_acc": -0.08, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29652352431345586, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2887927903013129, "mean_acc_ci": 0.0, "mean_centered_acc": -0.013511622447276432, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29652352431345586], "mean_acc_runs": [0.2887927903013129], "mean_centered_acc_runs": [-0.013511622447276432], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-5184", "checkpoint": "checkpoint-5184", "global_step": 5184, "tokens_seen": 5435817984, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25562636924915355, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25562636924915355], "acc": 0.2604062935670185, "acc_ci": 0.0, "acc_runs": [0.2604062935670185], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004353212146198103, "centered_acc": 0.007501825665538068, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007501825665538068], "ci": 0.0}, "piqa": {"acc_norm": 0.5136017410228509, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5136017410228509], "acc": 0.530467899891186, "acc_ci": 0.0, "acc_runs": [0.530467899891186], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011661506839823792, "centered_acc": 0.027203482045701888, "centered_acc_ci": 0.0, "centered_acc_runs": [0.027203482045701888], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26557239057239057, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26557239057239057], "acc": 0.2588383838383838, "acc_ci": 0.0, "acc_runs": [0.2588383838383838], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00906221062697191, "centered_acc": 0.020763187429854096, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020763187429854096], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25853242320819114, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25853242320819114], "acc": 0.2158703071672355, "acc_ci": 0.0, "acc_runs": [0.2158703071672355], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012794553754288777, "centered_acc": 0.011376564277588189, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011376564277588189], "ci": 0.0}, "openbookqa": {"acc_norm": 0.266, "acc_norm_ci": 0.0, "acc_norm_runs": [0.266], "acc": 0.16, "acc_ci": 0.0, "acc_runs": [0.16], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019780559675655396, "centered_acc": 0.021333333333333353, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021333333333333353], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5153906866614049, "acc_ci": 0.0, "acc_runs": [0.5153906866614049], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014045826789783585, "centered_acc": 0.030781373322809724, "centered_acc_ci": 0.0, "centered_acc_runs": [0.030781373322809724], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5, "acc_ci": 0.0, "acc_runs": [0.5], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.050251890762960605, "centered_acc": 0.0, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 16251763.125292294, "perplexity_ci": 0.0, "perplexity_runs": [16251763.125292294], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29038882067543104, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2866455094610553, "mean_acc_ci": 0.0, "mean_centered_acc": -0.023755404856789706, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29038882067543104], "mean_acc_runs": [0.2866455094610553], "mean_centered_acc_runs": [-0.023755404856789706], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-5472", "checkpoint": "checkpoint-5472", "global_step": 5472, "tokens_seen": 5737807872, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2597092212706632, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2597092212706632], "acc": 0.2584146584345748, "acc_ci": 0.0, "acc_runs": [0.2584146584345748], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004375788991216637, "centered_acc": 0.012945628360884273, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012945628360884273], "ci": 0.0}, "piqa": {"acc_norm": 0.5179542981501633, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5179542981501633], "acc": 0.5217627856365615, "acc_ci": 0.0, "acc_runs": [0.5217627856365615], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011658300623287121, "centered_acc": 0.0359085963003265, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0359085963003265], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2676767676767677, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2676767676767677], "acc": 0.24915824915824916, "acc_ci": 0.0, "acc_runs": [0.24915824915824916], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009085000147099275, "centered_acc": 0.023569023569023573, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023569023569023573], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.27047781569965873, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27047781569965873], "acc": 0.21331058020477817, "acc_ci": 0.0, "acc_runs": [0.21331058020477817], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012980954547659553, "centered_acc": 0.02730375426621164, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02730375426621164], "ci": 0.0}, "openbookqa": {"acc_norm": 0.266, "acc_norm_ci": 0.0, "acc_norm_runs": [0.266], "acc": 0.164, "acc_ci": 0.0, "acc_runs": [0.164], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019780559675655396, "centered_acc": 0.021333333333333353, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021333333333333353], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5138121546961326, "acc_ci": 0.0, "acc_runs": [0.5138121546961326], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014047122916440271, "centered_acc": 0.027624309392265234, "centered_acc_ci": 0.0, "centered_acc_runs": [0.027624309392265234], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.53, "acc_ci": 0.0, "acc_runs": [0.53], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05016135580465919, "centered_acc": 0.06000000000000005, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06000000000000005], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 17905072.61922125, "perplexity_ci": 0.0, "perplexity_runs": [17905072.61922125], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2938030171328755, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2875939951615621, "mean_acc_ci": 0.0, "mean_centered_acc": -0.014082270136728303, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2938030171328755], "mean_acc_runs": [0.2875939951615621], "mean_centered_acc_runs": [-0.014082270136728303], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-5760", "checkpoint": "checkpoint-5760", "global_step": 5760, "tokens_seen": 6039797760, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2584146584345748, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2584146584345748], "acc": 0.2559251145190201, "acc_ci": 0.0, "acc_runs": [0.2559251145190201], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043686842556263684, "centered_acc": 0.011219544579433046, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011219544579433046], "ci": 0.0}, "piqa": {"acc_norm": 0.499455930359086, "acc_norm_ci": 0.0, "acc_norm_runs": [0.499455930359086], "acc": 0.5261153427638737, "acc_ci": 0.0, "acc_runs": [0.5261153427638737], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665817258899291, "centered_acc": -0.0010881392818280489, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0010881392818280489], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26262626262626265, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26262626262626265], "acc": 0.25126262626262624, "acc_ci": 0.0, "acc_runs": [0.25126262626262624], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009029861776763957, "centered_acc": 0.01683501683501687, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01683501683501687], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2636518771331058, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2636518771331058], "acc": 0.2158703071672355, "acc_ci": 0.0, "acc_runs": [0.2158703071672355], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012875929151296968, "centered_acc": 0.018202502844141044, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018202502844141044], "ci": 0.0}, "openbookqa": {"acc_norm": 0.268, "acc_norm_ci": 0.0, "acc_norm_runs": [0.268], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019827714859587477, "centered_acc": 0.02400000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02400000000000002], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5122336227308603, "acc_ci": 0.0, "acc_runs": [0.5122336227308603], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014048278820405729, "centered_acc": 0.02446724546172052, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02446724546172052], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5, "acc_ci": 0.0, "acc_runs": [0.5], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.050251890762960605, "centered_acc": 0.0, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 15540285.037326762, "perplexity_ci": 0.0, "perplexity_runs": [15540285.037326762], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.182, "acc_norm_ci": 0.0, "acc_norm_runs": [0.182], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0122075806376622, "centered_acc": -0.09066666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09066666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2890247880921715, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2842694475217316, "mean_acc_ci": 0.0, "mean_centered_acc": -0.026717285853420286, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2890247880921715], "mean_acc_runs": [0.2842694475217316], "mean_centered_acc_runs": [-0.026717285853420286], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6048", "checkpoint": "checkpoint-6048", "global_step": 6048, "tokens_seen": 6341787648, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2606054570802629, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2606054570802629], "acc": 0.2566221868153754, "acc_ci": 0.0, "acc_runs": [0.2566221868153754], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004380678585341082, "centered_acc": 0.014140609440350524, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014140609440350524], "ci": 0.0}, "piqa": {"acc_norm": 0.499455930359086, "acc_norm_ci": 0.0, "acc_norm_runs": [0.499455930359086], "acc": 0.5331882480957563, "acc_ci": 0.0, "acc_runs": [0.5331882480957563], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665817258899291, "centered_acc": -0.0010881392818280489, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0010881392818280489], "ci": 0.0}, "arc_easy": {"acc_norm": 0.27146464646464646, "acc_norm_ci": 0.0, "acc_norm_runs": [0.27146464646464646], "acc": 0.25547138047138046, "acc_ci": 0.0, "acc_runs": [0.25547138047138046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00912536297036068, "centered_acc": 0.028619528619528618, "centered_acc_ci": 0.0, "centered_acc_runs": [0.028619528619528618], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2636518771331058, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2636518771331058], "acc": 0.2167235494880546, "acc_ci": 0.0, "acc_runs": [0.2167235494880546], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012875929151296968, "centered_acc": 0.018202502844141044, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018202502844141044], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.17, "acc_ci": 0.0, "acc_runs": [0.17], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5067087608524072, "acc_ci": 0.0, "acc_runs": [0.5067087608524072], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051220692330247, "centered_acc": 0.013417521704814472, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013417521704814472], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.51, "acc_ci": 0.0, "acc_runs": [0.51], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05024183937956913, "centered_acc": 0.020000000000000018, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020000000000000018], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 11857136.932694603, "perplexity_ci": 0.0, "perplexity_runs": [11857136.932694603], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.185, "acc_norm_ci": 0.0, "acc_norm_runs": [0.185], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012285191326386655, "centered_acc": -0.08666666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08666666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2923629851728502, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2873195649208299, "mean_acc_ci": 0.0, "mean_centered_acc": -0.022755524421362153, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2923629851728502], "mean_acc_runs": [0.2873195649208299], "mean_centered_acc_runs": [-0.022755524421362153], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6336", "checkpoint": "checkpoint-6336", "global_step": 6336, "tokens_seen": 6643777536, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2636924915355507, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2636924915355507], "acc": 0.2577175861382195, "acc_ci": 0.0, "acc_runs": [0.2577175861382195], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004397339661695592, "centered_acc": 0.01825665538073425, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01825665538073425], "ci": 0.0}, "piqa": {"acc_norm": 0.5048966267682263, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5048966267682263], "acc": 0.5261153427638737, "acc_ci": 0.0, "acc_runs": [0.5261153427638737], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665264730078137, "centered_acc": 0.009793253536452662, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009793253536452662], "ci": 0.0}, "arc_easy": {"acc_norm": 0.265993265993266, "acc_norm_ci": 0.0, "acc_norm_runs": [0.265993265993266], "acc": 0.2558922558922559, "acc_ci": 0.0, "acc_runs": [0.2558922558922559], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009066789565615746, "centered_acc": 0.021324354657688005, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021324354657688005], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2738907849829352, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2738907849829352], "acc": 0.21416382252559726, "acc_ci": 0.0, "acc_runs": [0.21416382252559726], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.013032004972989623, "centered_acc": 0.031854379977246904, "centered_acc_ci": 0.0, "centered_acc_runs": [0.031854379977246904], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.164, "acc_ci": 0.0, "acc_runs": [0.164], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.510655090765588, "acc_ci": 0.0, "acc_runs": [0.510655090765588], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014049294536290372, "centered_acc": 0.02131018153117603, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02131018153117603], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.49, "acc_ci": 0.0, "acc_runs": [0.49], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05024183937956913, "centered_acc": -0.020000000000000018, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.020000000000000018], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 20496188.555832285, "perplexity_ci": 0.0, "perplexity_runs": [20496188.555832285], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.184, "acc_norm_ci": 0.0, "acc_norm_runs": [0.184], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01225945734093864, "centered_acc": -0.08800000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08800000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29441219487999637, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28438315598592345, "mean_acc_ci": 0.0, "mean_centered_acc": -0.024098472374366835, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29441219487999637], "mean_acc_runs": [0.28438315598592345], "mean_centered_acc_runs": [-0.024098472374366835], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6624", "checkpoint": "checkpoint-6624", "global_step": 6624, "tokens_seen": 6945767424, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2599083847839076, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2599083847839076], "acc": 0.2577175861382195, "acc_ci": 0.0, "acc_runs": [0.2577175861382195], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004376877619233649, "centered_acc": 0.013211179711876797, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013211179711876797], "ci": 0.0}, "piqa": {"acc_norm": 0.5136017410228509, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5136017410228509], "acc": 0.5315560391730142, "acc_ci": 0.0, "acc_runs": [0.5315560391730142], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011661506839823792, "centered_acc": 0.027203482045701888, "centered_acc_ci": 0.0, "centered_acc_runs": [0.027203482045701888], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26851851851851855, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26851851851851855], "acc": 0.2622053872053872, "acc_ci": 0.0, "acc_runs": [0.2622053872053872], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009094042554994908, "centered_acc": 0.024691358024691395, "centered_acc_ci": 0.0, "centered_acc_runs": [0.024691358024691395], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2636518771331058, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2636518771331058], "acc": 0.21843003412969283, "acc_ci": 0.0, "acc_runs": [0.21843003412969283], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012875929151296968, "centered_acc": 0.018202502844141044, "centered_acc_ci": 0.0, "centered_acc_runs": [0.018202502844141044], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.16, "acc_ci": 0.0, "acc_runs": [0.16], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5114443567482242, "acc_ci": 0.0, "acc_runs": [0.5114443567482242], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014048804199859391, "centered_acc": 0.022888713496448387, "centered_acc_ci": 0.0, "centered_acc_runs": [0.022888713496448387], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.47, "acc_ci": 0.0, "acc_runs": [0.47], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05016135580465919, "centered_acc": -0.06000000000000005, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.06000000000000005], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 29291163.031463638, "perplexity_ci": 0.0, "perplexity_runs": [29291163.031463638], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.184, "acc_norm_ci": 0.0, "acc_norm_runs": [0.184], "acc": 0.05, "acc_ci": 0.0, "acc_runs": [0.05], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01225945734093864, "centered_acc": -0.08800000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08800000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29394675357639716, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2839834926879862, "mean_acc_ci": 0.0, "mean_centered_acc": -0.02813642670330443, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29394675357639716], "mean_acc_runs": [0.2839834926879862], "mean_centered_acc_runs": [-0.02813642670330443], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-6912", "checkpoint": "checkpoint-6912", "global_step": 6912, "tokens_seen": 7247757312, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26090420235012945, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26090420235012945], "acc": 0.25951005775741887, "acc_ci": 0.0, "acc_runs": [0.25951005775741887], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004382303181183701, "centered_acc": 0.014538936466839273, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014538936466839273], "ci": 0.0}, "piqa": {"acc_norm": 0.5021762785636561, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5021762785636561], "acc": 0.5261153427638737, "acc_ci": 0.0, "acc_runs": [0.5261153427638737], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665713661738719, "centered_acc": 0.0043525571273121955, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0043525571273121955], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2697811447811448, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2697811447811448], "acc": 0.2516835016835017, "acc_ci": 0.0, "acc_runs": [0.2516835016835017], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009107527914671116, "centered_acc": 0.02637485970819305, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02637485970819305], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2619453924914676, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2619453924914676], "acc": 0.21075085324232082, "acc_ci": 0.0, "acc_runs": [0.21075085324232082], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012849054826857987, "centered_acc": 0.015927189988623452, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015927189988623452], "ci": 0.0}, "openbookqa": {"acc_norm": 0.26, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26], "acc": 0.168, "acc_ci": 0.0, "acc_runs": [0.168], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019635965529725526, "centered_acc": 0.013333333333333345, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013333333333333345], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.510655090765588, "acc_ci": 0.0, "acc_runs": [0.510655090765588], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014049294536290372, "centered_acc": 0.02131018153117603, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02131018153117603], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48, "acc_ci": 0.0, "acc_runs": [0.48], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05021167315686783, "centered_acc": -0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.040000000000000036], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 30090393.269364145, "perplexity_ci": 0.0, "perplexity_runs": [30090393.269364145], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.182, "acc_norm_ci": 0.0, "acc_norm_runs": [0.182], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0122075806376622, "centered_acc": -0.09066666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09066666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.28946783636439966, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28310023079864033, "mean_acc_ci": 0.0, "mean_centered_acc": -0.03091718721819875, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.28946783636439966], "mean_acc_runs": [0.28310023079864033], "mean_centered_acc_runs": [-0.03091718721819875], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-7200", "checkpoint": "checkpoint-7200", "global_step": 7200, "tokens_seen": 7549747200, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2629954192391954, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2629954192391954], "acc": 0.25891256721768574, "acc_ci": 0.0, "acc_runs": [0.25891256721768574], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004393601887506574, "centered_acc": 0.017327225652260525, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017327225652260525], "ci": 0.0}, "piqa": {"acc_norm": 0.5070729053318824, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5070729053318824], "acc": 0.5310119695321001, "acc_ci": 0.0, "acc_runs": [0.5310119695321001], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011664656918146136, "centered_acc": 0.014145810663764857, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014145810663764857], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26851851851851855, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26851851851851855], "acc": 0.26052188552188554, "acc_ci": 0.0, "acc_runs": [0.26052188552188554], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009094042554994908, "centered_acc": 0.024691358024691395, "centered_acc_ci": 0.0, "centered_acc_runs": [0.024691358024691395], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26791808873720135, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26791808873720135], "acc": 0.2167235494880546, "acc_ci": 0.0, "acc_runs": [0.2167235494880546], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012942030195136482, "centered_acc": 0.023890784982935138, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023890784982935138], "ci": 0.0}, "openbookqa": {"acc_norm": 0.266, "acc_norm_ci": 0.0, "acc_norm_runs": [0.266], "acc": 0.164, "acc_ci": 0.0, "acc_runs": [0.164], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019780559675655396, "centered_acc": 0.021333333333333353, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021333333333333353], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5059194948697711, "acc_ci": 0.0, "acc_runs": [0.5059194948697711], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051500838485923, "centered_acc": 0.011838989739542116, "centered_acc_ci": 0.0, "centered_acc_runs": [0.011838989739542116], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5, "acc_ci": 0.0, "acc_runs": [0.5], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.050251890762960605, "centered_acc": 0.0, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 31246624.518439956, "perplexity_ci": 0.0, "perplexity_runs": [31246624.518439956], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.186, "acc_norm_ci": 0.0, "acc_norm_runs": [0.186], "acc": 0.049, "acc_ci": 0.0, "acc_runs": [0.049], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012310790208412926, "centered_acc": -0.08533333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08533333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2930841553044663, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2864570990114822, "mean_acc_ci": 0.0, "mean_centered_acc": -0.02394787859882281, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2930841553044663], "mean_acc_runs": [0.2864570990114822], "mean_centered_acc_runs": [-0.02394787859882281], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-7488", "checkpoint": "checkpoint-7488", "global_step": 7488, "tokens_seen": 7851737088, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25960963951404104, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25960963951404104], "acc": 0.25851424019119695, "acc_ci": 0.0, "acc_runs": [0.25851424019119695], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004375244237044889, "centered_acc": 0.012812852685388046, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012812852685388046], "ci": 0.0}, "piqa": {"acc_norm": 0.5021762785636561, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5021762785636561], "acc": 0.5223068552774756, "acc_ci": 0.0, "acc_runs": [0.5223068552774756], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665713661738719, "centered_acc": 0.0043525571273121955, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0043525571273121955], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2718855218855219, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2718855218855219], "acc": 0.25252525252525254, "acc_ci": 0.0, "acc_runs": [0.25252525252525254], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009129795867310605, "centered_acc": 0.02918069584736253, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02918069584736253], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26706484641638223, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26706484641638223], "acc": 0.20819112627986347, "acc_ci": 0.0, "acc_runs": [0.20819112627986347], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01292893319649643, "centered_acc": 0.022753128555176305, "centered_acc_ci": 0.0, "centered_acc_runs": [0.022753128555176305], "ci": 0.0}, "openbookqa": {"acc_norm": 0.268, "acc_norm_ci": 0.0, "acc_norm_runs": [0.268], "acc": 0.162, "acc_ci": 0.0, "acc_runs": [0.162], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019827714859587477, "centered_acc": 0.02400000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02400000000000002], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4925019731649566, "acc_ci": 0.0, "acc_runs": [0.4925019731649566], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050905521228646, "centered_acc": -0.014996053670086829, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.014996053670086829], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.51, "acc_ci": 0.0, "acc_runs": [0.51], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05024183937956913, "centered_acc": 0.020000000000000018, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020000000000000018], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 28540135.2715266, "perplexity_ci": 0.0, "perplexity_runs": [28540135.2715266], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.188, "acc_norm_ci": 0.0, "acc_norm_runs": [0.188], "acc": 0.05, "acc_ci": 0.0, "acc_runs": [0.05], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01236158601510376, "centered_acc": -0.08266666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08266666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29278938106326685, "mean_acc_norm_ci": 0.0, "mean_acc": 0.283452097092407, "mean_acc_ci": 0.0, "mean_centered_acc": -0.02533206250823486, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29278938106326685], "mean_acc_runs": [0.283452097092407], "mean_centered_acc_runs": [-0.02533206250823486], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-7776", "checkpoint": "checkpoint-7776", "global_step": 7776, "tokens_seen": 8153726976, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2593108942441745, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2593108942441745], "acc": 0.25712009559848636, "acc_ci": 0.0, "acc_runs": [0.25712009559848636], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043736082125607105, "centered_acc": 0.012414525658899297, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012414525658899297], "ci": 0.0}, "piqa": {"acc_norm": 0.5076169749727966, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5076169749727966], "acc": 0.5282916213275299, "acc_ci": 0.0, "acc_runs": [0.5282916213275299], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011664470424044771, "centered_acc": 0.015233949945593128, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015233949945593128], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2706228956228956, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2706228956228956], "acc": 0.2529461279461279, "acc_ci": 0.0, "acc_runs": [0.2529461279461279], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009116466166404035, "centered_acc": 0.0274971941638608, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0274971941638608], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26109215017064846, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26109215017064846], "acc": 0.21245733788395904, "acc_ci": 0.0, "acc_runs": [0.21245733788395904], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012835523909473708, "centered_acc": 0.014789533560864617, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014789533560864617], "ci": 0.0}, "openbookqa": {"acc_norm": 0.276, "acc_norm_ci": 0.0, "acc_norm_runs": [0.276], "acc": 0.158, "acc_ci": 0.0, "acc_runs": [0.158], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.02001121929807355, "centered_acc": 0.0346666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0346666666666667], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4909234411996843, "acc_ci": 0.0, "acc_runs": [0.4909234411996843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050170094497657, "centered_acc": -0.01815311760063143, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.01815311760063143], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.52, "acc_ci": 0.0, "acc_runs": [0.52], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05021167315686783, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 20094626.55806883, "perplexity_ci": 0.0, "perplexity_runs": [20094626.55806883], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.182, "acc_norm_ci": 0.0, "acc_norm_runs": [0.182], "acc": 0.048, "acc_ci": 0.0, "acc_runs": [0.048], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0122075806376622, "centered_acc": -0.09066666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09066666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29277381916841916, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28462201474411125, "mean_acc_ci": 0.0, "mean_centered_acc": -0.02307144341377921, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29277381916841916], "mean_acc_runs": [0.28462201474411125], "mean_centered_acc_runs": [-0.02307144341377921], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-8064", "checkpoint": "checkpoint-8064", "global_step": 8064, "tokens_seen": 8455716864, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2591117307309301, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2591117307309301], "acc": 0.2577175861382195, "acc_ci": 0.0, "acc_runs": [0.2577175861382195], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004372516060164899, "centered_acc": 0.012148974307906771, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012148974307906771], "ci": 0.0}, "piqa": {"acc_norm": 0.5027203482045702, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5027203482045702], "acc": 0.5217627856365615, "acc_ci": 0.0, "acc_runs": [0.5217627856365615], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665651503000776, "centered_acc": 0.005440696409140466, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005440696409140466], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2702020202020202, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2702020202020202], "acc": 0.25084175084175087, "acc_ci": 0.0, "acc_runs": [0.25084175084175087], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009112002229119622, "centered_acc": 0.026936026936026963, "centered_acc_ci": 0.0, "centered_acc_runs": [0.026936026936026963], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26023890784982934, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26023890784982934], "acc": 0.20733788395904437, "acc_ci": 0.0, "acc_runs": [0.20733788395904437], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012821930225112672, "centered_acc": 0.013651877133105783, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013651877133105783], "ci": 0.0}, "openbookqa": {"acc_norm": 0.268, "acc_norm_ci": 0.0, "acc_norm_runs": [0.268], "acc": 0.166, "acc_ci": 0.0, "acc_runs": [0.166], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019827714859587477, "centered_acc": 0.02400000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02400000000000002], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48382004735595896, "acc_ci": 0.0, "acc_runs": [0.48382004735595896], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014045126130978676, "centered_acc": -0.03235990528808208, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.03235990528808208], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.49, "acc_ci": 0.0, "acc_runs": [0.49], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05024183937956913, "centered_acc": -0.020000000000000018, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.020000000000000018], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 27210636.409949224, "perplexity_ci": 0.0, "perplexity_runs": [27210636.409949224], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.187, "acc_norm_ci": 0.0, "acc_norm_runs": [0.187], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012336254828074166, "centered_acc": -0.084, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.084], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.291212167831225, "mean_acc_norm_ci": 0.0, "mean_acc": 0.280296157741686, "mean_acc_ci": 0.0, "mean_centered_acc": -0.03306748966161127, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.291212167831225], "mean_acc_runs": [0.280296157741686], "mean_centered_acc_runs": [-0.03306748966161127], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/mask30-fulle-lm", "run": "mask30-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/mask30-fulle-lm/checkpoint-8352", "checkpoint": "checkpoint-8352", "global_step": 8352, "tokens_seen": 8757706752, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26010754829715194, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26010754829715194], "acc": 0.25721967735510853, "acc_ci": 0.0, "acc_runs": [0.25721967735510853], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004377965074211591, "centered_acc": 0.013476731062869248, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013476731062869248], "ci": 0.0}, "piqa": {"acc_norm": 0.5179542981501633, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5179542981501633], "acc": 0.5266594124047879, "acc_ci": 0.0, "acc_runs": [0.5266594124047879], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011658300623287121, "centered_acc": 0.0359085963003265, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0359085963003265], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2702020202020202, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2702020202020202], "acc": 0.2516835016835017, "acc_ci": 0.0, "acc_runs": [0.2516835016835017], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009112002229119622, "centered_acc": 0.026936026936026963, "centered_acc_ci": 0.0, "centered_acc_runs": [0.026936026936026963], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25341296928327645, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25341296928327645], "acc": 0.20648464163822525, "acc_ci": 0.0, "acc_runs": [0.20648464163822525], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012710896778378677, "centered_acc": 0.004550625711035261, "centered_acc_ci": 0.0, "centered_acc_runs": [0.004550625711035261], "ci": 0.0}, "openbookqa": {"acc_norm": 0.262, "acc_norm_ci": 0.0, "acc_norm_runs": [0.262], "acc": 0.162, "acc_ci": 0.0, "acc_runs": [0.162], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019684688820194637, "centered_acc": 0.016000000000000014, "centered_acc_ci": 0.0, "centered_acc_runs": [0.016000000000000014], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48697711128650356, "acc_ci": 0.0, "acc_runs": [0.48697711128650356], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014047718393997596, "centered_acc": -0.026045777426992878, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.026045777426992878], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5, "acc_ci": 0.0, "acc_runs": [0.5], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.050251890762960605, "centered_acc": 0.0, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 29409355.486137535, "perplexity_ci": 0.0, "perplexity_runs": [29409355.486137535], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.188, "acc_norm_ci": 0.0, "acc_norm_runs": [0.188], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01236158601510376, "centered_acc": -0.08266666666666667, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08266666666666667], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29194613932210195, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28165058678534516, "mean_acc_ci": 0.0, "mean_centered_acc": -0.028362837837333437, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29194613932210195], "mean_acc_runs": [0.28165058678534516], "mean_centered_acc_runs": [-0.028362837837333437], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-10080.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-10080.json new file mode 100644 index 0000000000000000000000000000000000000000..0c2040a51b32368b3e0129308cf52bf7236e871c --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-10080.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-10080", + "checkpoint": "checkpoint-10080", + "global_step": 10080, + "eval_loss": 5.884442329406738, + "metrics": { + "eval_loss": 5.884442329406738, + "eval_model_preparation_time": 0.0054, + "eval_runtime": 130.3007, + "eval_samples_per_second": 18.741, + "eval_steps_per_second": 2.348, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-10368.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-10368.json new file mode 100644 index 0000000000000000000000000000000000000000..b9802178d6f75917a2c2d5aca3c2634b6d79196e --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-10368.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-10368", + "checkpoint": "checkpoint-10368", + "global_step": 10368, + "eval_loss": 5.880742073059082, + "metrics": { + "eval_loss": 5.880742073059082, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 130.3692, + "eval_samples_per_second": 18.731, + "eval_steps_per_second": 2.347, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-1152.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-1152.json new file mode 100644 index 0000000000000000000000000000000000000000..d3217d0093473095a19e9454629c650c1be9b727 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-1152.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-1152", + "checkpoint": "checkpoint-1152", + "global_step": 1152, + "eval_loss": 3.9473233222961426, + "metrics": { + "eval_loss": 3.9473233222961426, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 129.494, + "eval_samples_per_second": 18.858, + "eval_steps_per_second": 2.363, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-1440.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-1440.json new file mode 100644 index 0000000000000000000000000000000000000000..2fd2fe5ddac37d9a1381dece6bce4d34f425cae8 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-1440.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-1440", + "checkpoint": "checkpoint-1440", + "global_step": 1440, + "eval_loss": 3.9768552780151367, + "metrics": { + "eval_loss": 3.9768552780151367, + "eval_model_preparation_time": 0.0043, + "eval_runtime": 129.4431, + "eval_samples_per_second": 18.865, + "eval_steps_per_second": 2.364, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-1728.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-1728.json new file mode 100644 index 0000000000000000000000000000000000000000..acc5984757b9e5a871bab64f60c32566e7ca4a6f --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-1728.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-1728", + "checkpoint": "checkpoint-1728", + "global_step": 1728, + "eval_loss": 4.0437912940979, + "metrics": { + "eval_loss": 4.0437912940979, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 129.4641, + "eval_samples_per_second": 18.862, + "eval_steps_per_second": 2.364, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-2016.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-2016.json new file mode 100644 index 0000000000000000000000000000000000000000..ae1a723dee804259c7d7b071fea52ab0def259a7 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-2016.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2016", + "checkpoint": "checkpoint-2016", + "global_step": 2016, + "eval_loss": 4.126010894775391, + "metrics": { + "eval_loss": 4.126010894775391, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.9538, + "eval_samples_per_second": 18.791, + "eval_steps_per_second": 2.355, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-2304.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-2304.json new file mode 100644 index 0000000000000000000000000000000000000000..aa7d10b207269d9e37f3de03dea92dd0768d628f --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-2304.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2304", + "checkpoint": "checkpoint-2304", + "global_step": 2304, + "eval_loss": 4.216304302215576, + "metrics": { + "eval_loss": 4.216304302215576, + "eval_model_preparation_time": 0.0038, + "eval_runtime": 130.2199, + "eval_samples_per_second": 18.753, + "eval_steps_per_second": 2.35, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-2592.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-2592.json new file mode 100644 index 0000000000000000000000000000000000000000..1cdfff836f82ef29df9c8fa986f0fd048d6478a1 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-2592.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2592", + "checkpoint": "checkpoint-2592", + "global_step": 2592, + "eval_loss": 4.3146514892578125, + "metrics": { + "eval_loss": 4.3146514892578125, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 130.4072, + "eval_samples_per_second": 18.726, + "eval_steps_per_second": 2.346, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-288.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-288.json new file mode 100644 index 0000000000000000000000000000000000000000..808b8b0983ad7222ccc9c01f0929b76ab6b32aa2 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-288.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-288", + "checkpoint": "checkpoint-288", + "global_step": 288, + "eval_loss": 5.458173751831055, + "metrics": { + "eval_loss": 5.458173751831055, + "eval_model_preparation_time": 0.0038, + "eval_runtime": 130.1941, + "eval_samples_per_second": 18.757, + "eval_steps_per_second": 2.35, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-2880.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-2880.json new file mode 100644 index 0000000000000000000000000000000000000000..379d2854ae964683155d261fc91f89361f201ce4 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-2880.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2880", + "checkpoint": "checkpoint-2880", + "global_step": 2880, + "eval_loss": 4.405431747436523, + "metrics": { + "eval_loss": 4.405431747436523, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 130.4215, + "eval_samples_per_second": 18.724, + "eval_steps_per_second": 2.346, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-3168.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-3168.json new file mode 100644 index 0000000000000000000000000000000000000000..18a98e309614d1a10e988290e11c868ba5dde5c4 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-3168.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-3168", + "checkpoint": "checkpoint-3168", + "global_step": 3168, + "eval_loss": 4.508022785186768, + "metrics": { + "eval_loss": 4.508022785186768, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 130.4211, + "eval_samples_per_second": 18.724, + "eval_steps_per_second": 2.346, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-3456.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-3456.json new file mode 100644 index 0000000000000000000000000000000000000000..91783fffca05a0e1507af05f13622448b0f6d403 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-3456.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-3456", + "checkpoint": "checkpoint-3456", + "global_step": 3456, + "eval_loss": 4.594779014587402, + "metrics": { + "eval_loss": 4.594779014587402, + "eval_model_preparation_time": 0.0038, + "eval_runtime": 130.4593, + "eval_samples_per_second": 18.718, + "eval_steps_per_second": 2.346, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-3744.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-3744.json new file mode 100644 index 0000000000000000000000000000000000000000..576aa03bab74af0066d3b5a9889928b40ae746b2 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-3744.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-3744", + "checkpoint": "checkpoint-3744", + "global_step": 3744, + "eval_loss": 4.692500591278076, + "metrics": { + "eval_loss": 4.692500591278076, + "eval_model_preparation_time": 0.004, + "eval_runtime": 130.3904, + "eval_samples_per_second": 18.728, + "eval_steps_per_second": 2.347, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-4032.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-4032.json new file mode 100644 index 0000000000000000000000000000000000000000..c1bbae0cc7fa586c5db3e7f0493b792ecc040865 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-4032.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4032", + "checkpoint": "checkpoint-4032", + "global_step": 4032, + "eval_loss": 4.745551586151123, + "metrics": { + "eval_loss": 4.745551586151123, + "eval_model_preparation_time": 0.0041, + "eval_runtime": 130.4167, + "eval_samples_per_second": 18.725, + "eval_steps_per_second": 2.346, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-4320.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-4320.json new file mode 100644 index 0000000000000000000000000000000000000000..fd005e92139573bca64a0cdb77ced3cf1aea8aa0 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-4320.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4320", + "checkpoint": "checkpoint-4320", + "global_step": 4320, + "eval_loss": 4.841882228851318, + "metrics": { + "eval_loss": 4.841882228851318, + "eval_model_preparation_time": 0.0053, + "eval_runtime": 130.1186, + "eval_samples_per_second": 18.767, + "eval_steps_per_second": 2.352, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-4608.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-4608.json new file mode 100644 index 0000000000000000000000000000000000000000..a364d6a0ae773e97872536907f97e48421e4c39a --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-4608.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4608", + "checkpoint": "checkpoint-4608", + "global_step": 4608, + "eval_loss": 4.894923210144043, + "metrics": { + "eval_loss": 4.894923210144043, + "eval_model_preparation_time": 0.0042, + "eval_runtime": 129.6477, + "eval_samples_per_second": 18.836, + "eval_steps_per_second": 2.36, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-4896.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-4896.json new file mode 100644 index 0000000000000000000000000000000000000000..db46deadbb52f05ec50f2f590c0760b1b55eb432 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-4896.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4896", + "checkpoint": "checkpoint-4896", + "global_step": 4896, + "eval_loss": 4.98560094833374, + "metrics": { + "eval_loss": 4.98560094833374, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 129.5908, + "eval_samples_per_second": 18.844, + "eval_steps_per_second": 2.361, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-5184.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-5184.json new file mode 100644 index 0000000000000000000000000000000000000000..efe074f0be4c6602c3e4cdafaf451ba5a6f61524 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-5184.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-5184", + "checkpoint": "checkpoint-5184", + "global_step": 5184, + "eval_loss": 5.054779529571533, + "metrics": { + "eval_loss": 5.054779529571533, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.6213, + "eval_samples_per_second": 18.84, + "eval_steps_per_second": 2.361, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-5472.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-5472.json new file mode 100644 index 0000000000000000000000000000000000000000..ead85cbab7fff2245958e1ab91a265bcb85a58d3 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-5472.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-5472", + "checkpoint": "checkpoint-5472", + "global_step": 5472, + "eval_loss": 5.127499580383301, + "metrics": { + "eval_loss": 5.127499580383301, + "eval_model_preparation_time": 0.0054, + "eval_runtime": 130.0474, + "eval_samples_per_second": 18.778, + "eval_steps_per_second": 2.353, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-576.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-576.json new file mode 100644 index 0000000000000000000000000000000000000000..ad6edeb8544153db8ee14b38a5be1fbd0b522332 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-576.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-576", + "checkpoint": "checkpoint-576", + "global_step": 576, + "eval_loss": 4.356330394744873, + "metrics": { + "eval_loss": 4.356330394744873, + "eval_model_preparation_time": 0.0035, + "eval_runtime": 130.2771, + "eval_samples_per_second": 18.745, + "eval_steps_per_second": 2.349, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-5760.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-5760.json new file mode 100644 index 0000000000000000000000000000000000000000..d500cbda9f46814fdd1e613439a4aedd2b67bd76 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-5760.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-5760", + "checkpoint": "checkpoint-5760", + "global_step": 5760, + "eval_loss": 5.181248664855957, + "metrics": { + "eval_loss": 5.181248664855957, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 130.2243, + "eval_samples_per_second": 18.752, + "eval_steps_per_second": 2.35, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-6048.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-6048.json new file mode 100644 index 0000000000000000000000000000000000000000..613a4f67400aebd8fddcd48d31547bc20f70c807 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-6048.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6048", + "checkpoint": "checkpoint-6048", + "global_step": 6048, + "eval_loss": 5.242222309112549, + "metrics": { + "eval_loss": 5.242222309112549, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 130.2109, + "eval_samples_per_second": 18.754, + "eval_steps_per_second": 2.35, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-6336.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-6336.json new file mode 100644 index 0000000000000000000000000000000000000000..53d598bd01341494dd2d353099d219bef8b9da5c --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-6336.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6336", + "checkpoint": "checkpoint-6336", + "global_step": 6336, + "eval_loss": 5.273025035858154, + "metrics": { + "eval_loss": 5.273025035858154, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 130.2806, + "eval_samples_per_second": 18.744, + "eval_steps_per_second": 2.349, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-6624.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-6624.json new file mode 100644 index 0000000000000000000000000000000000000000..5783fdefa63c964ebb5a575efccd17b85ca09152 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-6624.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6624", + "checkpoint": "checkpoint-6624", + "global_step": 6624, + "eval_loss": 5.367213249206543, + "metrics": { + "eval_loss": 5.367213249206543, + "eval_model_preparation_time": 0.004, + "eval_runtime": 130.1799, + "eval_samples_per_second": 18.759, + "eval_steps_per_second": 2.351, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-6912.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-6912.json new file mode 100644 index 0000000000000000000000000000000000000000..e50d1add748f4055a755ccc79098d963ccf0b0f4 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-6912.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6912", + "checkpoint": "checkpoint-6912", + "global_step": 6912, + "eval_loss": 5.398253440856934, + "metrics": { + "eval_loss": 5.398253440856934, + "eval_model_preparation_time": 0.0039, + "eval_runtime": 130.2518, + "eval_samples_per_second": 18.748, + "eval_steps_per_second": 2.349, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-72.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-72.json new file mode 100644 index 0000000000000000000000000000000000000000..813c773d8ed5354518966edefb0b1d32bfd36bb1 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-72.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-72", + "checkpoint": "checkpoint-72", + "global_step": 72, + "eval_loss": 7.358715057373047, + "metrics": { + "eval_loss": 7.358715057373047, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 129.5366, + "eval_samples_per_second": 18.852, + "eval_steps_per_second": 2.362, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-7200.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-7200.json new file mode 100644 index 0000000000000000000000000000000000000000..392c4f3d589660308b7f7ac318aef52436832d44 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-7200.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-7200", + "checkpoint": "checkpoint-7200", + "global_step": 7200, + "eval_loss": 5.4624834060668945, + "metrics": { + "eval_loss": 5.4624834060668945, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 130.2516, + "eval_samples_per_second": 18.748, + "eval_steps_per_second": 2.349, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-7488.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-7488.json new file mode 100644 index 0000000000000000000000000000000000000000..95d67232d5ab47679f97dbdeb4492ba99961383a --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-7488.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-7488", + "checkpoint": "checkpoint-7488", + "global_step": 7488, + "eval_loss": 5.509789943695068, + "metrics": { + "eval_loss": 5.509789943695068, + "eval_model_preparation_time": 0.0034, + "eval_runtime": 130.2096, + "eval_samples_per_second": 18.754, + "eval_steps_per_second": 2.35, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-7776.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-7776.json new file mode 100644 index 0000000000000000000000000000000000000000..b8203eb213ef8b7f280a103c1d68a7b809e2ab17 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-7776.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-7776", + "checkpoint": "checkpoint-7776", + "global_step": 7776, + "eval_loss": 5.563534259796143, + "metrics": { + "eval_loss": 5.563534259796143, + "eval_model_preparation_time": 0.0042, + "eval_runtime": 130.2939, + "eval_samples_per_second": 18.742, + "eval_steps_per_second": 2.349, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-8064.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-8064.json new file mode 100644 index 0000000000000000000000000000000000000000..d75d1caef0713d9e767899364da2d78101ef4bc5 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-8064.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-8064", + "checkpoint": "checkpoint-8064", + "global_step": 8064, + "eval_loss": 5.606389999389648, + "metrics": { + "eval_loss": 5.606389999389648, + "eval_model_preparation_time": 0.0037, + "eval_runtime": 130.3616, + "eval_samples_per_second": 18.733, + "eval_steps_per_second": 2.347, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-8352.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-8352.json new file mode 100644 index 0000000000000000000000000000000000000000..48a1ed43da84738a438259cafb0ae92bc7a4fa78 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-8352.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-8352", + "checkpoint": "checkpoint-8352", + "global_step": 8352, + "eval_loss": 5.634708404541016, + "metrics": { + "eval_loss": 5.634708404541016, + "eval_model_preparation_time": 0.004, + "eval_runtime": 130.264, + "eval_samples_per_second": 18.747, + "eval_steps_per_second": 2.349, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-864.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-864.json new file mode 100644 index 0000000000000000000000000000000000000000..bdfd997577bf9d1717a64b350845d26532e6f600 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-864.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-864", + "checkpoint": "checkpoint-864", + "global_step": 864, + "eval_loss": 4.034478187561035, + "metrics": { + "eval_loss": 4.034478187561035, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 130.0421, + "eval_samples_per_second": 18.779, + "eval_steps_per_second": 2.353, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-8640.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-8640.json new file mode 100644 index 0000000000000000000000000000000000000000..5d01dab76693cd1895e741b0ae8cc73a0aca4a11 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-8640.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-8640", + "checkpoint": "checkpoint-8640", + "global_step": 8640, + "eval_loss": 5.657890319824219, + "metrics": { + "eval_loss": 5.657890319824219, + "eval_model_preparation_time": 0.0039, + "eval_runtime": 130.2744, + "eval_samples_per_second": 18.745, + "eval_steps_per_second": 2.349, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-8928.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-8928.json new file mode 100644 index 0000000000000000000000000000000000000000..3b75013d9838bad8ed54b2a5e6fcfa7a7429b57e --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-8928.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-8928", + "checkpoint": "checkpoint-8928", + "global_step": 8928, + "eval_loss": 5.713588237762451, + "metrics": { + "eval_loss": 5.713588237762451, + "eval_model_preparation_time": 0.004, + "eval_runtime": 130.3033, + "eval_samples_per_second": 18.741, + "eval_steps_per_second": 2.348, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-9216.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-9216.json new file mode 100644 index 0000000000000000000000000000000000000000..76df4827496b3144350b5e280401837f9fbe148d --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-9216.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-9216", + "checkpoint": "checkpoint-9216", + "global_step": 9216, + "eval_loss": 5.754378795623779, + "metrics": { + "eval_loss": 5.754378795623779, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 130.2756, + "eval_samples_per_second": 18.745, + "eval_steps_per_second": 2.349, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-9504.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-9504.json new file mode 100644 index 0000000000000000000000000000000000000000..14d7578aedf79c61720ab668d03513e2eac93ea0 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-9504.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-9504", + "checkpoint": "checkpoint-9504", + "global_step": 9504, + "eval_loss": 5.817337512969971, + "metrics": { + "eval_loss": 5.817337512969971, + "eval_model_preparation_time": 0.0036, + "eval_runtime": 130.323, + "eval_samples_per_second": 18.738, + "eval_steps_per_second": 2.348, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-9792.json b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-9792.json new file mode 100644 index 0000000000000000000000000000000000000000..0bb192be84e5d3348070568bef971979ab30cf80 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/checkpoint-9792.json @@ -0,0 +1,35 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-9792", + "checkpoint": "checkpoint-9792", + "global_step": 9792, + "eval_loss": 5.826162338256836, + "metrics": { + "eval_loss": 5.826162338256836, + "eval_model_preparation_time": 0.0044, + "eval_runtime": 130.3165, + "eval_samples_per_second": 18.739, + "eval_steps_per_second": 2.348, + "epoch": 0.0 + }, + "model_type": "default", + "objective": "LM loss", + "l2r": false, + "l2r_percent": 0.0, + "embed_window_size": 1, + "embed_loss_func": "cosine", + "logical_per_device_eval_batch_size": 512, + "per_device_eval_micro_batch_size": 8, + "eval_gradient_accumulation_steps": 64, + "world_size_env": 1, + "world_size_runtime": 1, + "global_logical_eval_batch": 512, + "global_eval_batch_size_arg": 512, + "batch_size_per_device_arg": 64, + "dataloader_num_workers": 4, + "dataloader_prefetch_factor": 4, + "dataloader_pin_memory": true, + "remove_unused_columns": true, + "next_i_enabled": false, + "max_next_i_train": 1 +} diff --git a/results/psm25-spm25-fulle-lm/l2r_n1_eval/summary.jsonl b/results/psm25-spm25-fulle-lm/l2r_n1_eval/summary.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..582f1026ae93edc8d11a50fac540325aca480034 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/l2r_n1_eval/summary.jsonl @@ -0,0 +1,37 @@ +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-72", "checkpoint": "checkpoint-72", "global_step": 72, "eval_loss": 7.358715057373047, "metrics": {"eval_loss": 7.358715057373047, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.5366, "eval_samples_per_second": 18.852, "eval_steps_per_second": 2.362, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-288", "checkpoint": "checkpoint-288", "global_step": 288, "eval_loss": 5.458173751831055, "metrics": {"eval_loss": 5.458173751831055, "eval_model_preparation_time": 0.0038, "eval_runtime": 130.1941, "eval_samples_per_second": 18.757, "eval_steps_per_second": 2.35, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-576", "checkpoint": "checkpoint-576", "global_step": 576, "eval_loss": 4.356330394744873, "metrics": {"eval_loss": 4.356330394744873, "eval_model_preparation_time": 0.0035, "eval_runtime": 130.2771, "eval_samples_per_second": 18.745, "eval_steps_per_second": 2.349, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-864", "checkpoint": "checkpoint-864", "global_step": 864, "eval_loss": 4.034478187561035, "metrics": {"eval_loss": 4.034478187561035, "eval_model_preparation_time": 0.0036, "eval_runtime": 130.0421, "eval_samples_per_second": 18.779, "eval_steps_per_second": 2.353, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-1152", "checkpoint": "checkpoint-1152", "global_step": 1152, "eval_loss": 3.9473233222961426, "metrics": {"eval_loss": 3.9473233222961426, "eval_model_preparation_time": 0.0034, "eval_runtime": 129.494, "eval_samples_per_second": 18.858, "eval_steps_per_second": 2.363, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-1440", "checkpoint": "checkpoint-1440", "global_step": 1440, "eval_loss": 3.9768552780151367, "metrics": {"eval_loss": 3.9768552780151367, "eval_model_preparation_time": 0.0043, "eval_runtime": 129.4431, "eval_samples_per_second": 18.865, "eval_steps_per_second": 2.364, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-1728", "checkpoint": "checkpoint-1728", "global_step": 1728, "eval_loss": 4.0437912940979, "metrics": {"eval_loss": 4.0437912940979, "eval_model_preparation_time": 0.0037, "eval_runtime": 129.4641, "eval_samples_per_second": 18.862, "eval_steps_per_second": 2.364, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2016", "checkpoint": "checkpoint-2016", "global_step": 2016, "eval_loss": 4.126010894775391, "metrics": {"eval_loss": 4.126010894775391, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.9538, "eval_samples_per_second": 18.791, "eval_steps_per_second": 2.355, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2304", "checkpoint": "checkpoint-2304", "global_step": 2304, "eval_loss": 4.216304302215576, "metrics": {"eval_loss": 4.216304302215576, "eval_model_preparation_time": 0.0038, "eval_runtime": 130.2199, "eval_samples_per_second": 18.753, "eval_steps_per_second": 2.35, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2592", "checkpoint": "checkpoint-2592", "global_step": 2592, "eval_loss": 4.3146514892578125, "metrics": {"eval_loss": 4.3146514892578125, "eval_model_preparation_time": 0.0037, "eval_runtime": 130.4072, "eval_samples_per_second": 18.726, "eval_steps_per_second": 2.346, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2880", "checkpoint": "checkpoint-2880", "global_step": 2880, "eval_loss": 4.405431747436523, "metrics": {"eval_loss": 4.405431747436523, "eval_model_preparation_time": 0.0037, "eval_runtime": 130.4215, "eval_samples_per_second": 18.724, "eval_steps_per_second": 2.346, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-3168", "checkpoint": "checkpoint-3168", "global_step": 3168, "eval_loss": 4.508022785186768, "metrics": {"eval_loss": 4.508022785186768, "eval_model_preparation_time": 0.0035, "eval_runtime": 130.4211, "eval_samples_per_second": 18.724, "eval_steps_per_second": 2.346, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-3456", "checkpoint": "checkpoint-3456", "global_step": 3456, "eval_loss": 4.594779014587402, "metrics": {"eval_loss": 4.594779014587402, "eval_model_preparation_time": 0.0038, "eval_runtime": 130.4593, "eval_samples_per_second": 18.718, "eval_steps_per_second": 2.346, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-3744", "checkpoint": "checkpoint-3744", "global_step": 3744, "eval_loss": 4.692500591278076, "metrics": {"eval_loss": 4.692500591278076, "eval_model_preparation_time": 0.004, "eval_runtime": 130.3904, "eval_samples_per_second": 18.728, "eval_steps_per_second": 2.347, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4032", "checkpoint": "checkpoint-4032", "global_step": 4032, "eval_loss": 4.745551586151123, "metrics": {"eval_loss": 4.745551586151123, "eval_model_preparation_time": 0.0041, "eval_runtime": 130.4167, "eval_samples_per_second": 18.725, "eval_steps_per_second": 2.346, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4320", "checkpoint": "checkpoint-4320", "global_step": 4320, "eval_loss": 4.841882228851318, "metrics": {"eval_loss": 4.841882228851318, "eval_model_preparation_time": 0.0053, "eval_runtime": 130.1186, "eval_samples_per_second": 18.767, "eval_steps_per_second": 2.352, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4608", "checkpoint": "checkpoint-4608", "global_step": 4608, "eval_loss": 4.894923210144043, "metrics": {"eval_loss": 4.894923210144043, "eval_model_preparation_time": 0.0042, "eval_runtime": 129.6477, "eval_samples_per_second": 18.836, "eval_steps_per_second": 2.36, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4896", "checkpoint": "checkpoint-4896", "global_step": 4896, "eval_loss": 4.98560094833374, "metrics": {"eval_loss": 4.98560094833374, "eval_model_preparation_time": 0.0035, "eval_runtime": 129.5908, "eval_samples_per_second": 18.844, "eval_steps_per_second": 2.361, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-5184", "checkpoint": "checkpoint-5184", "global_step": 5184, "eval_loss": 5.054779529571533, "metrics": {"eval_loss": 5.054779529571533, "eval_model_preparation_time": 0.0036, "eval_runtime": 129.6213, "eval_samples_per_second": 18.84, "eval_steps_per_second": 2.361, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-5472", "checkpoint": "checkpoint-5472", "global_step": 5472, "eval_loss": 5.127499580383301, "metrics": {"eval_loss": 5.127499580383301, "eval_model_preparation_time": 0.0054, "eval_runtime": 130.0474, "eval_samples_per_second": 18.778, "eval_steps_per_second": 2.353, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-5760", "checkpoint": "checkpoint-5760", "global_step": 5760, "eval_loss": 5.181248664855957, "metrics": {"eval_loss": 5.181248664855957, "eval_model_preparation_time": 0.0036, "eval_runtime": 130.2243, "eval_samples_per_second": 18.752, "eval_steps_per_second": 2.35, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6048", "checkpoint": "checkpoint-6048", "global_step": 6048, "eval_loss": 5.242222309112549, "metrics": {"eval_loss": 5.242222309112549, "eval_model_preparation_time": 0.0037, "eval_runtime": 130.2109, "eval_samples_per_second": 18.754, "eval_steps_per_second": 2.35, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6336", "checkpoint": "checkpoint-6336", "global_step": 6336, "eval_loss": 5.273025035858154, "metrics": {"eval_loss": 5.273025035858154, "eval_model_preparation_time": 0.0037, "eval_runtime": 130.2806, "eval_samples_per_second": 18.744, "eval_steps_per_second": 2.349, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6624", "checkpoint": "checkpoint-6624", "global_step": 6624, "eval_loss": 5.367213249206543, "metrics": {"eval_loss": 5.367213249206543, "eval_model_preparation_time": 0.004, "eval_runtime": 130.1799, "eval_samples_per_second": 18.759, "eval_steps_per_second": 2.351, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6912", "checkpoint": "checkpoint-6912", "global_step": 6912, "eval_loss": 5.398253440856934, "metrics": {"eval_loss": 5.398253440856934, "eval_model_preparation_time": 0.0039, "eval_runtime": 130.2518, "eval_samples_per_second": 18.748, "eval_steps_per_second": 2.349, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-7200", "checkpoint": "checkpoint-7200", "global_step": 7200, "eval_loss": 5.4624834060668945, "metrics": {"eval_loss": 5.4624834060668945, "eval_model_preparation_time": 0.0037, "eval_runtime": 130.2516, "eval_samples_per_second": 18.748, "eval_steps_per_second": 2.349, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-7488", "checkpoint": "checkpoint-7488", "global_step": 7488, "eval_loss": 5.509789943695068, "metrics": {"eval_loss": 5.509789943695068, "eval_model_preparation_time": 0.0034, "eval_runtime": 130.2096, "eval_samples_per_second": 18.754, "eval_steps_per_second": 2.35, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-7776", "checkpoint": "checkpoint-7776", "global_step": 7776, "eval_loss": 5.563534259796143, "metrics": {"eval_loss": 5.563534259796143, "eval_model_preparation_time": 0.0042, "eval_runtime": 130.2939, "eval_samples_per_second": 18.742, "eval_steps_per_second": 2.349, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-8064", "checkpoint": "checkpoint-8064", "global_step": 8064, "eval_loss": 5.606389999389648, "metrics": {"eval_loss": 5.606389999389648, "eval_model_preparation_time": 0.0037, "eval_runtime": 130.3616, "eval_samples_per_second": 18.733, "eval_steps_per_second": 2.347, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-8352", "checkpoint": "checkpoint-8352", "global_step": 8352, "eval_loss": 5.634708404541016, "metrics": {"eval_loss": 5.634708404541016, "eval_model_preparation_time": 0.004, "eval_runtime": 130.264, "eval_samples_per_second": 18.747, "eval_steps_per_second": 2.349, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-8640", "checkpoint": "checkpoint-8640", "global_step": 8640, "eval_loss": 5.657890319824219, "metrics": {"eval_loss": 5.657890319824219, "eval_model_preparation_time": 0.0039, "eval_runtime": 130.2744, "eval_samples_per_second": 18.745, "eval_steps_per_second": 2.349, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-8928", "checkpoint": "checkpoint-8928", "global_step": 8928, "eval_loss": 5.713588237762451, "metrics": {"eval_loss": 5.713588237762451, "eval_model_preparation_time": 0.004, "eval_runtime": 130.3033, "eval_samples_per_second": 18.741, "eval_steps_per_second": 2.348, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-9216", "checkpoint": "checkpoint-9216", "global_step": 9216, "eval_loss": 5.754378795623779, "metrics": {"eval_loss": 5.754378795623779, "eval_model_preparation_time": 0.0036, "eval_runtime": 130.2756, "eval_samples_per_second": 18.745, "eval_steps_per_second": 2.349, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-9504", "checkpoint": "checkpoint-9504", "global_step": 9504, "eval_loss": 5.817337512969971, "metrics": {"eval_loss": 5.817337512969971, "eval_model_preparation_time": 0.0036, "eval_runtime": 130.323, "eval_samples_per_second": 18.738, "eval_steps_per_second": 2.348, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-9792", "checkpoint": "checkpoint-9792", "global_step": 9792, "eval_loss": 5.826162338256836, "metrics": {"eval_loss": 5.826162338256836, "eval_model_preparation_time": 0.0044, "eval_runtime": 130.3165, "eval_samples_per_second": 18.739, "eval_steps_per_second": 2.348, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-10080", "checkpoint": "checkpoint-10080", "global_step": 10080, "eval_loss": 5.884442329406738, "metrics": {"eval_loss": 5.884442329406738, "eval_model_preparation_time": 0.0054, "eval_runtime": 130.3007, "eval_samples_per_second": 18.741, "eval_steps_per_second": 2.348, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-10368", "checkpoint": "checkpoint-10368", "global_step": 10368, "eval_loss": 5.880742073059082, "metrics": {"eval_loss": 5.880742073059082, "eval_model_preparation_time": 0.0036, "eval_runtime": 130.3692, "eval_samples_per_second": 18.731, "eval_steps_per_second": 2.347, "epoch": 0.0}, "model_type": "default", "objective": "LM loss", "l2r": false, "l2r_percent": 0.0, "embed_window_size": 1, "embed_loss_func": "cosine", "logical_per_device_eval_batch_size": 512, "per_device_eval_micro_batch_size": 8, "eval_gradient_accumulation_steps": 64, "world_size_env": 1, "world_size_runtime": 1, "global_logical_eval_batch": 512, "global_eval_batch_size_arg": 512, "batch_size_per_device_arg": 64, "dataloader_num_workers": 4, "dataloader_prefetch_factor": 4, "dataloader_pin_memory": true, "remove_unused_columns": true, "next_i_enabled": false, "max_next_i_train": 1} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-1152.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-1152.json new file mode 100644 index 0000000000000000000000000000000000000000..21488e45af9c000eda35d9e15d0003f34070e037 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-1152.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-1152", + "checkpoint": "checkpoint-1152", + "global_step": 1152, + "tokens_seen": 1207959552, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2566221868153754, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2566221868153754 + ], + "acc": 0.25473013343955386, + "acc_ci": 0.0, + "acc_runs": [ + 0.25473013343955386 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004358764596401022, + "centered_acc": 0.008829582420500545, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008829582420500545 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5179542981501633, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5179542981501633 + ], + "acc": 0.543525571273123, + "acc_ci": 0.0, + "acc_runs": [ + 0.543525571273123 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011658300623287121, + "centered_acc": 0.0359085963003265, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0359085963003265 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2718855218855219, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2718855218855219 + ], + "acc": 0.26430976430976433, + "acc_ci": 0.0, + "acc_runs": [ + 0.26430976430976433 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009129795867310605, + "centered_acc": 0.02918069584736253, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02918069584736253 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24914675767918087, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24914675767918087 + ], + "acc": 0.21160409556313994, + "acc_ci": 0.0, + "acc_runs": [ + 0.21160409556313994 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012639407111926505, + "centered_acc": -0.0011376564277588337, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0011376564277588337 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.284, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.284 + ], + "acc": 0.166, + "acc_ci": 0.0, + "acc_runs": [ + 0.166 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020186703693570774, + "centered_acc": 0.0453333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0453333333333333 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.47198105761641673, + "acc_ci": 0.0, + "acc_runs": [ + 0.47198105761641673 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014030404213405703, + "centered_acc": -0.056037884767166535, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.056037884767166535 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 12722601.352029193, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 12722601.352029193 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.187, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.187 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012336254828074166, + "centered_acc": -0.084, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.084 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2944347940883736, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2906438083975698, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.013927601082889105, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2944347940883736 + ], + "mean_acc_runs": [ + 0.2906438083975698 + ], + "mean_centered_acc_runs": [ + -0.013927601082889105 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-1440.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-1440.json new file mode 100644 index 0000000000000000000000000000000000000000..d6533dda8fad24649540b25a5db1ad1cd99df013 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-1440.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-1440", + "checkpoint": "checkpoint-1440", + "global_step": 1440, + "tokens_seen": 1509949440, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2566221868153754, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2566221868153754 + ], + "acc": 0.253734315873332, + "acc_ci": 0.0, + "acc_runs": [ + 0.253734315873332 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004358764596401022, + "centered_acc": 0.008829582420500545, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008829582420500545 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5266594124047879, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5266594124047879 + ], + "acc": 0.5369967355821545, + "acc_ci": 0.0, + "acc_runs": [ + 0.5369967355821545 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01164922999434743, + "centered_acc": 0.05331882480957573, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05331882480957573 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26725589225589225, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26725589225589225 + ], + "acc": 0.2596801346801347, + "acc_ci": 0.0, + "acc_runs": [ + 0.2596801346801347 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009080463246017426, + "centered_acc": 0.023007856341189663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023007856341189663 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.25, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25 + ], + "acc": 0.21245733788395904, + "acc_ci": 0.0, + "acc_runs": [ + 0.21245733788395904 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012653835621466646, + "centered_acc": 0.0, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.29, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.29 + ], + "acc": 0.164, + "acc_ci": 0.0, + "acc_runs": [ + 0.164 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020313179231745138, + "centered_acc": 0.05333333333333331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05333333333333331 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48855564325177586, + "acc_ci": 0.0, + "acc_runs": [ + 0.48855564325177586 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014048804199859391, + "centered_acc": -0.022888713496448276, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.022888713496448276 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.00019406171162429653, + "acc_ci": 0.0, + "acc_runs": [ + 0.00019406171162429653 + ], + "perplexity": 19299504.75313499, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 19299504.75313499 + ], + "stderr": 0.00019406171162429653, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.177, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.177 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012075463420375087, + "centered_acc": -0.09733333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09733333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2945895819126759, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.291890569075668, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.007239725153086869, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2945895819126759 + ], + "mean_acc_runs": [ + 0.291890569075668 + ], + "mean_centered_acc_runs": [ + -0.007239725153086869 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-1728.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-1728.json new file mode 100644 index 0000000000000000000000000000000000000000..f9be6d838f70992740f489f9969956ee98c95814 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-1728.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-1728", + "checkpoint": "checkpoint-1728", + "global_step": 1728, + "tokens_seen": 1811939328, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25144393547102173, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25144393547102173 + ], + "acc": 0.25492929695279826, + "acc_ci": 0.0, + "acc_runs": [ + 0.25492929695279826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043295650165269565, + "centered_acc": 0.0019252472946956385, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0019252472946956385 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5359085963003264, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5359085963003264 + ], + "acc": 0.5424374319912949, + "acc_ci": 0.0, + "acc_runs": [ + 0.5424374319912949 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011635700809215522, + "centered_acc": 0.07181719260065278, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.07181719260065278 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2735690235690236, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2735690235690236 + ], + "acc": 0.265993265993266, + "acc_ci": 0.0, + "acc_runs": [ + 0.265993265993266 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009147424438490918, + "centered_acc": 0.031425364758698095, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.031425364758698095 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24744027303754265, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24744027303754265 + ], + "acc": 0.2158703071672355, + "acc_ci": 0.0, + "acc_runs": [ + 0.2158703071672355 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01261035266329266, + "centered_acc": -0.0034129692832764644, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0034129692832764644 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.292, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.292 + ], + "acc": 0.16, + "acc_ci": 0.0, + "acc_runs": [ + 0.16 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020354375480530096, + "centered_acc": 0.05599999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05599999999999997 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4940805051302289, + "acc_ci": 0.0, + "acc_runs": [ + 0.4940805051302289 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051500838485925, + "centered_acc": -0.011838989739542227, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.011838989739542227 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 26026105.85764453, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 26026105.85764453 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29689363806298574, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2936598269008524, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0032788034245968523, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29689363806298574 + ], + "mean_acc_runs": [ + 0.2936598269008524 + ], + "mean_centered_acc_runs": [ + -0.0032788034245968523 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-2016.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-2016.json new file mode 100644 index 0000000000000000000000000000000000000000..d9b2bfc5ffdb9d622066f3c530db157b75cd45a2 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-2016.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2016", + "checkpoint": "checkpoint-2016", + "global_step": 2016, + "tokens_seen": 2113929216, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2544313881696873, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2544313881696873 + ], + "acc": 0.25273849830711015, + "acc_ci": 0.0, + "acc_runs": [ + 0.25273849830711015 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00434650985067984, + "centered_acc": 0.005908517559583067, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005908517559583067 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5321001088139282, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5321001088139282 + ], + "acc": 0.5386289445048966, + "acc_ci": 0.0, + "acc_runs": [ + 0.5386289445048966 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011641758014820223, + "centered_acc": 0.06420021762785644, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06420021762785644 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26725589225589225, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26725589225589225 + ], + "acc": 0.255050505050505, + "acc_ci": 0.0, + "acc_runs": [ + 0.255050505050505 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009080463246017426, + "centered_acc": 0.023007856341189663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023007856341189663 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24744027303754265, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24744027303754265 + ], + "acc": 0.20051194539249148, + "acc_ci": 0.0, + "acc_runs": [ + 0.20051194539249148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01261035266329266, + "centered_acc": -0.0034129692832764644, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0034129692832764644 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.29, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.29 + ], + "acc": 0.17, + "acc_ci": 0.0, + "acc_runs": [ + 0.17 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020313179231745138, + "centered_acc": 0.05333333333333331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05333333333333331 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5217048145224941, + "acc_ci": 0.0, + "acc_runs": [ + 0.5217048145224941 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014039239216484667, + "centered_acc": 0.04340962904498813, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04340962904498813 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.6, + "acc_ci": 0.0, + "acc_runs": [ + 0.6 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0492365963917331, + "centered_acc": 0.19999999999999996, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.19999999999999996 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 48697070.09879444, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 48697070.09879444 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.173, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.173 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011967214137559927, + "centered_acc": -0.10266666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10266666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29403794371284175, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29629221695511976, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.00448387127826756, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29403794371284175 + ], + "mean_acc_runs": [ + 0.29629221695511976 + ], + "mean_centered_acc_runs": [ + 0.00448387127826756 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-2304.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-2304.json new file mode 100644 index 0000000000000000000000000000000000000000..f22f44ade1d055573c82e3edc727c41eb4d9d505 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-2304.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2304", + "checkpoint": "checkpoint-2304", + "global_step": 2304, + "tokens_seen": 2415919104, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25403306114319857, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25403306114319857 + ], + "acc": 0.2546305516829317, + "acc_ci": 0.0, + "acc_runs": [ + 0.2546305516829317 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004344266179634878, + "centered_acc": 0.005377414857598091, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005377414857598091 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.528835690968444, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.528835690968444 + ], + "acc": 0.5408052230685527, + "acc_ci": 0.0, + "acc_runs": [ + 0.5408052230685527 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011646407809944788, + "centered_acc": 0.05767138193688792, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05767138193688792 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26725589225589225, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26725589225589225 + ], + "acc": 0.265993265993266, + "acc_ci": 0.0, + "acc_runs": [ + 0.265993265993266 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009080463246017426, + "centered_acc": 0.023007856341189663, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.023007856341189663 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.23293515358361774, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.23293515358361774 + ], + "acc": 0.20392491467576793, + "acc_ci": 0.0, + "acc_runs": [ + 0.20392491467576793 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012352507042617355, + "centered_acc": -0.022753128555176343, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.022753128555176343 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.3, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.3 + ], + "acc": 0.176, + "acc_ci": 0.0, + "acc_runs": [ + 0.176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020514426225627987, + "centered_acc": 0.06666666666666665, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06666666666666665 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48855564325177586, + "acc_ci": 0.0, + "acc_runs": [ + 0.48855564325177586 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014048804199859391, + "centered_acc": -0.022888713496448276, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.022888713496448276 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.61, + "acc_ci": 0.0, + "acc_runs": [ + 0.61 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04902071300001973, + "centered_acc": 0.21999999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.21999999999999997 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 81215739.36168465, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 81215739.36168465 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.176, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.176 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012048616898597498, + "centered_acc": -0.09866666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09866666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29317663299185875, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29641970604459944, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.001667807263172041, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29317663299185875 + ], + "mean_acc_runs": [ + 0.29641970604459944 + ], + "mean_centered_acc_runs": [ + -0.001667807263172041 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-2592.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-2592.json new file mode 100644 index 0000000000000000000000000000000000000000..6260d16d0562e6c5865bd852b83631930d2ed667 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-2592.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2592", + "checkpoint": "checkpoint-2592", + "global_step": 2592, + "tokens_seen": 2717908992, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2545309699263095, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2545309699263095 + ], + "acc": 0.25542720573590916, + "acc_ci": 0.0, + "acc_runs": [ + 0.25542720573590916 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043470700195274055, + "centered_acc": 0.006041293235079366, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006041293235079366 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5244831338411317, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5244831338411317 + ], + "acc": 0.5310119695321001, + "acc_ci": 0.0, + "acc_runs": [ + 0.5310119695321001 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011651830225710194, + "centered_acc": 0.04896626768226331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04896626768226331 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2668350168350168, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2668350168350168 + ], + "acc": 0.257996632996633, + "acc_ci": 0.0, + "acc_runs": [ + 0.257996632996633 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009075915859266983, + "centered_acc": 0.02244668911335575, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02244668911335575 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24061433447098976, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24061433447098976 + ], + "acc": 0.20733788395904437, + "acc_ci": 0.0, + "acc_runs": [ + 0.20733788395904437 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012491468532390682, + "centered_acc": -0.012514220705346987, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.012514220705346987 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.288, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.288 + ], + "acc": 0.16, + "acc_ci": 0.0, + "acc_runs": [ + 0.16 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020271503835075255, + "centered_acc": 0.05066666666666664, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05066666666666664 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4861878453038674, + "acc_ci": 0.0, + "acc_runs": [ + 0.4861878453038674 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014047122916440271, + "centered_acc": -0.027624309392265234, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.027624309392265234 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 98230320.95594785, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 98230320.95594785 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.18, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.18 + ], + "acc": 0.048, + "acc_ci": 0.0, + "acc_runs": [ + 0.048 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012155153135512024, + "centered_acc": -0.09333333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09333333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2924105758455746, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29142489993012544, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.007641780354019991, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2924105758455746 + ], + "mean_acc_runs": [ + 0.29142489993012544 + ], + "mean_centered_acc_runs": [ + -0.007641780354019991 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-288.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-288.json new file mode 100644 index 0000000000000000000000000000000000000000..e2c87a7bba68c95241ade9c4582b9c7292b88e27 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-288.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-288", + "checkpoint": "checkpoint-288", + "global_step": 288, + "tokens_seen": 301989888, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.253734315873332, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.253734315873332 + ], + "acc": 0.2555267874925314, + "acc_ci": 0.0, + "acc_runs": [ + 0.2555267874925314 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004342580277662561, + "centered_acc": 0.004979087831109341, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.004979087831109341 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.49347116430903154, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.49347116430903154 + ], + "acc": 0.5233949945593036, + "acc_ci": 0.0, + "acc_runs": [ + 0.5233949945593036 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011664829595211118, + "centered_acc": -0.01305767138193692, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.01305767138193692 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.257996632996633, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.257996632996633 + ], + "acc": 0.25715488215488214, + "acc_ci": 0.0, + "acc_runs": [ + 0.25715488215488214 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00897797000520329, + "centered_acc": 0.010662177328844002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.010662177328844002 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26109215017064846, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26109215017064846 + ], + "acc": 0.2098976109215017, + "acc_ci": 0.0, + "acc_runs": [ + 0.2098976109215017 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012835523909473708, + "centered_acc": 0.014789533560864617, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014789533560864617 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.3, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.3 + ], + "acc": 0.164, + "acc_ci": 0.0, + "acc_runs": [ + 0.164 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020514426225627987, + "centered_acc": 0.06666666666666665, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06666666666666665 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.48539857932123126, + "acc_ci": 0.0, + "acc_runs": [ + 0.48539857932123126 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014046492383275955, + "centered_acc": -0.02920284135753748, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.02920284135753748 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 549623.6452280454, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 549623.6452280454 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.171, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.171 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011912216456264588, + "centered_acc": -0.10533333333333332, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10533333333333332 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2895490438916075, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.288866031622315, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.017102384126435843, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2895490438916075 + ], + "mean_acc_runs": [ + 0.288866031622315 + ], + "mean_centered_acc_runs": [ + -0.017102384126435843 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-2880.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-2880.json new file mode 100644 index 0000000000000000000000000000000000000000..c82894d426746471d36265ea3485d07485ec4a0b --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-2880.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2880", + "checkpoint": "checkpoint-2880", + "global_step": 2880, + "tokens_seen": 3019898880, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.26030671181039633, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26030671181039633 + ], + "acc": 0.25473013343955386, + "acc_ci": 0.0, + "acc_runs": [ + 0.25473013343955386 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004379051357024372, + "centered_acc": 0.013742282413861773, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013742282413861773 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.528835690968444, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.528835690968444 + ], + "acc": 0.5331882480957563, + "acc_ci": 0.0, + "acc_runs": [ + 0.5331882480957563 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011646407809944788, + "centered_acc": 0.05767138193688792, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05767138193688792 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26515151515151514, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26515151515151514 + ], + "acc": 0.26557239057239057, + "acc_ci": 0.0, + "acc_runs": [ + 0.26557239057239057 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009057621139172679, + "centered_acc": 0.020202020202020183, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020202020202020183 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24061433447098976, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24061433447098976 + ], + "acc": 0.20051194539249148, + "acc_ci": 0.0, + "acc_runs": [ + 0.20051194539249148 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012491468532390682, + "centered_acc": -0.012514220705346987, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.012514220705346987 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.306, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.306 + ], + "acc": 0.168, + "acc_ci": 0.0, + "acc_runs": [ + 0.168 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020629569998345414, + "centered_acc": 0.07466666666666666, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.07466666666666666 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5067087608524072, + "acc_ci": 0.0, + "acc_runs": [ + 0.5067087608524072 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051220692330247, + "centered_acc": 0.013417521704814472, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013417521704814472 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.61, + "acc_ci": 0.0, + "acc_runs": [ + 0.61 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04902071300001973, + "centered_acc": 0.21999999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.21999999999999997 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 181099864.8703926, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 181099864.8703926 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.178, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.178 + ], + "acc": 0.042, + "acc_ci": 0.0, + "acc_runs": [ + 0.042 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012102167676183571, + "centered_acc": -0.09600000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09600000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29648470873355753, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29589989401262995, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.0053067306407005126, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29648470873355753 + ], + "mean_acc_runs": [ + 0.29589989401262995 + ], + "mean_centered_acc_runs": [ + 0.0053067306407005126 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-3168.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-3168.json new file mode 100644 index 0000000000000000000000000000000000000000..03f31b355b85cec17576cf4851b0f7dbed15ca47 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-3168.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-3168", + "checkpoint": "checkpoint-3168", + "global_step": 3168, + "tokens_seen": 3321888768, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2555267874925314, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2555267874925314 + ], + "acc": 0.2568213503286198, + "acc_ci": 0.0, + "acc_runs": [ + 0.2568213503286198 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00435265526368279, + "centered_acc": 0.007369049990041843, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.007369049990041843 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5201305767138193, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5201305767138193 + ], + "acc": 0.5244831338411317, + "acc_ci": 0.0, + "acc_runs": [ + 0.5244831338411317 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011656365410780507, + "centered_acc": 0.040261153427638696, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040261153427638696 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26851851851851855, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26851851851851855 + ], + "acc": 0.25841750841750843, + "acc_ci": 0.0, + "acc_runs": [ + 0.25841750841750843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009094042554994908, + "centered_acc": 0.024691358024691395, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.024691358024691395 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2440273037542662, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2440273037542662 + ], + "acc": 0.1962457337883959, + "acc_ci": 0.0, + "acc_runs": [ + 0.1962457337883959 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012551447627856151, + "centered_acc": -0.007963594994311726, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.007963594994311726 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.288, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.288 + ], + "acc": 0.18, + "acc_ci": 0.0, + "acc_runs": [ + 0.18 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020271503835075255, + "centered_acc": 0.05066666666666664, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05066666666666664 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4988161010260458, + "acc_ci": 0.0, + "acc_runs": [ + 0.4988161010260458 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052446290528871, + "centered_acc": -0.0023677979479084232, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0023677979479084232 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.67, + "acc_ci": 0.0, + "acc_runs": [ + 0.67 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04725815626252609, + "centered_acc": 0.3400000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.3400000000000001 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 262312736.31570685, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 262312736.31570685 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.178, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.178 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012102167676183571, + "centered_acc": -0.09600000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09600000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2923671977465226, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.3006071289175402, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.012581306523802125, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2923671977465226 + ], + "mean_acc_runs": [ + 0.3006071289175402 + ], + "mean_centered_acc_runs": [ + 0.012581306523802125 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-3456.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-3456.json new file mode 100644 index 0000000000000000000000000000000000000000..d618140b210b988207930b0360a308ac16037e45 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-3456.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-3456", + "checkpoint": "checkpoint-3456", + "global_step": 3456, + "tokens_seen": 3623878656, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2550288787094204, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2550288787094204 + ], + "acc": 0.2557259510057757, + "acc_ci": 0.0, + "acc_runs": [ + 0.2557259510057757 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004349866376069043, + "centered_acc": 0.006705171612560568, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006705171612560568 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5223068552774756, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5223068552774756 + ], + "acc": 0.5277475516866159, + "acc_ci": 0.0, + "acc_runs": [ + 0.5277475516866159 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011654208652596384, + "centered_acc": 0.044613710554951114, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.044613710554951114 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2702020202020202, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2702020202020202 + ], + "acc": 0.2588383838383838, + "acc_ci": 0.0, + "acc_runs": [ + 0.2588383838383838 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009112002229119622, + "centered_acc": 0.026936026936026963, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.026936026936026963 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24744027303754265, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24744027303754265 + ], + "acc": 0.19880546075085323, + "acc_ci": 0.0, + "acc_runs": [ + 0.19880546075085323 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01261035266329266, + "centered_acc": -0.0034129692832764644, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0034129692832764644 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.292, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.292 + ], + "acc": 0.176, + "acc_ci": 0.0, + "acc_runs": [ + 0.176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020354375480530096, + "centered_acc": 0.05599999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05599999999999997 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4956590370955012, + "acc_ci": 0.0, + "acc_runs": [ + 0.4956590370955012 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051956064076884, + "centered_acc": -0.008681925808997626, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.008681925808997626 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.64, + "acc_ci": 0.0, + "acc_runs": [ + 0.64 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.048241815132442176, + "centered_acc": 0.28, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.28 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 259234493.37969702, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 259234493.37969702 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.178, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.178 + ], + "acc": 0.042, + "acc_ci": 0.0, + "acc_runs": [ + 0.042 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012102167676183571, + "centered_acc": -0.09600000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09600000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2941630045377432, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.297306384615083, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.006970548617629464, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2941630045377432 + ], + "mean_acc_runs": [ + 0.297306384615083 + ], + "mean_centered_acc_runs": [ + 0.006970548617629464 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-3744.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-3744.json new file mode 100644 index 0000000000000000000000000000000000000000..42f389ddd3c7e9749e49bda7189c3395b957e893 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-3744.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-3744", + "checkpoint": "checkpoint-3744", + "global_step": 3744, + "tokens_seen": 3925868544, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2533359888468433, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2533359888468433 + ], + "acc": 0.255327623979287, + "acc_ci": 0.0, + "acc_runs": [ + 0.255327623979287 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004340328204135388, + "centered_acc": 0.0044479851291243655, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0044479851291243655 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5179542981501633, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5179542981501633 + ], + "acc": 0.529923830250272, + "acc_ci": 0.0, + "acc_runs": [ + 0.529923830250272 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011658300623287121, + "centered_acc": 0.0359085963003265, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0359085963003265 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26430976430976433, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26430976430976433 + ], + "acc": 0.2542087542087542, + "acc_ci": 0.0, + "acc_runs": [ + 0.2542087542087542 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009048410451862887, + "centered_acc": 0.019079685746352437, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019079685746352437 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.23378839590443687, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.23378839590443687 + ], + "acc": 0.19368600682593856, + "acc_ci": 0.0, + "acc_runs": [ + 0.19368600682593856 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01236822537850722, + "centered_acc": -0.021615472127417507, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.021615472127417507 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.288, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.288 + ], + "acc": 0.18, + "acc_ci": 0.0, + "acc_runs": [ + 0.18 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020271503835075255, + "centered_acc": 0.05066666666666664, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05066666666666664 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5074980268350434, + "acc_ci": 0.0, + "acc_runs": [ + 0.5074980268350434 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014050905521228646, + "centered_acc": 0.014996053670086829, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014996053670086829 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 244735075.0605765, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 244735075.0605765 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.178, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.178 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012102167676183571, + "centered_acc": -0.09600000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09600000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.28923140786853463, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2922931703872996, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0084379512297178, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.28923140786853463 + ], + "mean_acc_runs": [ + 0.2922931703872996 + ], + "mean_centered_acc_runs": [ + -0.0084379512297178 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-4032.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-4032.json new file mode 100644 index 0000000000000000000000000000000000000000..723a61300ef2a4704c175cda4efc81384b257cb4 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-4032.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4032", + "checkpoint": "checkpoint-4032", + "global_step": 4032, + "tokens_seen": 4227858432, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25692093208524197, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25692093208524197 + ], + "acc": 0.2555267874925314, + "acc_ci": 0.0, + "acc_runs": [ + 0.2555267874925314 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004360424536144852, + "centered_acc": 0.009227909446989294, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009227909446989294 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5223068552774756, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5223068552774756 + ], + "acc": 0.529379760609358, + "acc_ci": 0.0, + "acc_runs": [ + 0.529379760609358 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011654208652596384, + "centered_acc": 0.044613710554951114, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.044613710554951114 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26851851851851855, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26851851851851855 + ], + "acc": 0.25336700336700335, + "acc_ci": 0.0, + "acc_runs": [ + 0.25336700336700335 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009094042554994908, + "centered_acc": 0.024691358024691395, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.024691358024691395 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24146757679180889, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24146757679180889 + ], + "acc": 0.20392491467576793, + "acc_ci": 0.0, + "acc_runs": [ + 0.20392491467576793 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012506564839739446, + "centered_acc": -0.011376564277588153, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.011376564277588153 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.296, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.296 + ], + "acc": 0.178, + "acc_ci": 0.0, + "acc_runs": [ + 0.178 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020435342091896222, + "centered_acc": 0.061333333333333316, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.061333333333333316 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.49171270718232046, + "acc_ci": 0.0, + "acc_runs": [ + 0.49171270718232046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405055532282417, + "centered_acc": -0.016574585635359074, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.016574585635359074 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.61, + "acc_ci": 0.0, + "acc_runs": [ + 0.61 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04902071300001973, + "centered_acc": 0.21999999999999997, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.21999999999999997 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 283131681.85078496, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 283131681.85078496 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.183, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.183 + ], + "acc": 0.046, + "acc_ci": 0.0, + "acc_runs": [ + 0.046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012233587399477745, + "centered_acc": -0.08933333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.08933333333333333 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2947023137788409, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2946198635100681, + "mean_acc_ci": 0.0, + "mean_centered_acc": -9.369425987942726e-05, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2947023137788409 + ], + "mean_acc_runs": [ + 0.2946198635100681 + ], + "mean_centered_acc_runs": [ + -9.369425987942726e-05 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-4320.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-4320.json new file mode 100644 index 0000000000000000000000000000000000000000..785ce6add807566ff0eb3ae980e18865222dc377 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-4320.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4320", + "checkpoint": "checkpoint-4320", + "global_step": 4320, + "tokens_seen": 4529848320, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2551284604660426, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2551284604660426 + ], + "acc": 0.2557259510057757, + "acc_ci": 0.0, + "acc_runs": [ + 0.2557259510057757 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043504247506458175, + "centered_acc": 0.006837947288056793, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006837947288056793 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5228509249183896, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5228509249183896 + ], + "acc": 0.5326441784548422, + "acc_ci": 0.0, + "acc_runs": [ + 0.5326441784548422 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011653634832401168, + "centered_acc": 0.04570184983677916, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.04570184983677916 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26557239057239057, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26557239057239057 + ], + "acc": 0.25336700336700335, + "acc_ci": 0.0, + "acc_runs": [ + 0.25336700336700335 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00906221062697191, + "centered_acc": 0.020763187429854096, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020763187429854096 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2363481228668942, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2363481228668942 + ], + "acc": 0.20392491467576793, + "acc_ci": 0.0, + "acc_runs": [ + 0.20392491467576793 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012414960524301884, + "centered_acc": -0.018202502844141082, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.018202502844141082 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.29, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.29 + ], + "acc": 0.192, + "acc_ci": 0.0, + "acc_runs": [ + 0.192 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020313179231745138, + "centered_acc": 0.05333333333333331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05333333333333331 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4980268350434096, + "acc_ci": 0.0, + "acc_runs": [ + 0.4980268350434096 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052376259225643, + "centered_acc": -0.003946329913180779, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.003946329913180779 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.62, + "acc_ci": 0.0, + "acc_runs": [ + 0.62 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04878317312145634, + "centered_acc": 0.24, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.24 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 509003311.822311, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 509003311.822311 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.176, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.176 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012048616898597498, + "centered_acc": -0.09866666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09866666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2909833164706195, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29809763443204995, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.00026619355682616023, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2909833164706195 + ], + "mean_acc_runs": [ + 0.29809763443204995 + ], + "mean_centered_acc_runs": [ + 0.00026619355682616023 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-4608.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-4608.json new file mode 100644 index 0000000000000000000000000000000000000000..db78cc431cc2d4b1b3e1c28e9f7a614ac07429e2 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-4608.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4608", + "checkpoint": "checkpoint-4608", + "global_step": 4608, + "tokens_seen": 4831838208, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25363473411670984, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25363473411670984 + ], + "acc": 0.25433180641306513, + "acc_ci": 0.0, + "acc_runs": [ + 0.25433180641306513 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004342017709967513, + "centered_acc": 0.0048463121556131155, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0048463121556131155 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5184983677910773, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5184983677910773 + ], + "acc": 0.5277475516866159, + "acc_ci": 0.0, + "acc_runs": [ + 0.5277475516866159 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01165783758381818, + "centered_acc": 0.03699673558215455, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03699673558215455 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26851851851851855, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26851851851851855 + ], + "acc": 0.2537878787878788, + "acc_ci": 0.0, + "acc_runs": [ + 0.2537878787878788 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009094042554994908, + "centered_acc": 0.024691358024691395, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.024691358024691395 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.23464163822525597, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.23464163822525597 + ], + "acc": 0.19795221843003413, + "acc_ci": 0.0, + "acc_runs": [ + 0.19795221843003413 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01238387356076873, + "centered_acc": -0.020477815699658713, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.020477815699658713 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.176, + "acc_ci": 0.0, + "acc_runs": [ + 0.176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.489344909234412, + "acc_ci": 0.0, + "acc_runs": [ + 0.489344909234412 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014049294536290372, + "centered_acc": -0.02131018153117603, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.02131018153117603 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.6, + "acc_ci": 0.0, + "acc_runs": [ + 0.6 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0492365963917331, + "centered_acc": 0.19999999999999996, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.19999999999999996 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 522256372.93410695, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 522256372.93410695 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.179, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.179 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012128730605719054, + "centered_acc": -0.09466666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09466666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29004887644192695, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2920451826325706, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0072605927319602005, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29004887644192695 + ], + "mean_acc_runs": [ + 0.2920451826325706 + ], + "mean_centered_acc_runs": [ + -0.0072605927319602005 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-4896.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-4896.json new file mode 100644 index 0000000000000000000000000000000000000000..b0f6ac159d9222d6a35a4e4e8b923173d6fb030d --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-4896.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4896", + "checkpoint": "checkpoint-4896", + "global_step": 4896, + "tokens_seen": 5133828096, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2552280422226648, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2552280422226648 + ], + "acc": 0.2568213503286198, + "acc_ci": 0.0, + "acc_runs": [ + 0.2568213503286198 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004350982826580739, + "centered_acc": 0.006970722963553093, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006970722963553093 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5223068552774756, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5223068552774756 + ], + "acc": 0.5266594124047879, + "acc_ci": 0.0, + "acc_runs": [ + 0.5266594124047879 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011654208652596384, + "centered_acc": 0.044613710554951114, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.044613710554951114 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26641414141414144, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26641414141414144 + ], + "acc": 0.26052188552188554, + "acc_ci": 0.0, + "acc_runs": [ + 0.26052188552188554 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009071357971078843, + "centered_acc": 0.021885521885521914, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021885521885521914 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.23464163822525597, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.23464163822525597 + ], + "acc": 0.19880546075085323, + "acc_ci": 0.0, + "acc_runs": [ + 0.19880546075085323 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01238387356076873, + "centered_acc": -0.020477815699658713, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.020477815699658713 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.29, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.29 + ], + "acc": 0.178, + "acc_ci": 0.0, + "acc_runs": [ + 0.178 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020313179231745138, + "centered_acc": 0.05333333333333331, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.05333333333333331 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.49171270718232046, + "acc_ci": 0.0, + "acc_runs": [ + 0.49171270718232046 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405055532282417, + "centered_acc": -0.016574585635359074, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.016574585635359074 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.63, + "acc_ci": 0.0, + "acc_runs": [ + 0.63 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.048523658709390974, + "centered_acc": 0.26, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.26 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 701259286.0416783, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 701259286.0416783 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.176, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.176 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012048616898597498, + "centered_acc": -0.09866666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09866666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2907651128565896, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2964808277962167, + "mean_acc_ci": 0.0, + "mean_centered_acc": 0.0008510160314528441, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2907651128565896 + ], + "mean_acc_runs": [ + 0.2964808277962167 + ], + "mean_centered_acc_runs": [ + 0.0008510160314528441 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-5184.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-5184.json new file mode 100644 index 0000000000000000000000000000000000000000..2677b56ce8e85cb2b809db574c08e17681b05730 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-5184.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-5184", + "checkpoint": "checkpoint-5184", + "global_step": 5184, + "tokens_seen": 5435817984, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2575184226249751, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2575184226249751 + ], + "acc": 0.2568213503286198, + "acc_ci": 0.0, + "acc_runs": [ + 0.2568213503286198 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004363736410689568, + "centered_acc": 0.010024563499966796, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.010024563499966796 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5223068552774756, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5223068552774756 + ], + "acc": 0.5266594124047879, + "acc_ci": 0.0, + "acc_runs": [ + 0.5266594124047879 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011654208652596384, + "centered_acc": 0.044613710554951114, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.044613710554951114 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2748316498316498, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2748316498316498 + ], + "acc": 0.25925925925925924, + "acc_ci": 0.0, + "acc_runs": [ + 0.25925925925925924 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009160538115254864, + "centered_acc": 0.03310886644219976, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03310886644219976 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2380546075085324, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2380546075085324 + ], + "acc": 0.19539249146757678, + "acc_ci": 0.0, + "acc_runs": [ + 0.19539249146757678 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012445770028026118, + "centered_acc": -0.015927189988623452, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.015927189988623452 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.3, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.3 + ], + "acc": 0.18, + "acc_ci": 0.0, + "acc_runs": [ + 0.18 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020514426225627987, + "centered_acc": 0.06666666666666665, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06666666666666665 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.505130228887135, + "acc_ci": 0.0, + "acc_runs": [ + 0.505130228887135 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051745961790525, + "centered_acc": 0.010260457774269982, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.010260457774269982 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 578011929.8043449, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 578011929.8043449 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.04, + "acc_ci": 0.0, + "acc_runs": [ + 0.04 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2956185892071055, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2921550204121079, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0029642223892409566, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2956185892071055 + ], + "mean_acc_runs": [ + 0.2921550204121079 + ], + "mean_centered_acc_runs": [ + -0.0029642223892409566 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-5472.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-5472.json new file mode 100644 index 0000000000000000000000000000000000000000..6d967a921c56852f55ca2fade1cb5ec031c95196 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-5472.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-5472", + "checkpoint": "checkpoint-5472", + "global_step": 5472, + "tokens_seen": 5737807872, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2552280422226648, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2552280422226648 + ], + "acc": 0.2559251145190201, + "acc_ci": 0.0, + "acc_runs": [ + 0.2559251145190201 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004350982826580739, + "centered_acc": 0.006970722963553093, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006970722963553093 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5190424374319913, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5190424374319913 + ], + "acc": 0.5255712731229597, + "acc_ci": 0.0, + "acc_runs": [ + 0.5255712731229597 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011657360703051524, + "centered_acc": 0.0380848748639826, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0380848748639826 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2718855218855219, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2718855218855219 + ], + "acc": 0.2609427609427609, + "acc_ci": 0.0, + "acc_runs": [ + 0.2609427609427609 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009129795867310605, + "centered_acc": 0.02918069584736253, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02918069584736253 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.23720136518771331, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.23720136518771331 + ], + "acc": 0.19965870307167236, + "acc_ci": 0.0, + "acc_runs": [ + 0.19965870307167236 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01243039982926071, + "centered_acc": -0.017064846416382246, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.017064846416382246 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.17, + "acc_ci": 0.0, + "acc_runs": [ + 0.17 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5011838989739542, + "acc_ci": 0.0, + "acc_runs": [ + 0.5011838989739542 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052446290528871, + "centered_acc": 0.0023677979479084232, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0023677979479084232 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 728174583.5411192, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 728174583.5411192 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.174, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.174 + ], + "acc": 0.048, + "acc_ci": 0.0, + "acc_runs": [ + 0.048 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011994493230973449, + "centered_acc": -0.10133333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10133333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29055956112131526, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29295692124040673, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.006357684953278712, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29055956112131526 + ], + "mean_acc_runs": [ + 0.29295692124040673 + ], + "mean_centered_acc_runs": [ + -0.006357684953278712 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-576.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-576.json new file mode 100644 index 0000000000000000000000000000000000000000..364371ace8ec83ae9f786838b0f48172cd0b3da4 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-576.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-576", + "checkpoint": "checkpoint-576", + "global_step": 576, + "tokens_seen": 603979776, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2560246962756423, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2560246962756423 + ], + "acc": 0.2524397530372436, + "acc_ci": 0.0, + "acc_runs": [ + 0.2524397530372436 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004355436696716235, + "centered_acc": 0.008032928367523043, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008032928367523043 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5108813928182807, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5108813928182807 + ], + "acc": 0.5375408052230686, + "acc_ci": 0.0, + "acc_runs": [ + 0.5375408052230686 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011663061261117717, + "centered_acc": 0.02176278563656142, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02176278563656142 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26304713804713803, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26304713804713803 + ], + "acc": 0.2558922558922559, + "acc_ci": 0.0, + "acc_runs": [ + 0.2558922558922559 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009034514898865944, + "centered_acc": 0.017396184062850706, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017396184062850706 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.26706484641638223, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26706484641638223 + ], + "acc": 0.20819112627986347, + "acc_ci": 0.0, + "acc_runs": [ + 0.20819112627986347 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01292893319649643, + "centered_acc": 0.022753128555176305, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.022753128555176305 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.298, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.298 + ], + "acc": 0.166, + "acc_ci": 0.0, + "acc_runs": [ + 0.166 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020475118092988954, + "centered_acc": 0.06399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06399999999999999 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4956590370955012, + "acc_ci": 0.0, + "acc_runs": [ + 0.4956590370955012 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051956064076884, + "centered_acc": -0.008681925808997626, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.008681925808997626 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.57, + "acc_ci": 0.0, + "acc_runs": [ + 0.57 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049756985195624305, + "centered_acc": 0.1399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1399999999999999 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 3166132.9466289175, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 3166132.9466289175 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.179, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.179 + ], + "acc": 0.045, + "acc_ci": 0.0, + "acc_runs": [ + 0.045 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012128730605719054, + "centered_acc": -0.09466666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09466666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2956696789262406, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29090104393016325, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.008092071367350257, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2956696789262406 + ], + "mean_acc_runs": [ + 0.29090104393016325 + ], + "mean_centered_acc_runs": [ + -0.008092071367350257 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-5760.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-5760.json new file mode 100644 index 0000000000000000000000000000000000000000..c5059e7739691f81fac169785c171d863cbdf54d --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-5760.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-5760", + "checkpoint": "checkpoint-5760", + "global_step": 5760, + "tokens_seen": 6039797760, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2560246962756423, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2560246962756423 + ], + "acc": 0.2552280422226648, + "acc_ci": 0.0, + "acc_runs": [ + 0.2552280422226648 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004355436696716235, + "centered_acc": 0.008032928367523043, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.008032928367523043 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5119695321001088, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5119695321001088 + ], + "acc": 0.5272034820457019, + "acc_ci": 0.0, + "acc_runs": [ + 0.5272034820457019 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011662480968070035, + "centered_acc": 0.02393906420021752, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02393906420021752 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2769360269360269, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2769360269360269 + ], + "acc": 0.25673400673400676, + "acc_ci": 0.0, + "acc_runs": [ + 0.25673400673400676 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009182190173795887, + "centered_acc": 0.03591470258136923, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03591470258136923 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.23464163822525597, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.23464163822525597 + ], + "acc": 0.2022184300341297, + "acc_ci": 0.0, + "acc_runs": [ + 0.2022184300341297 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01238387356076873, + "centered_acc": -0.020477815699658713, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.020477815699658713 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.176, + "acc_ci": 0.0, + "acc_runs": [ + 0.176 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5098658247829518, + "acc_ci": 0.0, + "acc_runs": [ + 0.5098658247829518 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014049749833367674, + "centered_acc": 0.019731649565903675, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019731649565903675 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.6, + "acc_ci": 0.0, + "acc_runs": [ + 0.6 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0492365963917331, + "centered_acc": 0.19999999999999996, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.19999999999999996 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 924589344.7186648, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 924589344.7186648 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.178, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.178 + ], + "acc": 0.049, + "acc_ci": 0.0, + "acc_runs": [ + 0.049 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012102167676183571, + "centered_acc": -0.09600000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09600000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.28959531558950563, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29545372475931553, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.0035871719374716265, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.28959531558950563 + ], + "mean_acc_runs": [ + 0.29545372475931553 + ], + "mean_centered_acc_runs": [ + -0.0035871719374716265 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-6048.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-6048.json new file mode 100644 index 0000000000000000000000000000000000000000..79611133dd97395355ef846207d8bf92f78b6774 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-6048.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6048", + "checkpoint": "checkpoint-6048", + "global_step": 6048, + "tokens_seen": 6341787648, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2574188408683529, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2574188408683529 + ], + "acc": 0.25423222465644296, + "acc_ci": 0.0, + "acc_runs": [ + 0.25423222465644296 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043631851720466585, + "centered_acc": 0.009891787824470569, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009891787824470569 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5201305767138193, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5201305767138193 + ], + "acc": 0.5261153427638737, + "acc_ci": 0.0, + "acc_runs": [ + 0.5261153427638737 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011656365410780507, + "centered_acc": 0.040261153427638696, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040261153427638696 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26515151515151514, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26515151515151514 + ], + "acc": 0.2601010101010101, + "acc_ci": 0.0, + "acc_runs": [ + 0.2601010101010101 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009057621139172679, + "centered_acc": 0.020202020202020183, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020202020202020183 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24061433447098976, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24061433447098976 + ], + "acc": 0.20477815699658702, + "acc_ci": 0.0, + "acc_runs": [ + 0.20477815699658702 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012491468532390682, + "centered_acc": -0.012514220705346987, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.012514220705346987 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.298, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.298 + ], + "acc": 0.178, + "acc_ci": 0.0, + "acc_runs": [ + 0.178 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020475118092988954, + "centered_acc": 0.06399999999999999, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06399999999999999 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4996053670086819, + "acc_ci": 0.0, + "acc_runs": [ + 0.4996053670086819 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405248130604961, + "centered_acc": -0.0007892659826361781, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.0007892659826361781 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 1083041077.3872633, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 1083041077.3872633 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.181, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.181 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012181436179178025, + "centered_acc": -0.09200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09200000000000001 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29371921120077954, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2938119563300296, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.003819289076272574, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29371921120077954 + ], + "mean_acc_runs": [ + 0.2938119563300296 + ], + "mean_centered_acc_runs": [ + -0.003819289076272574 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-6336.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-6336.json new file mode 100644 index 0000000000000000000000000000000000000000..967be7a14780a478c2851f00b878e7662a7431ad --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-6336.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6336", + "checkpoint": "checkpoint-6336", + "global_step": 6336, + "tokens_seen": 6643777536, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2545309699263095, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2545309699263095 + ], + "acc": 0.25721967735510853, + "acc_ci": 0.0, + "acc_runs": [ + 0.25721967735510853 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043470700195274055, + "centered_acc": 0.006041293235079366, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.006041293235079366 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5179542981501633, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5179542981501633 + ], + "acc": 0.5342763873775843, + "acc_ci": 0.0, + "acc_runs": [ + 0.5342763873775843 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011658300623287121, + "centered_acc": 0.0359085963003265, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0359085963003265 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2647306397306397, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2647306397306397 + ], + "acc": 0.26136363636363635, + "acc_ci": 0.0, + "acc_runs": [ + 0.26136363636363635 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009053021086174166, + "centered_acc": 0.019640852974186274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019640852974186274 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.24146757679180889, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.24146757679180889 + ], + "acc": 0.2090443686006826, + "acc_ci": 0.0, + "acc_runs": [ + 0.2090443686006826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012506564839739446, + "centered_acc": -0.011376564277588153, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.011376564277588153 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.18, + "acc_ci": 0.0, + "acc_runs": [ + 0.18 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.49013417521704816, + "acc_ci": 0.0, + "acc_runs": [ + 0.49013417521704816 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014049749833367674, + "centered_acc": -0.019731649565903675, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.019731649565903675 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.56, + "acc_ci": 0.0, + "acc_runs": [ + 0.56 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049888765156985884, + "centered_acc": 0.1200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1200000000000001 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 808986448.4541316, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 808986448.4541316 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.178, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.178 + ], + "acc": 0.043, + "acc_ci": 0.0, + "acc_runs": [ + 0.043 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012102167676183571, + "centered_acc": -0.09600000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09600000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29044724743315353, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29133257066877605, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.015660283087388777, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29044724743315353 + ], + "mean_acc_runs": [ + 0.29133257066877605 + ], + "mean_centered_acc_runs": [ + -0.015660283087388777 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-6624.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-6624.json new file mode 100644 index 0000000000000000000000000000000000000000..35b8ba2db1db7ae4e9baad61b662dae7b30eb157 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-6624.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6624", + "checkpoint": "checkpoint-6624", + "global_step": 6624, + "tokens_seen": 6945767424, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2529376618203545, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2529376618203545 + ], + "acc": 0.25492929695279826, + "acc_ci": 0.0, + "acc_runs": [ + 0.25492929695279826 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.0043380713189125845, + "centered_acc": 0.003916882427139316, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.003916882427139316 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5087051142546246, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5087051142546246 + ], + "acc": 0.5261153427638737, + "acc_ci": 0.0, + "acc_runs": [ + 0.5261153427638737 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011664055982032772, + "centered_acc": 0.017410228509249226, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017410228509249226 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26557239057239057, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26557239057239057 + ], + "acc": 0.2563131313131313, + "acc_ci": 0.0, + "acc_runs": [ + 0.2563131313131313 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00906221062697191, + "centered_acc": 0.020763187429854096, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.020763187429854096 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2440273037542662, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2440273037542662 + ], + "acc": 0.19880546075085323, + "acc_ci": 0.0, + "acc_runs": [ + 0.19880546075085323 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012551447627856151, + "centered_acc": -0.007963594994311726, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.007963594994311726 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.274, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.274 + ], + "acc": 0.172, + "acc_ci": 0.0, + "acc_runs": [ + 0.172 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.019966103540279518, + "centered_acc": 0.03200000000000003, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.03200000000000003 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5067087608524072, + "acc_ci": 0.0, + "acc_runs": [ + 0.5067087608524072 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014051220692330247, + "centered_acc": 0.013417521704814472, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.013417521704814472 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.55, + "acc_ci": 0.0, + "acc_runs": [ + 0.55 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05, + "centered_acc": 0.10000000000000009, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.10000000000000009 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 1203424385.6715648, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 1203424385.6715648 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.172, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.172 + ], + "acc": 0.042, + "acc_ci": 0.0, + "acc_runs": [ + 0.042 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011939788882495308, + "centered_acc": -0.10400000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10400000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.28620707840027265, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2885159454406764, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.018653427930650432, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.28620707840027265 + ], + "mean_acc_runs": [ + 0.2885159454406764 + ], + "mean_centered_acc_runs": [ + -0.018653427930650432 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-6912.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-6912.json new file mode 100644 index 0000000000000000000000000000000000000000..8e9030dcffe97c6274343a8e642f1b9bfc0c4d15 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-6912.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6912", + "checkpoint": "checkpoint-6912", + "global_step": 6912, + "tokens_seen": 7247757312, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25731925911173076, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25731925911173076 + ], + "acc": 0.2546305516829317, + "acc_ci": 0.0, + "acc_runs": [ + 0.2546305516829317 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004362633637374433, + "centered_acc": 0.009759012148974344, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009759012148974344 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.514689880304679, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.514689880304679 + ], + "acc": 0.528835690968444, + "acc_ci": 0.0, + "acc_runs": [ + 0.528835690968444 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011660788281735552, + "centered_acc": 0.029379760609357986, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.029379760609357986 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.265993265993266, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.265993265993266 + ], + "acc": 0.255050505050505, + "acc_ci": 0.0, + "acc_runs": [ + 0.255050505050505 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009066789565615746, + "centered_acc": 0.021324354657688005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.021324354657688005 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.23208191126279865, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.23208191126279865 + ], + "acc": 0.2022184300341297, + "acc_ci": 0.0, + "acc_runs": [ + 0.2022184300341297 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012336718284948816, + "centered_acc": -0.023890784982935138, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.023890784982935138 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.284, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.284 + ], + "acc": 0.172, + "acc_ci": 0.0, + "acc_runs": [ + 0.172 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020186703693570774, + "centered_acc": 0.0453333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0453333333333333 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5122336227308603, + "acc_ci": 0.0, + "acc_runs": [ + 0.5122336227308603 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014048278820405729, + "centered_acc": 0.02446724546172052, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02446724546172052 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 1379801887.5433037, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 1379801887.5433037 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.176, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.176 + ], + "acc": 0.048, + "acc_ci": 0.0, + "acc_runs": [ + 0.048 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012048616898597498, + "centered_acc": -0.09866666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09866666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.28834738611207905, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.2931256262240571, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.008413202432347455, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.28834738611207905 + ], + "mean_acc_runs": [ + 0.2931256262240571 + ], + "mean_centered_acc_runs": [ + -0.008413202432347455 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-72.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-72.json new file mode 100644 index 0000000000000000000000000000000000000000..18d48c56636871c645c7bae8ad02e873f0d52fd8 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-72.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-72", + "checkpoint": "checkpoint-72", + "global_step": 72, + "tokens_seen": 75497472, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2591117307309301, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2591117307309301 + ], + "acc": 0.25632344154550885, + "acc_ci": 0.0, + "acc_runs": [ + 0.25632344154550885 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004372516060164899, + "centered_acc": 0.012148974307906771, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.012148974307906771 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5048966267682263, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5048966267682263 + ], + "acc": 0.5315560391730142, + "acc_ci": 0.0, + "acc_runs": [ + 0.5315560391730142 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011665264730078137, + "centered_acc": 0.009793253536452662, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.009793253536452662 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.26052188552188554, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.26052188552188554 + ], + "acc": 0.25, + "acc_ci": 0.0, + "acc_runs": [ + 0.25 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.00900643589033641, + "centered_acc": 0.01402918069584739, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01402918069584739 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2619453924914676, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2619453924914676 + ], + "acc": 0.197098976109215, + "acc_ci": 0.0, + "acc_runs": [ + 0.197098976109215 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012849054826857987, + "centered_acc": 0.015927189988623452, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.015927189988623452 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.284, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.284 + ], + "acc": 0.15, + "acc_ci": 0.0, + "acc_runs": [ + 0.15 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020186703693570774, + "centered_acc": 0.0453333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0453333333333333 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4964483030781373, + "acc_ci": 0.0, + "acc_runs": [ + 0.4964483030781373 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052131146915853, + "centered_acc": -0.007103393843725381, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.007103393843725381 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.53, + "acc_ci": 0.0, + "acc_runs": [ + 0.53 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.05016135580465919, + "centered_acc": 0.06000000000000005, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.06000000000000005 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 16124.486786954227, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 16124.486786954227 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.175, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.175 + ], + "acc": 0.041, + "acc_ci": 0.0, + "acc_runs": [ + 0.041 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012021627157731998, + "centered_acc": -0.10000000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.10000000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2909126059187516, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.28307142216795755, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.021477393159351234, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2909126059187516 + ], + "mean_acc_runs": [ + 0.28307142216795755 + ], + "mean_centered_acc_runs": [ + -0.021477393159351234 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-7200.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-7200.json new file mode 100644 index 0000000000000000000000000000000000000000..becd5f312895a931196e39eed0429c8a3bcd5708 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-7200.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-7200", + "checkpoint": "checkpoint-7200", + "global_step": 7200, + "tokens_seen": 7549747200, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.2583150766779526, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2583150766779526 + ], + "acc": 0.2574188408683529, + "acc_ci": 0.0, + "acc_runs": [ + 0.2574188408683529 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004368135676213571, + "centered_acc": 0.01108676890393682, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.01108676890393682 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5125136017410229, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5125136017410229 + ], + "acc": 0.5212187159956474, + "acc_ci": 0.0, + "acc_runs": [ + 0.5212187159956474 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011662170084916882, + "centered_acc": 0.02502720348204579, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02502720348204579 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2609427609427609, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2609427609427609 + ], + "acc": 0.2529461279461279, + "acc_ci": 0.0, + "acc_runs": [ + 0.2529461279461279 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009011142493235922, + "centered_acc": 0.014590347923681227, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.014590347923681227 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.23464163822525597, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.23464163822525597 + ], + "acc": 0.20733788395904437, + "acc_ci": 0.0, + "acc_runs": [ + 0.20733788395904437 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01238387356076873, + "centered_acc": -0.020477815699658713, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.020477815699658713 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.28, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.28 + ], + "acc": 0.166, + "acc_ci": 0.0, + "acc_runs": [ + 0.166 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020099950647503192, + "centered_acc": 0.040000000000000036, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.040000000000000036 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.5035516969218626, + "acc_ci": 0.0, + "acc_runs": [ + 0.5035516969218626 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052131146915852, + "centered_acc": 0.00710339384372527, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.00710339384372527 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.58, + "acc_ci": 0.0, + "acc_runs": [ + 0.58 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04960449637488582, + "centered_acc": 0.15999999999999992, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.15999999999999992 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 1695076667.3291082, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 1695076667.3291082 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.178, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.178 + ], + "acc": 0.044, + "acc_ci": 0.0, + "acc_runs": [ + 0.044 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012102167676183571, + "centered_acc": -0.09600000000000002, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09600000000000002 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.28740217959783204, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29107607274647357, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.011343908666541003, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.28740217959783204 + ], + "mean_acc_runs": [ + 0.29107607274647357 + ], + "mean_centered_acc_runs": [ + -0.011343908666541003 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-7488.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-7488.json new file mode 100644 index 0000000000000000000000000000000000000000..8c397f4bc15131e78afd8b31dcd0d159f2b30fc5 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-7488.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-7488", + "checkpoint": "checkpoint-7488", + "global_step": 7488, + "tokens_seen": 7851737088, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25403306114319857, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25403306114319857 + ], + "acc": 0.25363473411670984, + "acc_ci": 0.0, + "acc_runs": [ + 0.25363473411670984 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004344266179634878, + "centered_acc": 0.005377414857598091, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005377414857598091 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.514689880304679, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.514689880304679 + ], + "acc": 0.5201305767138193, + "acc_ci": 0.0, + "acc_runs": [ + 0.5201305767138193 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011660788281735552, + "centered_acc": 0.029379760609357986, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.029379760609357986 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2647306397306397, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2647306397306397 + ], + "acc": 0.25252525252525254, + "acc_ci": 0.0, + "acc_runs": [ + 0.25252525252525254 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009053021086174166, + "centered_acc": 0.019640852974186274, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.019640852974186274 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.23037542662116042, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.23037542662116042 + ], + "acc": 0.21075085324232082, + "acc_ci": 0.0, + "acc_runs": [ + 0.21075085324232082 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012304928418747618, + "centered_acc": -0.02616609783845277, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.02616609783845277 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.284, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.284 + ], + "acc": 0.154, + "acc_ci": 0.0, + "acc_runs": [ + 0.154 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020186703693570774, + "centered_acc": 0.0453333333333333, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0453333333333333 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.500394632991318, + "acc_ci": 0.0, + "acc_runs": [ + 0.500394632991318 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.01405248130604961, + "centered_acc": 0.0007892659826360671, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.0007892659826360671 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.59, + "acc_ci": 0.0, + "acc_runs": [ + 0.59 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.04943110704237104, + "centered_acc": 0.17999999999999994, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.17999999999999994 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 2016319301.444169, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 2016319301.444169 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.177, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.177 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012075463420375087, + "centered_acc": -0.09733333333333334, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09733333333333334 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.2874715012999463, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.29067235113631207, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.009600431096363757, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.2874715012999463 + ], + "mean_acc_runs": [ + 0.29067235113631207 + ], + "mean_centered_acc_runs": [ + -0.009600431096363757 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-864.json b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-864.json new file mode 100644 index 0000000000000000000000000000000000000000..7e851dfc517b69f46e1d8cf1eb3a20a86b869439 --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/checkpoint-864.json @@ -0,0 +1,261 @@ +{ + "run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", + "run": "psm25-spm25-fulle-lm", + "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-864", + "checkpoint": "checkpoint-864", + "global_step": 864, + "tokens_seen": 905969664, + "num_fewshot": 0, + "runs_per_task": 1, + "base_seed": 42, + "tasks_evaluated": [ + "hellaswag", + "piqa", + "arc_easy", + "arc_challenge", + "openbookqa", + "winogrande", + "boolq", + "copa", + "lambada_openai", + "sciq" + ], + "tasks": { + "hellaswag": { + "acc_norm": 0.25403306114319857, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.25403306114319857 + ], + "acc": 0.2546305516829317, + "acc_ci": 0.0, + "acc_runs": [ + 0.2546305516829317 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.004344266179634878, + "centered_acc": 0.005377414857598091, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.005377414857598091 + ], + "ci": 0.0 + }, + "piqa": { + "acc_norm": 0.5206746463547334, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.5206746463547334 + ], + "acc": 0.5364526659412405, + "acc_ci": 0.0, + "acc_runs": [ + 0.5364526659412405 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.011655846995729838, + "centered_acc": 0.041349292709466745, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.041349292709466745 + ], + "ci": 0.0 + }, + "arc_easy": { + "acc_norm": 0.2718855218855219, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2718855218855219 + ], + "acc": 0.26430976430976433, + "acc_ci": 0.0, + "acc_runs": [ + 0.26430976430976433 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.009129795867310605, + "centered_acc": 0.02918069584736253, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.02918069584736253 + ], + "ci": 0.0 + }, + "arc_challenge": { + "acc_norm": 0.2627986348122867, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.2627986348122867 + ], + "acc": 0.20819112627986347, + "acc_ci": 0.0, + "acc_runs": [ + 0.20819112627986347 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012862523175351397, + "centered_acc": 0.017064846416382284, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.017064846416382284 + ], + "ci": 0.0 + }, + "openbookqa": { + "acc_norm": 0.286, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.286 + ], + "acc": 0.164, + "acc_ci": 0.0, + "acc_runs": [ + 0.164 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.020229346329177604, + "centered_acc": 0.047999999999999966, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.047999999999999966 + ], + "ci": 0.0 + }, + "winogrande": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.4980268350434096, + "acc_ci": 0.0, + "acc_runs": [ + 0.4980268350434096 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.014052376259225643, + "centered_acc": -0.003946329913180779, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.003946329913180779 + ], + "ci": 0.0 + }, + "boolq": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.3782874617737003, + "acc_ci": 0.0, + "acc_runs": [ + 0.3782874617737003 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.008482001133931097, + "centered_acc": -0.24342507645259936, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.24342507645259936 + ], + "ci": 0.0 + }, + "copa": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.56, + "acc_ci": 0.0, + "acc_runs": [ + 0.56 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.049888765156985884, + "centered_acc": 0.1200000000000001, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + 0.1200000000000001 + ], + "ci": 0.0 + }, + "lambada_openai": { + "acc_norm": null, + "acc_norm_ci": 0.0, + "acc_norm_runs": [], + "acc": 0.0, + "acc_ci": 0.0, + "acc_runs": [ + 0.0 + ], + "perplexity": 6469188.807142904, + "perplexity_ci": 0.0, + "perplexity_runs": [ + 6469188.807142904 + ], + "stderr": 0.0, + "centered_acc": null, + "centered_acc_ci": 0.0, + "centered_acc_runs": [], + "ci": 0.0 + }, + "sciq": { + "acc_norm": 0.179, + "acc_norm_ci": 0.0, + "acc_norm_runs": [ + 0.179 + ], + "acc": 0.047, + "acc_ci": 0.0, + "acc_runs": [ + 0.047 + ], + "perplexity": null, + "perplexity_ci": 0.0, + "perplexity_runs": [], + "stderr": 0.012128730605719054, + "centered_acc": -0.09466666666666668, + "centered_acc_ci": 0.0, + "centered_acc_runs": [ + -0.09466666666666668 + ], + "ci": 0.0 + } + }, + "summary": { + "mean_acc_norm": 0.29573197736595674, + "mean_acc_norm_ci": 0.0, + "mean_acc": 0.291089840503091, + "mean_acc_ci": 0.0, + "mean_centered_acc": -0.00900731368907079, + "mean_centered_acc_ci": 0.0, + "mean_acc_norm_runs": [ + 0.29573197736595674 + ], + "mean_acc_runs": [ + 0.291089840503091 + ], + "mean_centered_acc_runs": [ + -0.00900731368907079 + ], + "ci": 0.0 + }, + "batch_size_per_device": 8, + "device": "cuda:0", + "dtype": "float16", + "next_i_enabled": false, + "max_next_i_train": 1, + "l2r_prefix_enabled": true +} diff --git a/results/psm25-spm25-fulle-lm/zero_shot_eval/summary.jsonl b/results/psm25-spm25-fulle-lm/zero_shot_eval/summary.jsonl new file mode 100644 index 0000000000000000000000000000000000000000..a613033f535392763efb3fc054036e3a901b346b --- /dev/null +++ b/results/psm25-spm25-fulle-lm/zero_shot_eval/summary.jsonl @@ -0,0 +1,27 @@ +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-72", "checkpoint": "checkpoint-72", "global_step": 72, "tokens_seen": 75497472, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2591117307309301, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2591117307309301], "acc": 0.25632344154550885, "acc_ci": 0.0, "acc_runs": [0.25632344154550885], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004372516060164899, "centered_acc": 0.012148974307906771, "centered_acc_ci": 0.0, "centered_acc_runs": [0.012148974307906771], "ci": 0.0}, "piqa": {"acc_norm": 0.5048966267682263, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5048966267682263], "acc": 0.5315560391730142, "acc_ci": 0.0, "acc_runs": [0.5315560391730142], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011665264730078137, "centered_acc": 0.009793253536452662, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009793253536452662], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26052188552188554, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26052188552188554], "acc": 0.25, "acc_ci": 0.0, "acc_runs": [0.25], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00900643589033641, "centered_acc": 0.01402918069584739, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01402918069584739], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2619453924914676, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2619453924914676], "acc": 0.197098976109215, "acc_ci": 0.0, "acc_runs": [0.197098976109215], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012849054826857987, "centered_acc": 0.015927189988623452, "centered_acc_ci": 0.0, "centered_acc_runs": [0.015927189988623452], "ci": 0.0}, "openbookqa": {"acc_norm": 0.284, "acc_norm_ci": 0.0, "acc_norm_runs": [0.284], "acc": 0.15, "acc_ci": 0.0, "acc_runs": [0.15], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020186703693570774, "centered_acc": 0.0453333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0453333333333333], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4964483030781373, "acc_ci": 0.0, "acc_runs": [0.4964483030781373], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052131146915853, "centered_acc": -0.007103393843725381, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.007103393843725381], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.53, "acc_ci": 0.0, "acc_runs": [0.53], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05016135580465919, "centered_acc": 0.06000000000000005, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06000000000000005], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 16124.486786954227, "perplexity_ci": 0.0, "perplexity_runs": [16124.486786954227], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.175, "acc_norm_ci": 0.0, "acc_norm_runs": [0.175], "acc": 0.041, "acc_ci": 0.0, "acc_runs": [0.041], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012021627157731998, "centered_acc": -0.10000000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10000000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2909126059187516, "mean_acc_norm_ci": 0.0, "mean_acc": 0.28307142216795755, "mean_acc_ci": 0.0, "mean_centered_acc": -0.021477393159351234, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2909126059187516], "mean_acc_runs": [0.28307142216795755], "mean_centered_acc_runs": [-0.021477393159351234], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-288", "checkpoint": "checkpoint-288", "global_step": 288, "tokens_seen": 301989888, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.253734315873332, "acc_norm_ci": 0.0, "acc_norm_runs": [0.253734315873332], "acc": 0.2555267874925314, "acc_ci": 0.0, "acc_runs": [0.2555267874925314], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004342580277662561, "centered_acc": 0.004979087831109341, "centered_acc_ci": 0.0, "centered_acc_runs": [0.004979087831109341], "ci": 0.0}, "piqa": {"acc_norm": 0.49347116430903154, "acc_norm_ci": 0.0, "acc_norm_runs": [0.49347116430903154], "acc": 0.5233949945593036, "acc_ci": 0.0, "acc_runs": [0.5233949945593036], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011664829595211118, "centered_acc": -0.01305767138193692, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.01305767138193692], "ci": 0.0}, "arc_easy": {"acc_norm": 0.257996632996633, "acc_norm_ci": 0.0, "acc_norm_runs": [0.257996632996633], "acc": 0.25715488215488214, "acc_ci": 0.0, "acc_runs": [0.25715488215488214], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00897797000520329, "centered_acc": 0.010662177328844002, "centered_acc_ci": 0.0, "centered_acc_runs": [0.010662177328844002], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26109215017064846, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26109215017064846], "acc": 0.2098976109215017, "acc_ci": 0.0, "acc_runs": [0.2098976109215017], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012835523909473708, "centered_acc": 0.014789533560864617, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014789533560864617], "ci": 0.0}, "openbookqa": {"acc_norm": 0.3, "acc_norm_ci": 0.0, "acc_norm_runs": [0.3], "acc": 0.164, "acc_ci": 0.0, "acc_runs": [0.164], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020514426225627987, "centered_acc": 0.06666666666666665, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06666666666666665], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48539857932123126, "acc_ci": 0.0, "acc_runs": [0.48539857932123126], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014046492383275955, "centered_acc": -0.02920284135753748, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.02920284135753748], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 549623.6452280454, "perplexity_ci": 0.0, "perplexity_runs": [549623.6452280454], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.171, "acc_norm_ci": 0.0, "acc_norm_runs": [0.171], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011912216456264588, "centered_acc": -0.10533333333333332, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10533333333333332], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2895490438916075, "mean_acc_norm_ci": 0.0, "mean_acc": 0.288866031622315, "mean_acc_ci": 0.0, "mean_centered_acc": -0.017102384126435843, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2895490438916075], "mean_acc_runs": [0.288866031622315], "mean_centered_acc_runs": [-0.017102384126435843], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-576", "checkpoint": "checkpoint-576", "global_step": 576, "tokens_seen": 603979776, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2560246962756423, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2560246962756423], "acc": 0.2524397530372436, "acc_ci": 0.0, "acc_runs": [0.2524397530372436], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004355436696716235, "centered_acc": 0.008032928367523043, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008032928367523043], "ci": 0.0}, "piqa": {"acc_norm": 0.5108813928182807, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5108813928182807], "acc": 0.5375408052230686, "acc_ci": 0.0, "acc_runs": [0.5375408052230686], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011663061261117717, "centered_acc": 0.02176278563656142, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02176278563656142], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26304713804713803, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26304713804713803], "acc": 0.2558922558922559, "acc_ci": 0.0, "acc_runs": [0.2558922558922559], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009034514898865944, "centered_acc": 0.017396184062850706, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017396184062850706], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.26706484641638223, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26706484641638223], "acc": 0.20819112627986347, "acc_ci": 0.0, "acc_runs": [0.20819112627986347], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01292893319649643, "centered_acc": 0.022753128555176305, "centered_acc_ci": 0.0, "centered_acc_runs": [0.022753128555176305], "ci": 0.0}, "openbookqa": {"acc_norm": 0.298, "acc_norm_ci": 0.0, "acc_norm_runs": [0.298], "acc": 0.166, "acc_ci": 0.0, "acc_runs": [0.166], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020475118092988954, "centered_acc": 0.06399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06399999999999999], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4956590370955012, "acc_ci": 0.0, "acc_runs": [0.4956590370955012], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051956064076884, "centered_acc": -0.008681925808997626, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.008681925808997626], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 3166132.9466289175, "perplexity_ci": 0.0, "perplexity_runs": [3166132.9466289175], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.179, "acc_norm_ci": 0.0, "acc_norm_runs": [0.179], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012128730605719054, "centered_acc": -0.09466666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09466666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2956696789262406, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29090104393016325, "mean_acc_ci": 0.0, "mean_centered_acc": -0.008092071367350257, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2956696789262406], "mean_acc_runs": [0.29090104393016325], "mean_centered_acc_runs": [-0.008092071367350257], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-864", "checkpoint": "checkpoint-864", "global_step": 864, "tokens_seen": 905969664, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25403306114319857, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25403306114319857], "acc": 0.2546305516829317, "acc_ci": 0.0, "acc_runs": [0.2546305516829317], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004344266179634878, "centered_acc": 0.005377414857598091, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005377414857598091], "ci": 0.0}, "piqa": {"acc_norm": 0.5206746463547334, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5206746463547334], "acc": 0.5364526659412405, "acc_ci": 0.0, "acc_runs": [0.5364526659412405], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011655846995729838, "centered_acc": 0.041349292709466745, "centered_acc_ci": 0.0, "centered_acc_runs": [0.041349292709466745], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2718855218855219, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2718855218855219], "acc": 0.26430976430976433, "acc_ci": 0.0, "acc_runs": [0.26430976430976433], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009129795867310605, "centered_acc": 0.02918069584736253, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02918069584736253], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2627986348122867, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2627986348122867], "acc": 0.20819112627986347, "acc_ci": 0.0, "acc_runs": [0.20819112627986347], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012862523175351397, "centered_acc": 0.017064846416382284, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017064846416382284], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.164, "acc_ci": 0.0, "acc_runs": [0.164], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4980268350434096, "acc_ci": 0.0, "acc_runs": [0.4980268350434096], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052376259225643, "centered_acc": -0.003946329913180779, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.003946329913180779], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.56, "acc_ci": 0.0, "acc_runs": [0.56], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049888765156985884, "centered_acc": 0.1200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1200000000000001], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 6469188.807142904, "perplexity_ci": 0.0, "perplexity_runs": [6469188.807142904], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.179, "acc_norm_ci": 0.0, "acc_norm_runs": [0.179], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012128730605719054, "centered_acc": -0.09466666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09466666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29573197736595674, "mean_acc_norm_ci": 0.0, "mean_acc": 0.291089840503091, "mean_acc_ci": 0.0, "mean_centered_acc": -0.00900731368907079, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29573197736595674], "mean_acc_runs": [0.291089840503091], "mean_centered_acc_runs": [-0.00900731368907079], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-1152", "checkpoint": "checkpoint-1152", "global_step": 1152, "tokens_seen": 1207959552, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2566221868153754, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2566221868153754], "acc": 0.25473013343955386, "acc_ci": 0.0, "acc_runs": [0.25473013343955386], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004358764596401022, "centered_acc": 0.008829582420500545, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008829582420500545], "ci": 0.0}, "piqa": {"acc_norm": 0.5179542981501633, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5179542981501633], "acc": 0.543525571273123, "acc_ci": 0.0, "acc_runs": [0.543525571273123], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011658300623287121, "centered_acc": 0.0359085963003265, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0359085963003265], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2718855218855219, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2718855218855219], "acc": 0.26430976430976433, "acc_ci": 0.0, "acc_runs": [0.26430976430976433], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009129795867310605, "centered_acc": 0.02918069584736253, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02918069584736253], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24914675767918087, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24914675767918087], "acc": 0.21160409556313994, "acc_ci": 0.0, "acc_runs": [0.21160409556313994], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012639407111926505, "centered_acc": -0.0011376564277588337, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0011376564277588337], "ci": 0.0}, "openbookqa": {"acc_norm": 0.284, "acc_norm_ci": 0.0, "acc_norm_runs": [0.284], "acc": 0.166, "acc_ci": 0.0, "acc_runs": [0.166], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020186703693570774, "centered_acc": 0.0453333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0453333333333333], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.47198105761641673, "acc_ci": 0.0, "acc_runs": [0.47198105761641673], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014030404213405703, "centered_acc": -0.056037884767166535, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.056037884767166535], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.57, "acc_ci": 0.0, "acc_runs": [0.57], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049756985195624305, "centered_acc": 0.1399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1399999999999999], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 12722601.352029193, "perplexity_ci": 0.0, "perplexity_runs": [12722601.352029193], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.187, "acc_norm_ci": 0.0, "acc_norm_runs": [0.187], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012336254828074166, "centered_acc": -0.084, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.084], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2944347940883736, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2906438083975698, "mean_acc_ci": 0.0, "mean_centered_acc": -0.013927601082889105, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2944347940883736], "mean_acc_runs": [0.2906438083975698], "mean_centered_acc_runs": [-0.013927601082889105], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-1440", "checkpoint": "checkpoint-1440", "global_step": 1440, "tokens_seen": 1509949440, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2566221868153754, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2566221868153754], "acc": 0.253734315873332, "acc_ci": 0.0, "acc_runs": [0.253734315873332], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004358764596401022, "centered_acc": 0.008829582420500545, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008829582420500545], "ci": 0.0}, "piqa": {"acc_norm": 0.5266594124047879, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5266594124047879], "acc": 0.5369967355821545, "acc_ci": 0.0, "acc_runs": [0.5369967355821545], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01164922999434743, "centered_acc": 0.05331882480957573, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05331882480957573], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26725589225589225, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26725589225589225], "acc": 0.2596801346801347, "acc_ci": 0.0, "acc_runs": [0.2596801346801347], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009080463246017426, "centered_acc": 0.023007856341189663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023007856341189663], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.25, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25], "acc": 0.21245733788395904, "acc_ci": 0.0, "acc_runs": [0.21245733788395904], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012653835621466646, "centered_acc": 0.0, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0], "ci": 0.0}, "openbookqa": {"acc_norm": 0.29, "acc_norm_ci": 0.0, "acc_norm_runs": [0.29], "acc": 0.164, "acc_ci": 0.0, "acc_runs": [0.164], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020313179231745138, "centered_acc": 0.05333333333333331, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05333333333333331], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48855564325177586, "acc_ci": 0.0, "acc_runs": [0.48855564325177586], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014048804199859391, "centered_acc": -0.022888713496448276, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.022888713496448276], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.00019406171162429653, "acc_ci": 0.0, "acc_runs": [0.00019406171162429653], "perplexity": 19299504.75313499, "perplexity_ci": 0.0, "perplexity_runs": [19299504.75313499], "stderr": 0.00019406171162429653, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.177, "acc_norm_ci": 0.0, "acc_norm_runs": [0.177], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012075463420375087, "centered_acc": -0.09733333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09733333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2945895819126759, "mean_acc_norm_ci": 0.0, "mean_acc": 0.291890569075668, "mean_acc_ci": 0.0, "mean_centered_acc": -0.007239725153086869, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2945895819126759], "mean_acc_runs": [0.291890569075668], "mean_centered_acc_runs": [-0.007239725153086869], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-1728", "checkpoint": "checkpoint-1728", "global_step": 1728, "tokens_seen": 1811939328, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25144393547102173, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25144393547102173], "acc": 0.25492929695279826, "acc_ci": 0.0, "acc_runs": [0.25492929695279826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043295650165269565, "centered_acc": 0.0019252472946956385, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0019252472946956385], "ci": 0.0}, "piqa": {"acc_norm": 0.5359085963003264, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5359085963003264], "acc": 0.5424374319912949, "acc_ci": 0.0, "acc_runs": [0.5424374319912949], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011635700809215522, "centered_acc": 0.07181719260065278, "centered_acc_ci": 0.0, "centered_acc_runs": [0.07181719260065278], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2735690235690236, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2735690235690236], "acc": 0.265993265993266, "acc_ci": 0.0, "acc_runs": [0.265993265993266], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009147424438490918, "centered_acc": 0.031425364758698095, "centered_acc_ci": 0.0, "centered_acc_runs": [0.031425364758698095], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24744027303754265, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24744027303754265], "acc": 0.2158703071672355, "acc_ci": 0.0, "acc_runs": [0.2158703071672355], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01261035266329266, "centered_acc": -0.0034129692832764644, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0034129692832764644], "ci": 0.0}, "openbookqa": {"acc_norm": 0.292, "acc_norm_ci": 0.0, "acc_norm_runs": [0.292], "acc": 0.16, "acc_ci": 0.0, "acc_runs": [0.16], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020354375480530096, "centered_acc": 0.05599999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05599999999999997], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4940805051302289, "acc_ci": 0.0, "acc_runs": [0.4940805051302289], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051500838485925, "centered_acc": -0.011838989739542227, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.011838989739542227], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 26026105.85764453, "perplexity_ci": 0.0, "perplexity_runs": [26026105.85764453], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.045, "acc_ci": 0.0, "acc_runs": [0.045], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29689363806298574, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2936598269008524, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0032788034245968523, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29689363806298574], "mean_acc_runs": [0.2936598269008524], "mean_centered_acc_runs": [-0.0032788034245968523], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2016", "checkpoint": "checkpoint-2016", "global_step": 2016, "tokens_seen": 2113929216, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2544313881696873, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2544313881696873], "acc": 0.25273849830711015, "acc_ci": 0.0, "acc_runs": [0.25273849830711015], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00434650985067984, "centered_acc": 0.005908517559583067, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005908517559583067], "ci": 0.0}, "piqa": {"acc_norm": 0.5321001088139282, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5321001088139282], "acc": 0.5386289445048966, "acc_ci": 0.0, "acc_runs": [0.5386289445048966], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011641758014820223, "centered_acc": 0.06420021762785644, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06420021762785644], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26725589225589225, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26725589225589225], "acc": 0.255050505050505, "acc_ci": 0.0, "acc_runs": [0.255050505050505], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009080463246017426, "centered_acc": 0.023007856341189663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023007856341189663], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24744027303754265, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24744027303754265], "acc": 0.20051194539249148, "acc_ci": 0.0, "acc_runs": [0.20051194539249148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01261035266329266, "centered_acc": -0.0034129692832764644, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0034129692832764644], "ci": 0.0}, "openbookqa": {"acc_norm": 0.29, "acc_norm_ci": 0.0, "acc_norm_runs": [0.29], "acc": 0.17, "acc_ci": 0.0, "acc_runs": [0.17], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020313179231745138, "centered_acc": 0.05333333333333331, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05333333333333331], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5217048145224941, "acc_ci": 0.0, "acc_runs": [0.5217048145224941], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014039239216484667, "centered_acc": 0.04340962904498813, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04340962904498813], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.6, "acc_ci": 0.0, "acc_runs": [0.6], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0492365963917331, "centered_acc": 0.19999999999999996, "centered_acc_ci": 0.0, "centered_acc_runs": [0.19999999999999996], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 48697070.09879444, "perplexity_ci": 0.0, "perplexity_runs": [48697070.09879444], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.173, "acc_norm_ci": 0.0, "acc_norm_runs": [0.173], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011967214137559927, "centered_acc": -0.10266666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10266666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29403794371284175, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29629221695511976, "mean_acc_ci": 0.0, "mean_centered_acc": 0.00448387127826756, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29403794371284175], "mean_acc_runs": [0.29629221695511976], "mean_centered_acc_runs": [0.00448387127826756], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2304", "checkpoint": "checkpoint-2304", "global_step": 2304, "tokens_seen": 2415919104, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25403306114319857, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25403306114319857], "acc": 0.2546305516829317, "acc_ci": 0.0, "acc_runs": [0.2546305516829317], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004344266179634878, "centered_acc": 0.005377414857598091, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005377414857598091], "ci": 0.0}, "piqa": {"acc_norm": 0.528835690968444, "acc_norm_ci": 0.0, "acc_norm_runs": [0.528835690968444], "acc": 0.5408052230685527, "acc_ci": 0.0, "acc_runs": [0.5408052230685527], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011646407809944788, "centered_acc": 0.05767138193688792, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05767138193688792], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26725589225589225, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26725589225589225], "acc": 0.265993265993266, "acc_ci": 0.0, "acc_runs": [0.265993265993266], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009080463246017426, "centered_acc": 0.023007856341189663, "centered_acc_ci": 0.0, "centered_acc_runs": [0.023007856341189663], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.23293515358361774, "acc_norm_ci": 0.0, "acc_norm_runs": [0.23293515358361774], "acc": 0.20392491467576793, "acc_ci": 0.0, "acc_runs": [0.20392491467576793], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012352507042617355, "centered_acc": -0.022753128555176343, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.022753128555176343], "ci": 0.0}, "openbookqa": {"acc_norm": 0.3, "acc_norm_ci": 0.0, "acc_norm_runs": [0.3], "acc": 0.176, "acc_ci": 0.0, "acc_runs": [0.176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020514426225627987, "centered_acc": 0.06666666666666665, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06666666666666665], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.48855564325177586, "acc_ci": 0.0, "acc_runs": [0.48855564325177586], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014048804199859391, "centered_acc": -0.022888713496448276, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.022888713496448276], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.61, "acc_ci": 0.0, "acc_runs": [0.61], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04902071300001973, "centered_acc": 0.21999999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.21999999999999997], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 81215739.36168465, "perplexity_ci": 0.0, "perplexity_runs": [81215739.36168465], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.176, "acc_norm_ci": 0.0, "acc_norm_runs": [0.176], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012048616898597498, "centered_acc": -0.09866666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09866666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29317663299185875, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29641970604459944, "mean_acc_ci": 0.0, "mean_centered_acc": -0.001667807263172041, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29317663299185875], "mean_acc_runs": [0.29641970604459944], "mean_centered_acc_runs": [-0.001667807263172041], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2592", "checkpoint": "checkpoint-2592", "global_step": 2592, "tokens_seen": 2717908992, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2545309699263095, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2545309699263095], "acc": 0.25542720573590916, "acc_ci": 0.0, "acc_runs": [0.25542720573590916], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043470700195274055, "centered_acc": 0.006041293235079366, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006041293235079366], "ci": 0.0}, "piqa": {"acc_norm": 0.5244831338411317, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5244831338411317], "acc": 0.5310119695321001, "acc_ci": 0.0, "acc_runs": [0.5310119695321001], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011651830225710194, "centered_acc": 0.04896626768226331, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04896626768226331], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2668350168350168, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2668350168350168], "acc": 0.257996632996633, "acc_ci": 0.0, "acc_runs": [0.257996632996633], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009075915859266983, "centered_acc": 0.02244668911335575, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02244668911335575], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24061433447098976, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24061433447098976], "acc": 0.20733788395904437, "acc_ci": 0.0, "acc_runs": [0.20733788395904437], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012491468532390682, "centered_acc": -0.012514220705346987, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.012514220705346987], "ci": 0.0}, "openbookqa": {"acc_norm": 0.288, "acc_norm_ci": 0.0, "acc_norm_runs": [0.288], "acc": 0.16, "acc_ci": 0.0, "acc_runs": [0.16], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020271503835075255, "centered_acc": 0.05066666666666664, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05066666666666664], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4861878453038674, "acc_ci": 0.0, "acc_runs": [0.4861878453038674], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014047122916440271, "centered_acc": -0.027624309392265234, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.027624309392265234], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 98230320.95594785, "perplexity_ci": 0.0, "perplexity_runs": [98230320.95594785], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.18, "acc_norm_ci": 0.0, "acc_norm_runs": [0.18], "acc": 0.048, "acc_ci": 0.0, "acc_runs": [0.048], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012155153135512024, "centered_acc": -0.09333333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09333333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2924105758455746, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29142489993012544, "mean_acc_ci": 0.0, "mean_centered_acc": -0.007641780354019991, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2924105758455746], "mean_acc_runs": [0.29142489993012544], "mean_centered_acc_runs": [-0.007641780354019991], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-2880", "checkpoint": "checkpoint-2880", "global_step": 2880, "tokens_seen": 3019898880, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.26030671181039633, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26030671181039633], "acc": 0.25473013343955386, "acc_ci": 0.0, "acc_runs": [0.25473013343955386], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004379051357024372, "centered_acc": 0.013742282413861773, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013742282413861773], "ci": 0.0}, "piqa": {"acc_norm": 0.528835690968444, "acc_norm_ci": 0.0, "acc_norm_runs": [0.528835690968444], "acc": 0.5331882480957563, "acc_ci": 0.0, "acc_runs": [0.5331882480957563], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011646407809944788, "centered_acc": 0.05767138193688792, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05767138193688792], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26515151515151514, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26515151515151514], "acc": 0.26557239057239057, "acc_ci": 0.0, "acc_runs": [0.26557239057239057], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009057621139172679, "centered_acc": 0.020202020202020183, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020202020202020183], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24061433447098976, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24061433447098976], "acc": 0.20051194539249148, "acc_ci": 0.0, "acc_runs": [0.20051194539249148], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012491468532390682, "centered_acc": -0.012514220705346987, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.012514220705346987], "ci": 0.0}, "openbookqa": {"acc_norm": 0.306, "acc_norm_ci": 0.0, "acc_norm_runs": [0.306], "acc": 0.168, "acc_ci": 0.0, "acc_runs": [0.168], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020629569998345414, "centered_acc": 0.07466666666666666, "centered_acc_ci": 0.0, "centered_acc_runs": [0.07466666666666666], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5067087608524072, "acc_ci": 0.0, "acc_runs": [0.5067087608524072], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051220692330247, "centered_acc": 0.013417521704814472, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013417521704814472], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.61, "acc_ci": 0.0, "acc_runs": [0.61], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04902071300001973, "centered_acc": 0.21999999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.21999999999999997], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 181099864.8703926, "perplexity_ci": 0.0, "perplexity_runs": [181099864.8703926], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.178, "acc_norm_ci": 0.0, "acc_norm_runs": [0.178], "acc": 0.042, "acc_ci": 0.0, "acc_runs": [0.042], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012102167676183571, "centered_acc": -0.09600000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09600000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29648470873355753, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29589989401262995, "mean_acc_ci": 0.0, "mean_centered_acc": 0.0053067306407005126, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29648470873355753], "mean_acc_runs": [0.29589989401262995], "mean_centered_acc_runs": [0.0053067306407005126], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-3168", "checkpoint": "checkpoint-3168", "global_step": 3168, "tokens_seen": 3321888768, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2555267874925314, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2555267874925314], "acc": 0.2568213503286198, "acc_ci": 0.0, "acc_runs": [0.2568213503286198], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00435265526368279, "centered_acc": 0.007369049990041843, "centered_acc_ci": 0.0, "centered_acc_runs": [0.007369049990041843], "ci": 0.0}, "piqa": {"acc_norm": 0.5201305767138193, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5201305767138193], "acc": 0.5244831338411317, "acc_ci": 0.0, "acc_runs": [0.5244831338411317], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011656365410780507, "centered_acc": 0.040261153427638696, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040261153427638696], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26851851851851855, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26851851851851855], "acc": 0.25841750841750843, "acc_ci": 0.0, "acc_runs": [0.25841750841750843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009094042554994908, "centered_acc": 0.024691358024691395, "centered_acc_ci": 0.0, "centered_acc_runs": [0.024691358024691395], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2440273037542662, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2440273037542662], "acc": 0.1962457337883959, "acc_ci": 0.0, "acc_runs": [0.1962457337883959], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012551447627856151, "centered_acc": -0.007963594994311726, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.007963594994311726], "ci": 0.0}, "openbookqa": {"acc_norm": 0.288, "acc_norm_ci": 0.0, "acc_norm_runs": [0.288], "acc": 0.18, "acc_ci": 0.0, "acc_runs": [0.18], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020271503835075255, "centered_acc": 0.05066666666666664, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05066666666666664], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4988161010260458, "acc_ci": 0.0, "acc_runs": [0.4988161010260458], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052446290528871, "centered_acc": -0.0023677979479084232, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0023677979479084232], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.67, "acc_ci": 0.0, "acc_runs": [0.67], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04725815626252609, "centered_acc": 0.3400000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [0.3400000000000001], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 262312736.31570685, "perplexity_ci": 0.0, "perplexity_runs": [262312736.31570685], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.178, "acc_norm_ci": 0.0, "acc_norm_runs": [0.178], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012102167676183571, "centered_acc": -0.09600000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09600000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2923671977465226, "mean_acc_norm_ci": 0.0, "mean_acc": 0.3006071289175402, "mean_acc_ci": 0.0, "mean_centered_acc": 0.012581306523802125, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2923671977465226], "mean_acc_runs": [0.3006071289175402], "mean_centered_acc_runs": [0.012581306523802125], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-3456", "checkpoint": "checkpoint-3456", "global_step": 3456, "tokens_seen": 3623878656, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2550288787094204, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2550288787094204], "acc": 0.2557259510057757, "acc_ci": 0.0, "acc_runs": [0.2557259510057757], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004349866376069043, "centered_acc": 0.006705171612560568, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006705171612560568], "ci": 0.0}, "piqa": {"acc_norm": 0.5223068552774756, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5223068552774756], "acc": 0.5277475516866159, "acc_ci": 0.0, "acc_runs": [0.5277475516866159], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011654208652596384, "centered_acc": 0.044613710554951114, "centered_acc_ci": 0.0, "centered_acc_runs": [0.044613710554951114], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2702020202020202, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2702020202020202], "acc": 0.2588383838383838, "acc_ci": 0.0, "acc_runs": [0.2588383838383838], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009112002229119622, "centered_acc": 0.026936026936026963, "centered_acc_ci": 0.0, "centered_acc_runs": [0.026936026936026963], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24744027303754265, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24744027303754265], "acc": 0.19880546075085323, "acc_ci": 0.0, "acc_runs": [0.19880546075085323], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01261035266329266, "centered_acc": -0.0034129692832764644, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0034129692832764644], "ci": 0.0}, "openbookqa": {"acc_norm": 0.292, "acc_norm_ci": 0.0, "acc_norm_runs": [0.292], "acc": 0.176, "acc_ci": 0.0, "acc_runs": [0.176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020354375480530096, "centered_acc": 0.05599999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05599999999999997], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4956590370955012, "acc_ci": 0.0, "acc_runs": [0.4956590370955012], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051956064076884, "centered_acc": -0.008681925808997626, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.008681925808997626], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.64, "acc_ci": 0.0, "acc_runs": [0.64], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.048241815132442176, "centered_acc": 0.28, "centered_acc_ci": 0.0, "centered_acc_runs": [0.28], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 259234493.37969702, "perplexity_ci": 0.0, "perplexity_runs": [259234493.37969702], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.178, "acc_norm_ci": 0.0, "acc_norm_runs": [0.178], "acc": 0.042, "acc_ci": 0.0, "acc_runs": [0.042], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012102167676183571, "centered_acc": -0.09600000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09600000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2941630045377432, "mean_acc_norm_ci": 0.0, "mean_acc": 0.297306384615083, "mean_acc_ci": 0.0, "mean_centered_acc": 0.006970548617629464, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2941630045377432], "mean_acc_runs": [0.297306384615083], "mean_centered_acc_runs": [0.006970548617629464], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-3744", "checkpoint": "checkpoint-3744", "global_step": 3744, "tokens_seen": 3925868544, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2533359888468433, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2533359888468433], "acc": 0.255327623979287, "acc_ci": 0.0, "acc_runs": [0.255327623979287], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004340328204135388, "centered_acc": 0.0044479851291243655, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0044479851291243655], "ci": 0.0}, "piqa": {"acc_norm": 0.5179542981501633, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5179542981501633], "acc": 0.529923830250272, "acc_ci": 0.0, "acc_runs": [0.529923830250272], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011658300623287121, "centered_acc": 0.0359085963003265, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0359085963003265], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26430976430976433, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26430976430976433], "acc": 0.2542087542087542, "acc_ci": 0.0, "acc_runs": [0.2542087542087542], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009048410451862887, "centered_acc": 0.019079685746352437, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019079685746352437], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.23378839590443687, "acc_norm_ci": 0.0, "acc_norm_runs": [0.23378839590443687], "acc": 0.19368600682593856, "acc_ci": 0.0, "acc_runs": [0.19368600682593856], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01236822537850722, "centered_acc": -0.021615472127417507, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.021615472127417507], "ci": 0.0}, "openbookqa": {"acc_norm": 0.288, "acc_norm_ci": 0.0, "acc_norm_runs": [0.288], "acc": 0.18, "acc_ci": 0.0, "acc_runs": [0.18], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020271503835075255, "centered_acc": 0.05066666666666664, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05066666666666664], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5074980268350434, "acc_ci": 0.0, "acc_runs": [0.5074980268350434], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014050905521228646, "centered_acc": 0.014996053670086829, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014996053670086829], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 244735075.0605765, "perplexity_ci": 0.0, "perplexity_runs": [244735075.0605765], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.178, "acc_norm_ci": 0.0, "acc_norm_runs": [0.178], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012102167676183571, "centered_acc": -0.09600000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09600000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.28923140786853463, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2922931703872996, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0084379512297178, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.28923140786853463], "mean_acc_runs": [0.2922931703872996], "mean_centered_acc_runs": [-0.0084379512297178], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4032", "checkpoint": "checkpoint-4032", "global_step": 4032, "tokens_seen": 4227858432, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25692093208524197, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25692093208524197], "acc": 0.2555267874925314, "acc_ci": 0.0, "acc_runs": [0.2555267874925314], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004360424536144852, "centered_acc": 0.009227909446989294, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009227909446989294], "ci": 0.0}, "piqa": {"acc_norm": 0.5223068552774756, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5223068552774756], "acc": 0.529379760609358, "acc_ci": 0.0, "acc_runs": [0.529379760609358], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011654208652596384, "centered_acc": 0.044613710554951114, "centered_acc_ci": 0.0, "centered_acc_runs": [0.044613710554951114], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26851851851851855, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26851851851851855], "acc": 0.25336700336700335, "acc_ci": 0.0, "acc_runs": [0.25336700336700335], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009094042554994908, "centered_acc": 0.024691358024691395, "centered_acc_ci": 0.0, "centered_acc_runs": [0.024691358024691395], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24146757679180889, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24146757679180889], "acc": 0.20392491467576793, "acc_ci": 0.0, "acc_runs": [0.20392491467576793], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012506564839739446, "centered_acc": -0.011376564277588153, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.011376564277588153], "ci": 0.0}, "openbookqa": {"acc_norm": 0.296, "acc_norm_ci": 0.0, "acc_norm_runs": [0.296], "acc": 0.178, "acc_ci": 0.0, "acc_runs": [0.178], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020435342091896222, "centered_acc": 0.061333333333333316, "centered_acc_ci": 0.0, "centered_acc_runs": [0.061333333333333316], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.49171270718232046, "acc_ci": 0.0, "acc_runs": [0.49171270718232046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405055532282417, "centered_acc": -0.016574585635359074, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.016574585635359074], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.61, "acc_ci": 0.0, "acc_runs": [0.61], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04902071300001973, "centered_acc": 0.21999999999999997, "centered_acc_ci": 0.0, "centered_acc_runs": [0.21999999999999997], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 283131681.85078496, "perplexity_ci": 0.0, "perplexity_runs": [283131681.85078496], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.183, "acc_norm_ci": 0.0, "acc_norm_runs": [0.183], "acc": 0.046, "acc_ci": 0.0, "acc_runs": [0.046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012233587399477745, "centered_acc": -0.08933333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.08933333333333333], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2947023137788409, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2946198635100681, "mean_acc_ci": 0.0, "mean_centered_acc": -9.369425987942726e-05, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2947023137788409], "mean_acc_runs": [0.2946198635100681], "mean_centered_acc_runs": [-9.369425987942726e-05], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4320", "checkpoint": "checkpoint-4320", "global_step": 4320, "tokens_seen": 4529848320, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2551284604660426, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2551284604660426], "acc": 0.2557259510057757, "acc_ci": 0.0, "acc_runs": [0.2557259510057757], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043504247506458175, "centered_acc": 0.006837947288056793, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006837947288056793], "ci": 0.0}, "piqa": {"acc_norm": 0.5228509249183896, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5228509249183896], "acc": 0.5326441784548422, "acc_ci": 0.0, "acc_runs": [0.5326441784548422], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011653634832401168, "centered_acc": 0.04570184983677916, "centered_acc_ci": 0.0, "centered_acc_runs": [0.04570184983677916], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26557239057239057, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26557239057239057], "acc": 0.25336700336700335, "acc_ci": 0.0, "acc_runs": [0.25336700336700335], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00906221062697191, "centered_acc": 0.020763187429854096, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020763187429854096], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2363481228668942, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2363481228668942], "acc": 0.20392491467576793, "acc_ci": 0.0, "acc_runs": [0.20392491467576793], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012414960524301884, "centered_acc": -0.018202502844141082, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.018202502844141082], "ci": 0.0}, "openbookqa": {"acc_norm": 0.29, "acc_norm_ci": 0.0, "acc_norm_runs": [0.29], "acc": 0.192, "acc_ci": 0.0, "acc_runs": [0.192], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020313179231745138, "centered_acc": 0.05333333333333331, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05333333333333331], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4980268350434096, "acc_ci": 0.0, "acc_runs": [0.4980268350434096], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052376259225643, "centered_acc": -0.003946329913180779, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.003946329913180779], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.62, "acc_ci": 0.0, "acc_runs": [0.62], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04878317312145634, "centered_acc": 0.24, "centered_acc_ci": 0.0, "centered_acc_runs": [0.24], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 509003311.822311, "perplexity_ci": 0.0, "perplexity_runs": [509003311.822311], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.176, "acc_norm_ci": 0.0, "acc_norm_runs": [0.176], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012048616898597498, "centered_acc": -0.09866666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09866666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2909833164706195, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29809763443204995, "mean_acc_ci": 0.0, "mean_centered_acc": 0.00026619355682616023, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2909833164706195], "mean_acc_runs": [0.29809763443204995], "mean_centered_acc_runs": [0.00026619355682616023], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4608", "checkpoint": "checkpoint-4608", "global_step": 4608, "tokens_seen": 4831838208, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25363473411670984, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25363473411670984], "acc": 0.25433180641306513, "acc_ci": 0.0, "acc_runs": [0.25433180641306513], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004342017709967513, "centered_acc": 0.0048463121556131155, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0048463121556131155], "ci": 0.0}, "piqa": {"acc_norm": 0.5184983677910773, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5184983677910773], "acc": 0.5277475516866159, "acc_ci": 0.0, "acc_runs": [0.5277475516866159], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01165783758381818, "centered_acc": 0.03699673558215455, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03699673558215455], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26851851851851855, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26851851851851855], "acc": 0.2537878787878788, "acc_ci": 0.0, "acc_runs": [0.2537878787878788], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009094042554994908, "centered_acc": 0.024691358024691395, "centered_acc_ci": 0.0, "centered_acc_runs": [0.024691358024691395], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.23464163822525597, "acc_norm_ci": 0.0, "acc_norm_runs": [0.23464163822525597], "acc": 0.19795221843003413, "acc_ci": 0.0, "acc_runs": [0.19795221843003413], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01238387356076873, "centered_acc": -0.020477815699658713, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.020477815699658713], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.176, "acc_ci": 0.0, "acc_runs": [0.176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.489344909234412, "acc_ci": 0.0, "acc_runs": [0.489344909234412], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014049294536290372, "centered_acc": -0.02131018153117603, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.02131018153117603], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.6, "acc_ci": 0.0, "acc_runs": [0.6], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0492365963917331, "centered_acc": 0.19999999999999996, "centered_acc_ci": 0.0, "centered_acc_runs": [0.19999999999999996], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 522256372.93410695, "perplexity_ci": 0.0, "perplexity_runs": [522256372.93410695], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.179, "acc_norm_ci": 0.0, "acc_norm_runs": [0.179], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012128730605719054, "centered_acc": -0.09466666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09466666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29004887644192695, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2920451826325706, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0072605927319602005, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29004887644192695], "mean_acc_runs": [0.2920451826325706], "mean_centered_acc_runs": [-0.0072605927319602005], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-4896", "checkpoint": "checkpoint-4896", "global_step": 4896, "tokens_seen": 5133828096, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2552280422226648, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2552280422226648], "acc": 0.2568213503286198, "acc_ci": 0.0, "acc_runs": [0.2568213503286198], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004350982826580739, "centered_acc": 0.006970722963553093, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006970722963553093], "ci": 0.0}, "piqa": {"acc_norm": 0.5223068552774756, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5223068552774756], "acc": 0.5266594124047879, "acc_ci": 0.0, "acc_runs": [0.5266594124047879], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011654208652596384, "centered_acc": 0.044613710554951114, "centered_acc_ci": 0.0, "centered_acc_runs": [0.044613710554951114], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26641414141414144, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26641414141414144], "acc": 0.26052188552188554, "acc_ci": 0.0, "acc_runs": [0.26052188552188554], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009071357971078843, "centered_acc": 0.021885521885521914, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021885521885521914], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.23464163822525597, "acc_norm_ci": 0.0, "acc_norm_runs": [0.23464163822525597], "acc": 0.19880546075085323, "acc_ci": 0.0, "acc_runs": [0.19880546075085323], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01238387356076873, "centered_acc": -0.020477815699658713, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.020477815699658713], "ci": 0.0}, "openbookqa": {"acc_norm": 0.29, "acc_norm_ci": 0.0, "acc_norm_runs": [0.29], "acc": 0.178, "acc_ci": 0.0, "acc_runs": [0.178], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020313179231745138, "centered_acc": 0.05333333333333331, "centered_acc_ci": 0.0, "centered_acc_runs": [0.05333333333333331], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.49171270718232046, "acc_ci": 0.0, "acc_runs": [0.49171270718232046], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405055532282417, "centered_acc": -0.016574585635359074, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.016574585635359074], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.63, "acc_ci": 0.0, "acc_runs": [0.63], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.048523658709390974, "centered_acc": 0.26, "centered_acc_ci": 0.0, "centered_acc_runs": [0.26], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 701259286.0416783, "perplexity_ci": 0.0, "perplexity_runs": [701259286.0416783], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.176, "acc_norm_ci": 0.0, "acc_norm_runs": [0.176], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012048616898597498, "centered_acc": -0.09866666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09866666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2907651128565896, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2964808277962167, "mean_acc_ci": 0.0, "mean_centered_acc": 0.0008510160314528441, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2907651128565896], "mean_acc_runs": [0.2964808277962167], "mean_centered_acc_runs": [0.0008510160314528441], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-5184", "checkpoint": "checkpoint-5184", "global_step": 5184, "tokens_seen": 5435817984, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2575184226249751, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2575184226249751], "acc": 0.2568213503286198, "acc_ci": 0.0, "acc_runs": [0.2568213503286198], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004363736410689568, "centered_acc": 0.010024563499966796, "centered_acc_ci": 0.0, "centered_acc_runs": [0.010024563499966796], "ci": 0.0}, "piqa": {"acc_norm": 0.5223068552774756, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5223068552774756], "acc": 0.5266594124047879, "acc_ci": 0.0, "acc_runs": [0.5266594124047879], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011654208652596384, "centered_acc": 0.044613710554951114, "centered_acc_ci": 0.0, "centered_acc_runs": [0.044613710554951114], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2748316498316498, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2748316498316498], "acc": 0.25925925925925924, "acc_ci": 0.0, "acc_runs": [0.25925925925925924], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009160538115254864, "centered_acc": 0.03310886644219976, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03310886644219976], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2380546075085324, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2380546075085324], "acc": 0.19539249146757678, "acc_ci": 0.0, "acc_runs": [0.19539249146757678], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012445770028026118, "centered_acc": -0.015927189988623452, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.015927189988623452], "ci": 0.0}, "openbookqa": {"acc_norm": 0.3, "acc_norm_ci": 0.0, "acc_norm_runs": [0.3], "acc": 0.18, "acc_ci": 0.0, "acc_runs": [0.18], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020514426225627987, "centered_acc": 0.06666666666666665, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06666666666666665], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.505130228887135, "acc_ci": 0.0, "acc_runs": [0.505130228887135], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051745961790525, "centered_acc": 0.010260457774269982, "centered_acc_ci": 0.0, "centered_acc_runs": [0.010260457774269982], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 578011929.8043449, "perplexity_ci": 0.0, "perplexity_runs": [578011929.8043449], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.04, "acc_ci": 0.0, "acc_runs": [0.04], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2956185892071055, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2921550204121079, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0029642223892409566, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2956185892071055], "mean_acc_runs": [0.2921550204121079], "mean_centered_acc_runs": [-0.0029642223892409566], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-5472", "checkpoint": "checkpoint-5472", "global_step": 5472, "tokens_seen": 5737807872, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2552280422226648, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2552280422226648], "acc": 0.2559251145190201, "acc_ci": 0.0, "acc_runs": [0.2559251145190201], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004350982826580739, "centered_acc": 0.006970722963553093, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006970722963553093], "ci": 0.0}, "piqa": {"acc_norm": 0.5190424374319913, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5190424374319913], "acc": 0.5255712731229597, "acc_ci": 0.0, "acc_runs": [0.5255712731229597], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011657360703051524, "centered_acc": 0.0380848748639826, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0380848748639826], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2718855218855219, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2718855218855219], "acc": 0.2609427609427609, "acc_ci": 0.0, "acc_runs": [0.2609427609427609], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009129795867310605, "centered_acc": 0.02918069584736253, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02918069584736253], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.23720136518771331, "acc_norm_ci": 0.0, "acc_norm_runs": [0.23720136518771331], "acc": 0.19965870307167236, "acc_ci": 0.0, "acc_runs": [0.19965870307167236], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01243039982926071, "centered_acc": -0.017064846416382246, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.017064846416382246], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.17, "acc_ci": 0.0, "acc_runs": [0.17], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5011838989739542, "acc_ci": 0.0, "acc_runs": [0.5011838989739542], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052446290528871, "centered_acc": 0.0023677979479084232, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0023677979479084232], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 728174583.5411192, "perplexity_ci": 0.0, "perplexity_runs": [728174583.5411192], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.174, "acc_norm_ci": 0.0, "acc_norm_runs": [0.174], "acc": 0.048, "acc_ci": 0.0, "acc_runs": [0.048], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011994493230973449, "centered_acc": -0.10133333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10133333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29055956112131526, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29295692124040673, "mean_acc_ci": 0.0, "mean_centered_acc": -0.006357684953278712, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29055956112131526], "mean_acc_runs": [0.29295692124040673], "mean_centered_acc_runs": [-0.006357684953278712], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-5760", "checkpoint": "checkpoint-5760", "global_step": 5760, "tokens_seen": 6039797760, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2560246962756423, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2560246962756423], "acc": 0.2552280422226648, "acc_ci": 0.0, "acc_runs": [0.2552280422226648], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004355436696716235, "centered_acc": 0.008032928367523043, "centered_acc_ci": 0.0, "centered_acc_runs": [0.008032928367523043], "ci": 0.0}, "piqa": {"acc_norm": 0.5119695321001088, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5119695321001088], "acc": 0.5272034820457019, "acc_ci": 0.0, "acc_runs": [0.5272034820457019], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011662480968070035, "centered_acc": 0.02393906420021752, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02393906420021752], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2769360269360269, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2769360269360269], "acc": 0.25673400673400676, "acc_ci": 0.0, "acc_runs": [0.25673400673400676], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009182190173795887, "centered_acc": 0.03591470258136923, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03591470258136923], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.23464163822525597, "acc_norm_ci": 0.0, "acc_norm_runs": [0.23464163822525597], "acc": 0.2022184300341297, "acc_ci": 0.0, "acc_runs": [0.2022184300341297], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01238387356076873, "centered_acc": -0.020477815699658713, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.020477815699658713], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.176, "acc_ci": 0.0, "acc_runs": [0.176], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5098658247829518, "acc_ci": 0.0, "acc_runs": [0.5098658247829518], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014049749833367674, "centered_acc": 0.019731649565903675, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019731649565903675], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.6, "acc_ci": 0.0, "acc_runs": [0.6], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0492365963917331, "centered_acc": 0.19999999999999996, "centered_acc_ci": 0.0, "centered_acc_runs": [0.19999999999999996], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 924589344.7186648, "perplexity_ci": 0.0, "perplexity_runs": [924589344.7186648], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.178, "acc_norm_ci": 0.0, "acc_norm_runs": [0.178], "acc": 0.049, "acc_ci": 0.0, "acc_runs": [0.049], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012102167676183571, "centered_acc": -0.09600000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09600000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.28959531558950563, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29545372475931553, "mean_acc_ci": 0.0, "mean_centered_acc": -0.0035871719374716265, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.28959531558950563], "mean_acc_runs": [0.29545372475931553], "mean_centered_acc_runs": [-0.0035871719374716265], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6048", "checkpoint": "checkpoint-6048", "global_step": 6048, "tokens_seen": 6341787648, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2574188408683529, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2574188408683529], "acc": 0.25423222465644296, "acc_ci": 0.0, "acc_runs": [0.25423222465644296], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043631851720466585, "centered_acc": 0.009891787824470569, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009891787824470569], "ci": 0.0}, "piqa": {"acc_norm": 0.5201305767138193, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5201305767138193], "acc": 0.5261153427638737, "acc_ci": 0.0, "acc_runs": [0.5261153427638737], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011656365410780507, "centered_acc": 0.040261153427638696, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040261153427638696], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26515151515151514, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26515151515151514], "acc": 0.2601010101010101, "acc_ci": 0.0, "acc_runs": [0.2601010101010101], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009057621139172679, "centered_acc": 0.020202020202020183, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020202020202020183], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24061433447098976, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24061433447098976], "acc": 0.20477815699658702, "acc_ci": 0.0, "acc_runs": [0.20477815699658702], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012491468532390682, "centered_acc": -0.012514220705346987, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.012514220705346987], "ci": 0.0}, "openbookqa": {"acc_norm": 0.298, "acc_norm_ci": 0.0, "acc_norm_runs": [0.298], "acc": 0.178, "acc_ci": 0.0, "acc_runs": [0.178], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020475118092988954, "centered_acc": 0.06399999999999999, "centered_acc_ci": 0.0, "centered_acc_runs": [0.06399999999999999], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.4996053670086819, "acc_ci": 0.0, "acc_runs": [0.4996053670086819], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405248130604961, "centered_acc": -0.0007892659826361781, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.0007892659826361781], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 1083041077.3872633, "perplexity_ci": 0.0, "perplexity_runs": [1083041077.3872633], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.181, "acc_norm_ci": 0.0, "acc_norm_runs": [0.181], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012181436179178025, "centered_acc": -0.09200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09200000000000001], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29371921120077954, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2938119563300296, "mean_acc_ci": 0.0, "mean_centered_acc": -0.003819289076272574, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29371921120077954], "mean_acc_runs": [0.2938119563300296], "mean_centered_acc_runs": [-0.003819289076272574], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6336", "checkpoint": "checkpoint-6336", "global_step": 6336, "tokens_seen": 6643777536, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2545309699263095, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2545309699263095], "acc": 0.25721967735510853, "acc_ci": 0.0, "acc_runs": [0.25721967735510853], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043470700195274055, "centered_acc": 0.006041293235079366, "centered_acc_ci": 0.0, "centered_acc_runs": [0.006041293235079366], "ci": 0.0}, "piqa": {"acc_norm": 0.5179542981501633, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5179542981501633], "acc": 0.5342763873775843, "acc_ci": 0.0, "acc_runs": [0.5342763873775843], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011658300623287121, "centered_acc": 0.0359085963003265, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0359085963003265], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2647306397306397, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2647306397306397], "acc": 0.26136363636363635, "acc_ci": 0.0, "acc_runs": [0.26136363636363635], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009053021086174166, "centered_acc": 0.019640852974186274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019640852974186274], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.24146757679180889, "acc_norm_ci": 0.0, "acc_norm_runs": [0.24146757679180889], "acc": 0.2090443686006826, "acc_ci": 0.0, "acc_runs": [0.2090443686006826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012506564839739446, "centered_acc": -0.011376564277588153, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.011376564277588153], "ci": 0.0}, "openbookqa": {"acc_norm": 0.286, "acc_norm_ci": 0.0, "acc_norm_runs": [0.286], "acc": 0.18, "acc_ci": 0.0, "acc_runs": [0.18], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020229346329177604, "centered_acc": 0.047999999999999966, "centered_acc_ci": 0.0, "centered_acc_runs": [0.047999999999999966], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.49013417521704816, "acc_ci": 0.0, "acc_runs": [0.49013417521704816], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014049749833367674, "centered_acc": -0.019731649565903675, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.019731649565903675], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.56, "acc_ci": 0.0, "acc_runs": [0.56], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.049888765156985884, "centered_acc": 0.1200000000000001, "centered_acc_ci": 0.0, "centered_acc_runs": [0.1200000000000001], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 808986448.4541316, "perplexity_ci": 0.0, "perplexity_runs": [808986448.4541316], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.178, "acc_norm_ci": 0.0, "acc_norm_runs": [0.178], "acc": 0.043, "acc_ci": 0.0, "acc_runs": [0.043], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012102167676183571, "centered_acc": -0.09600000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09600000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.29044724743315353, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29133257066877605, "mean_acc_ci": 0.0, "mean_centered_acc": -0.015660283087388777, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.29044724743315353], "mean_acc_runs": [0.29133257066877605], "mean_centered_acc_runs": [-0.015660283087388777], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6624", "checkpoint": "checkpoint-6624", "global_step": 6624, "tokens_seen": 6945767424, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2529376618203545, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2529376618203545], "acc": 0.25492929695279826, "acc_ci": 0.0, "acc_runs": [0.25492929695279826], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.0043380713189125845, "centered_acc": 0.003916882427139316, "centered_acc_ci": 0.0, "centered_acc_runs": [0.003916882427139316], "ci": 0.0}, "piqa": {"acc_norm": 0.5087051142546246, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5087051142546246], "acc": 0.5261153427638737, "acc_ci": 0.0, "acc_runs": [0.5261153427638737], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011664055982032772, "centered_acc": 0.017410228509249226, "centered_acc_ci": 0.0, "centered_acc_runs": [0.017410228509249226], "ci": 0.0}, "arc_easy": {"acc_norm": 0.26557239057239057, "acc_norm_ci": 0.0, "acc_norm_runs": [0.26557239057239057], "acc": 0.2563131313131313, "acc_ci": 0.0, "acc_runs": [0.2563131313131313], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.00906221062697191, "centered_acc": 0.020763187429854096, "centered_acc_ci": 0.0, "centered_acc_runs": [0.020763187429854096], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.2440273037542662, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2440273037542662], "acc": 0.19880546075085323, "acc_ci": 0.0, "acc_runs": [0.19880546075085323], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012551447627856151, "centered_acc": -0.007963594994311726, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.007963594994311726], "ci": 0.0}, "openbookqa": {"acc_norm": 0.274, "acc_norm_ci": 0.0, "acc_norm_runs": [0.274], "acc": 0.172, "acc_ci": 0.0, "acc_runs": [0.172], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.019966103540279518, "centered_acc": 0.03200000000000003, "centered_acc_ci": 0.0, "centered_acc_runs": [0.03200000000000003], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5067087608524072, "acc_ci": 0.0, "acc_runs": [0.5067087608524072], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014051220692330247, "centered_acc": 0.013417521704814472, "centered_acc_ci": 0.0, "centered_acc_runs": [0.013417521704814472], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.55, "acc_ci": 0.0, "acc_runs": [0.55], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.05, "centered_acc": 0.10000000000000009, "centered_acc_ci": 0.0, "centered_acc_runs": [0.10000000000000009], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 1203424385.6715648, "perplexity_ci": 0.0, "perplexity_runs": [1203424385.6715648], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.172, "acc_norm_ci": 0.0, "acc_norm_runs": [0.172], "acc": 0.042, "acc_ci": 0.0, "acc_runs": [0.042], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011939788882495308, "centered_acc": -0.10400000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.10400000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.28620707840027265, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2885159454406764, "mean_acc_ci": 0.0, "mean_centered_acc": -0.018653427930650432, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.28620707840027265], "mean_acc_runs": [0.2885159454406764], "mean_centered_acc_runs": [-0.018653427930650432], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-6912", "checkpoint": "checkpoint-6912", "global_step": 6912, "tokens_seen": 7247757312, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25731925911173076, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25731925911173076], "acc": 0.2546305516829317, "acc_ci": 0.0, "acc_runs": [0.2546305516829317], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004362633637374433, "centered_acc": 0.009759012148974344, "centered_acc_ci": 0.0, "centered_acc_runs": [0.009759012148974344], "ci": 0.0}, "piqa": {"acc_norm": 0.514689880304679, "acc_norm_ci": 0.0, "acc_norm_runs": [0.514689880304679], "acc": 0.528835690968444, "acc_ci": 0.0, "acc_runs": [0.528835690968444], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011660788281735552, "centered_acc": 0.029379760609357986, "centered_acc_ci": 0.0, "centered_acc_runs": [0.029379760609357986], "ci": 0.0}, "arc_easy": {"acc_norm": 0.265993265993266, "acc_norm_ci": 0.0, "acc_norm_runs": [0.265993265993266], "acc": 0.255050505050505, "acc_ci": 0.0, "acc_runs": [0.255050505050505], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009066789565615746, "centered_acc": 0.021324354657688005, "centered_acc_ci": 0.0, "centered_acc_runs": [0.021324354657688005], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.23208191126279865, "acc_norm_ci": 0.0, "acc_norm_runs": [0.23208191126279865], "acc": 0.2022184300341297, "acc_ci": 0.0, "acc_runs": [0.2022184300341297], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012336718284948816, "centered_acc": -0.023890784982935138, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.023890784982935138], "ci": 0.0}, "openbookqa": {"acc_norm": 0.284, "acc_norm_ci": 0.0, "acc_norm_runs": [0.284], "acc": 0.172, "acc_ci": 0.0, "acc_runs": [0.172], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020186703693570774, "centered_acc": 0.0453333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0453333333333333], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5122336227308603, "acc_ci": 0.0, "acc_runs": [0.5122336227308603], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014048278820405729, "centered_acc": 0.02446724546172052, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02446724546172052], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 1379801887.5433037, "perplexity_ci": 0.0, "perplexity_runs": [1379801887.5433037], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.176, "acc_norm_ci": 0.0, "acc_norm_runs": [0.176], "acc": 0.048, "acc_ci": 0.0, "acc_runs": [0.048], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012048616898597498, "centered_acc": -0.09866666666666668, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09866666666666668], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.28834738611207905, "mean_acc_norm_ci": 0.0, "mean_acc": 0.2931256262240571, "mean_acc_ci": 0.0, "mean_centered_acc": -0.008413202432347455, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.28834738611207905], "mean_acc_runs": [0.2931256262240571], "mean_centered_acc_runs": [-0.008413202432347455], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-7200", "checkpoint": "checkpoint-7200", "global_step": 7200, "tokens_seen": 7549747200, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.2583150766779526, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2583150766779526], "acc": 0.2574188408683529, "acc_ci": 0.0, "acc_runs": [0.2574188408683529], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004368135676213571, "centered_acc": 0.01108676890393682, "centered_acc_ci": 0.0, "centered_acc_runs": [0.01108676890393682], "ci": 0.0}, "piqa": {"acc_norm": 0.5125136017410229, "acc_norm_ci": 0.0, "acc_norm_runs": [0.5125136017410229], "acc": 0.5212187159956474, "acc_ci": 0.0, "acc_runs": [0.5212187159956474], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011662170084916882, "centered_acc": 0.02502720348204579, "centered_acc_ci": 0.0, "centered_acc_runs": [0.02502720348204579], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2609427609427609, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2609427609427609], "acc": 0.2529461279461279, "acc_ci": 0.0, "acc_runs": [0.2529461279461279], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009011142493235922, "centered_acc": 0.014590347923681227, "centered_acc_ci": 0.0, "centered_acc_runs": [0.014590347923681227], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.23464163822525597, "acc_norm_ci": 0.0, "acc_norm_runs": [0.23464163822525597], "acc": 0.20733788395904437, "acc_ci": 0.0, "acc_runs": [0.20733788395904437], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01238387356076873, "centered_acc": -0.020477815699658713, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.020477815699658713], "ci": 0.0}, "openbookqa": {"acc_norm": 0.28, "acc_norm_ci": 0.0, "acc_norm_runs": [0.28], "acc": 0.166, "acc_ci": 0.0, "acc_runs": [0.166], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020099950647503192, "centered_acc": 0.040000000000000036, "centered_acc_ci": 0.0, "centered_acc_runs": [0.040000000000000036], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.5035516969218626, "acc_ci": 0.0, "acc_runs": [0.5035516969218626], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.014052131146915852, "centered_acc": 0.00710339384372527, "centered_acc_ci": 0.0, "centered_acc_runs": [0.00710339384372527], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.58, "acc_ci": 0.0, "acc_runs": [0.58], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04960449637488582, "centered_acc": 0.15999999999999992, "centered_acc_ci": 0.0, "centered_acc_runs": [0.15999999999999992], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 1695076667.3291082, "perplexity_ci": 0.0, "perplexity_runs": [1695076667.3291082], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.178, "acc_norm_ci": 0.0, "acc_norm_runs": [0.178], "acc": 0.044, "acc_ci": 0.0, "acc_runs": [0.044], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012102167676183571, "centered_acc": -0.09600000000000002, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09600000000000002], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.28740217959783204, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29107607274647357, "mean_acc_ci": 0.0, "mean_centered_acc": -0.011343908666541003, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.28740217959783204], "mean_acc_runs": [0.29107607274647357], "mean_centered_acc_runs": [-0.011343908666541003], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true} +{"run_dir": "/workspace/concept_model/runs/psm25-spm25-fulle-lm", "run": "psm25-spm25-fulle-lm", "checkpoint_path": "/workspace/concept_model/runs/psm25-spm25-fulle-lm/checkpoint-7488", "checkpoint": "checkpoint-7488", "global_step": 7488, "tokens_seen": 7851737088, "num_fewshot": 0, "runs_per_task": 1, "base_seed": 42, "tasks_evaluated": ["hellaswag", "piqa", "arc_easy", "arc_challenge", "openbookqa", "winogrande", "boolq", "copa", "lambada_openai", "sciq"], "tasks": {"hellaswag": {"acc_norm": 0.25403306114319857, "acc_norm_ci": 0.0, "acc_norm_runs": [0.25403306114319857], "acc": 0.25363473411670984, "acc_ci": 0.0, "acc_runs": [0.25363473411670984], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.004344266179634878, "centered_acc": 0.005377414857598091, "centered_acc_ci": 0.0, "centered_acc_runs": [0.005377414857598091], "ci": 0.0}, "piqa": {"acc_norm": 0.514689880304679, "acc_norm_ci": 0.0, "acc_norm_runs": [0.514689880304679], "acc": 0.5201305767138193, "acc_ci": 0.0, "acc_runs": [0.5201305767138193], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.011660788281735552, "centered_acc": 0.029379760609357986, "centered_acc_ci": 0.0, "centered_acc_runs": [0.029379760609357986], "ci": 0.0}, "arc_easy": {"acc_norm": 0.2647306397306397, "acc_norm_ci": 0.0, "acc_norm_runs": [0.2647306397306397], "acc": 0.25252525252525254, "acc_ci": 0.0, "acc_runs": [0.25252525252525254], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.009053021086174166, "centered_acc": 0.019640852974186274, "centered_acc_ci": 0.0, "centered_acc_runs": [0.019640852974186274], "ci": 0.0}, "arc_challenge": {"acc_norm": 0.23037542662116042, "acc_norm_ci": 0.0, "acc_norm_runs": [0.23037542662116042], "acc": 0.21075085324232082, "acc_ci": 0.0, "acc_runs": [0.21075085324232082], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012304928418747618, "centered_acc": -0.02616609783845277, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.02616609783845277], "ci": 0.0}, "openbookqa": {"acc_norm": 0.284, "acc_norm_ci": 0.0, "acc_norm_runs": [0.284], "acc": 0.154, "acc_ci": 0.0, "acc_runs": [0.154], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.020186703693570774, "centered_acc": 0.0453333333333333, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0453333333333333], "ci": 0.0}, "winogrande": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.500394632991318, "acc_ci": 0.0, "acc_runs": [0.500394632991318], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.01405248130604961, "centered_acc": 0.0007892659826360671, "centered_acc_ci": 0.0, "centered_acc_runs": [0.0007892659826360671], "ci": 0.0}, "boolq": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.3782874617737003, "acc_ci": 0.0, "acc_runs": [0.3782874617737003], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.008482001133931097, "centered_acc": -0.24342507645259936, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.24342507645259936], "ci": 0.0}, "copa": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.59, "acc_ci": 0.0, "acc_runs": [0.59], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.04943110704237104, "centered_acc": 0.17999999999999994, "centered_acc_ci": 0.0, "centered_acc_runs": [0.17999999999999994], "ci": 0.0}, "lambada_openai": {"acc_norm": null, "acc_norm_ci": 0.0, "acc_norm_runs": [], "acc": 0.0, "acc_ci": 0.0, "acc_runs": [0.0], "perplexity": 2016319301.444169, "perplexity_ci": 0.0, "perplexity_runs": [2016319301.444169], "stderr": 0.0, "centered_acc": null, "centered_acc_ci": 0.0, "centered_acc_runs": [], "ci": 0.0}, "sciq": {"acc_norm": 0.177, "acc_norm_ci": 0.0, "acc_norm_runs": [0.177], "acc": 0.047, "acc_ci": 0.0, "acc_runs": [0.047], "perplexity": null, "perplexity_ci": 0.0, "perplexity_runs": [], "stderr": 0.012075463420375087, "centered_acc": -0.09733333333333334, "centered_acc_ci": 0.0, "centered_acc_runs": [-0.09733333333333334], "ci": 0.0}}, "summary": {"mean_acc_norm": 0.2874715012999463, "mean_acc_norm_ci": 0.0, "mean_acc": 0.29067235113631207, "mean_acc_ci": 0.0, "mean_centered_acc": -0.009600431096363757, "mean_centered_acc_ci": 0.0, "mean_acc_norm_runs": [0.2874715012999463], "mean_acc_runs": [0.29067235113631207], "mean_centered_acc_runs": [-0.009600431096363757], "ci": 0.0}, "batch_size_per_device": 8, "device": "cuda:0", "dtype": "float16", "next_i_enabled": false, "max_next_i_train": 1, "l2r_prefix_enabled": true}