Add files using upload-large-folder tool

results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-live8-scalars-seed11/command.json

@@ -0,0 +1,79 @@
+{
+ "gpu": "2",
+ "command": [
+  "/home/heck2/sbhansali8/SpecRouter/envs/hedgespec/bin/python",
+  "/home/heck2/sbhansali8/SpecRouter/implementations/specrouter/scripts/train_mlp_screen.py",
+  "--store",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/hvd-training-10k-campaign-v2/mlp-sweeps/screening/llama31-8b/base",
+  "--output-dir",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-live8-scalars-seed11",
+  "--train-request-count",
+  "10000",
+  "--train-request-selection-mode",
+  "source-balanced",
+  "--request-selection-seed",
+  "42",
+  "--validation-fold-count",
+  "2",
+  "--validation-fold-index",
+  "0",
+  "--validation-fold-seed",
+  "42",
+  "--batch-size",
+  "8192",
+  "--max-epochs",
+  "60",
+  "--patience",
+  "20",
+  "--causal-refit-rounds",
+  "0",
+  "--device",
+  "cuda",
+  "--preload-hidden-to-device",
+  "--max-hidden-bytes",
+  "128849018880",
+  "--hidden-layer-slots",
+  "1",
+  "--hidden-sample-dim",
+  "384",
+  "--hidden-fusion",
+  "flat",
+  "--projection-dim",
+  "32",
+  "--width",
+  "64",
+  "--depth",
+  "2",
+  "--activation",
+  "gelu",
+  "--normalization",
+  "layernorm",
+  "--output-link",
+  "softplus",
+  "--loss",
+  "mse",
+  "--objective",
+  "hybrid",
+  "--routing-temperature",
+  "0.2",
+  "--regression-weight",
+  "0.3",
+  "--optimizer",
+  "adamw",
+  "--learning-rate",
+  "0.001",
+  "--weight-decay",
+  "0.001",
+  "--scalar-features",
+  "ngram_hit_len",
+  "log_ngram_hit",
+  "is_repetition",
+  "seg_pos",
+  "log_seg_pos",
+  "context_len",
+  "log_context_len",
+  "prompt_len",
+  "--model-seed",
+  "11"
+ ]
+}

results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-live8-scalars-seed23/command.json

@@ -0,0 +1,79 @@
+{
+ "gpu": "3",
+ "command": [
+  "/home/heck2/sbhansali8/SpecRouter/envs/hedgespec/bin/python",
+  "/home/heck2/sbhansali8/SpecRouter/implementations/specrouter/scripts/train_mlp_screen.py",
+  "--store",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/hvd-training-10k-campaign-v2/mlp-sweeps/screening/llama31-8b/base",
+  "--output-dir",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-live8-scalars-seed23",
+  "--train-request-count",
+  "10000",
+  "--train-request-selection-mode",
+  "source-balanced",
+  "--request-selection-seed",
+  "42",
+  "--validation-fold-count",
+  "2",
+  "--validation-fold-index",
+  "0",
+  "--validation-fold-seed",
+  "42",
+  "--batch-size",
+  "8192",
+  "--max-epochs",
+  "60",
+  "--patience",
+  "20",
+  "--causal-refit-rounds",
+  "0",
+  "--device",
+  "cuda",
+  "--preload-hidden-to-device",
+  "--max-hidden-bytes",
+  "128849018880",
+  "--hidden-layer-slots",
+  "1",
+  "--hidden-sample-dim",
+  "384",
+  "--hidden-fusion",
+  "flat",
+  "--projection-dim",
+  "32",
+  "--width",
+  "64",
+  "--depth",
+  "2",
+  "--activation",
+  "gelu",
+  "--normalization",
+  "layernorm",
+  "--output-link",
+  "softplus",
+  "--loss",
+  "mse",
+  "--objective",
+  "hybrid",
+  "--routing-temperature",
+  "0.2",
+  "--regression-weight",
+  "0.3",
+  "--optimizer",
+  "adamw",
+  "--learning-rate",
+  "0.001",
+  "--weight-decay",
+  "0.001",
+  "--scalar-features",
+  "ngram_hit_len",
+  "log_ngram_hit",
+  "is_repetition",
+  "seg_pos",
+  "log_seg_pos",
+  "context_len",
+  "log_context_len",
+  "prompt_len",
+  "--model-seed",
+  "23"
+ ]
+}

results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-live8-scalars-seed7/command.json

@@ -0,0 +1,79 @@
+{
+ "gpu": "1",
+ "command": [
+  "/home/heck2/sbhansali8/SpecRouter/envs/hedgespec/bin/python",
+  "/home/heck2/sbhansali8/SpecRouter/implementations/specrouter/scripts/train_mlp_screen.py",
+  "--store",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/hvd-training-10k-campaign-v2/mlp-sweeps/screening/llama31-8b/base",
+  "--output-dir",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-live8-scalars-seed7",
+  "--train-request-count",
+  "10000",
+  "--train-request-selection-mode",
+  "source-balanced",
+  "--request-selection-seed",
+  "42",
+  "--validation-fold-count",
+  "2",
+  "--validation-fold-index",
+  "0",
+  "--validation-fold-seed",
+  "42",
+  "--batch-size",
+  "8192",
+  "--max-epochs",
+  "60",
+  "--patience",
+  "20",
+  "--causal-refit-rounds",
+  "0",
+  "--device",
+  "cuda",
+  "--preload-hidden-to-device",
+  "--max-hidden-bytes",
+  "128849018880",
+  "--hidden-layer-slots",
+  "1",
+  "--hidden-sample-dim",
+  "384",
+  "--hidden-fusion",
+  "flat",
+  "--projection-dim",
+  "32",
+  "--width",
+  "64",
+  "--depth",
+  "2",
+  "--activation",
+  "gelu",
+  "--normalization",
+  "layernorm",
+  "--output-link",
+  "softplus",
+  "--loss",
+  "mse",
+  "--objective",
+  "hybrid",
+  "--routing-temperature",
+  "0.2",
+  "--regression-weight",
+  "0.3",
+  "--optimizer",
+  "adamw",
+  "--learning-rate",
+  "0.001",
+  "--weight-decay",
+  "0.001",
+  "--scalar-features",
+  "ngram_hit_len",
+  "log_ngram_hit",
+  "is_repetition",
+  "seg_pos",
+  "log_seg_pos",
+  "context_len",
+  "log_context_len",
+  "prompt_len",
+  "--model-seed",
+  "7"
+ ]
+}

results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-live8-scalars-seed7/epoch-events.jsonl

@@ -0,0 +1,44 @@
+{"checkpoint_improved": true, "checkpoint_metric": 11.743166550157545, "early_stop_improved": true, "early_stop_metric": 11.743166550157545, "epoch": 0, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 9.610650291847115, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 49.91178083866432, "drafter_accepted_token_mae": 2.0268206523143673, "drafter_accepted_token_normalized_mae_percent": 50.08821916133568, "drafter_accepted_token_rounded_exact_fraction": 0.3737931500237685, "drafter_accepted_token_within_one_fraction": 0.5245848702984437, "hard_routing_tpot": 11.762863744460253, "macro_source_drafter_accepted_token_accuracy_percent": 49.75218269254729, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 12.144817752260668, "objective_total": 15.195632374070952, "prediction_oracle_tpot_regret_percent": 5.0352204641193055, "regression_loss": 10.169382072700946, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 40.6281112030195}, "validation": {"drafter_accepted_token_accuracy_percent": 49.175710367516, "drafter_accepted_token_mae": 1.9689905756203805, "drafter_accepted_token_normalized_mae_percent": 50.824289632484, "drafter_accepted_token_rounded_exact_fraction": 0.38504346190323163, "drafter_accepted_token_within_one_fraction": 0.5324403198963289, "hard_routing_tpot": 11.743166550157545, "macro_source_drafter_accepted_token_accuracy_percent": 49.05556167192323, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 12.120482532175044, "objective_total": 15.003677619729178, "prediction_oracle_tpot_regret_percent": 6.256102920679041, "regression_loss": 9.610650291847115, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 39.25010643859163}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.617977580036133, "early_stop_improved": true, "early_stop_metric": 11.617977580036133, "epoch": 1, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 6.731484834131435, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 56.71930425104873, "drafter_accepted_token_mae": 1.751354100012084, "drafter_accepted_token_normalized_mae_percent": 43.28069574895127, "drafter_accepted_token_rounded_exact_fraction": 0.38612304253680474, "drafter_accepted_token_within_one_fraction": 0.5444685128648099, "hard_routing_tpot": 11.664511272770893, "macro_source_drafter_accepted_token_accuracy_percent": 53.39198837507534, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 12.046752362755091, "objective_total": 14.071433607516147, "prediction_oracle_tpot_regret_percent": 4.42208596326954, "regression_loss": 6.748937482536851, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 42.18613608494213}, "validation": {"drafter_accepted_token_accuracy_percent": 55.36369379201052, "drafter_accepted_token_mae": 1.7292610853898378, "drafter_accepted_token_normalized_mae_percent": 44.63630620798948, "drafter_accepted_token_rounded_exact_fraction": 0.3954889496772036, "drafter_accepted_token_within_one_fraction": 0.5490412777041873, "hard_routing_tpot": 11.617977580036133, "macro_source_drafter_accepted_token_accuracy_percent": 52.80408022142103, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 12.005816040865191, "objective_total": 14.025261491104622, "prediction_oracle_tpot_regret_percent": 5.5852716516002765, "regression_loss": 6.731484834131435, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 41.468276636470684}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.58371829514904, "early_stop_improved": true, "early_stop_metric": 11.58371829514904, "epoch": 2, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.965770255481891, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 60.060983619540636, "drafter_accepted_token_mae": 1.6161329867268175, "drafter_accepted_token_normalized_mae_percent": 39.939016380459364, "drafter_accepted_token_rounded_exact_fraction": 0.39907247277036684, "drafter_accepted_token_within_one_fraction": 0.5570181956853565, "hard_routing_tpot": 11.625023910820993, "macro_source_drafter_accepted_token_accuracy_percent": 55.578296677289515, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 12.010859183410188, "objective_total": 13.746442933650608, "prediction_oracle_tpot_regret_percent": 4.556065817953225, "regression_loss": 5.785279167468065, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.652844884480494}, "validation": {"drafter_accepted_token_accuracy_percent": 58.34902945554328, "drafter_accepted_token_mae": 1.6136057987333083, "drafter_accepted_token_normalized_mae_percent": 41.65097054445672, "drafter_accepted_token_rounded_exact_fraction": 0.4078625940237597, "drafter_accepted_token_within_one_fraction": 0.5598081249069233, "hard_routing_tpot": 11.58371829514904, "macro_source_drafter_accepted_token_accuracy_percent": 54.89139422491577, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.970080979690785, "objective_total": 13.759812056335353, "prediction_oracle_tpot_regret_percent": 5.668879132558313, "regression_loss": 5.965770255481891, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.055051324415835}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.571375717180189, "early_stop_improved": true, "early_stop_metric": 11.571375717180189, "epoch": 3, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.755888405989706, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 61.0091280427135, "drafter_accepted_token_mae": 1.5777663062889777, "drafter_accepted_token_normalized_mae_percent": 38.9908719572865, "drafter_accepted_token_rounded_exact_fraction": 0.40358737412859813, "drafter_accepted_token_within_one_fraction": 0.5637037032543726, "hard_routing_tpot": 11.603422110729275, "macro_source_drafter_accepted_token_accuracy_percent": 55.762594030596766, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.968634014601934, "objective_total": 13.623443971397464, "prediction_oracle_tpot_regret_percent": 4.387149238896626, "regression_loss": 5.516033189318435, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 45.24279622309468}, "validation": {"drafter_accepted_token_accuracy_percent": 59.161165509406786, "drafter_accepted_token_mae": 1.5821427276753173, "drafter_accepted_token_normalized_mae_percent": 40.838834490593214, "drafter_accepted_token_rounded_exact_fraction": 0.41143151255783816, "drafter_accepted_token_within_one_fraction": 0.5661464824205942, "hard_routing_tpot": 11.571375717180189, "macro_source_drafter_accepted_token_accuracy_percent": 54.9171666327758, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.93366626662271, "objective_total": 13.660432788419623, "prediction_oracle_tpot_regret_percent": 5.543120083590214, "regression_loss": 5.755888405989706, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 44.71165545397458}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.560451578580219, "early_stop_improved": true, "early_stop_metric": 11.560451578580219, "epoch": 4, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.572318837491732, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 61.74200110119923, "drafter_accepted_token_mae": 1.5481105853363293, "drafter_accepted_token_normalized_mae_percent": 38.25799889880077, "drafter_accepted_token_rounded_exact_fraction": 0.40448978824305853, "drafter_accepted_token_within_one_fraction": 0.5647413884964544, "hard_routing_tpot": 11.5798472461312, "macro_source_drafter_accepted_token_accuracy_percent": 56.364782963134886, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.949556199564366, "objective_total": 13.536316917088595, "prediction_oracle_tpot_regret_percent": 4.245555572156734, "regression_loss": 5.289202391747429, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 45.342354839564024}, "validation": {"drafter_accepted_token_accuracy_percent": 59.78340568420462, "drafter_accepted_token_mae": 1.5580364381667229, "drafter_accepted_token_normalized_mae_percent": 40.21659431579538, "drafter_accepted_token_rounded_exact_fraction": 0.4123354992436296, "drafter_accepted_token_within_one_fraction": 0.5663345952569439, "hard_routing_tpot": 11.560451578580219, "macro_source_drafter_accepted_token_accuracy_percent": 55.49970118608223, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.924466496048074, "objective_total": 13.596162147295594, "prediction_oracle_tpot_regret_percent": 5.422517958142592, "regression_loss": 5.572318837491732, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 44.78609596057874}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.551399689319894, "early_stop_improved": true, "early_stop_metric": 11.551399689319894, "epoch": 5, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.570109783348732, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 61.95756956333124, "drafter_accepted_token_mae": 1.5393876037979062, "drafter_accepted_token_normalized_mae_percent": 38.04243043666876, "drafter_accepted_token_rounded_exact_fraction": 0.40860554878525923, "drafter_accepted_token_within_one_fraction": 0.568571846256781, "hard_routing_tpot": 11.567792192602049, "macro_source_drafter_accepted_token_accuracy_percent": 56.85106251643289, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.94308630541795, "objective_total": 13.518603931876811, "prediction_oracle_tpot_regret_percent": 4.339250022500085, "regression_loss": 5.251725421529538, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 45.875969258925714}, "validation": {"drafter_accepted_token_accuracy_percent": 59.964645744271195, "drafter_accepted_token_mae": 1.5510149928543249, "drafter_accepted_token_normalized_mae_percent": 40.035354255728805, "drafter_accepted_token_rounded_exact_fraction": 0.4159331572388171, "drafter_accepted_token_within_one_fraction": 0.570305866246548, "hard_routing_tpot": 11.551399689319894, "macro_source_drafter_accepted_token_accuracy_percent": 55.96782337953371, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.922219253711903, "objective_total": 13.593252188716523, "prediction_oracle_tpot_regret_percent": 5.516359659219372, "regression_loss": 5.570109783348732, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.22543930416214}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.543359580136068, "early_stop_improved": true, "early_stop_metric": 11.543359580136068, "epoch": 6, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.484017219495094, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 62.33593333980133, "drafter_accepted_token_mae": 1.5240771070568042, "drafter_accepted_token_normalized_mae_percent": 37.66406666019867, "drafter_accepted_token_rounded_exact_fraction": 0.4068405537583435, "drafter_accepted_token_within_one_fraction": 0.5685944925442153, "hard_routing_tpot": 11.557384488247866, "macro_source_drafter_accepted_token_accuracy_percent": 56.78002109150523, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.927720365518097, "objective_total": 13.462228227498693, "prediction_oracle_tpot_regret_percent": 4.165930278239438, "regression_loss": 5.115026206601991, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 45.46984411932738}, "validation": {"drafter_accepted_token_accuracy_percent": 60.26477564833698, "drafter_accepted_token_mae": 1.5393876202566825, "drafter_accepted_token_normalized_mae_percent": 39.73522435166302, "drafter_accepted_token_rounded_exact_fraction": 0.4147626773681969, "drafter_accepted_token_within_one_fraction": 0.5704574015869408, "hard_routing_tpot": 11.543359580136068, "macro_source_drafter_accepted_token_accuracy_percent": 55.78133201024416, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.910234277606413, "objective_total": 13.555439443454942, "prediction_oracle_tpot_regret_percent": 5.386125408385878, "regression_loss": 5.484017219495094, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 44.84694146951002}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.531625154066372, "early_stop_improved": true, "early_stop_metric": 11.531625154066372, "epoch": 7, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.454649036492356, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 62.43044410259668, "drafter_accepted_token_mae": 1.5202527274101134, "drafter_accepted_token_normalized_mae_percent": 37.56955589740332, "drafter_accepted_token_rounded_exact_fraction": 0.4086649952897744, "drafter_accepted_token_within_one_fraction": 0.5690916998727966, "hard_routing_tpot": 11.548006254801594, "macro_source_drafter_accepted_token_accuracy_percent": 56.90540491499855, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.909989036547781, "objective_total": 13.432755425258481, "prediction_oracle_tpot_regret_percent": 4.158334486350568, "regression_loss": 5.075887962368998, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 45.962718703083276}, "validation": {"drafter_accepted_token_accuracy_percent": 60.28639695295361, "drafter_accepted_token_mae": 1.538549986414073, "drafter_accepted_token_normalized_mae_percent": 39.71360304704639, "drafter_accepted_token_rounded_exact_fraction": 0.41607424186607933, "drafter_accepted_token_within_one_fraction": 0.5704260494475492, "hard_routing_tpot": 11.531625154066372, "macro_source_drafter_accepted_token_accuracy_percent": 55.98720358483938, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.891572840789363, "objective_total": 13.52796755173707, "prediction_oracle_tpot_regret_percent": 5.3318199366205965, "regression_loss": 5.454649036492356, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.283896189914806}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.528083862563978, "early_stop_improved": true, "early_stop_metric": 11.528083862563978, "epoch": 8, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.423365249180659, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 62.73494142021785, "drafter_accepted_token_mae": 1.5079312382004249, "drafter_accepted_token_normalized_mae_percent": 37.26505857978215, "drafter_accepted_token_rounded_exact_fraction": 0.41078869133373086, "drafter_accepted_token_within_one_fraction": 0.5722975649377238, "hard_routing_tpot": 11.534466750637382, "macro_source_drafter_accepted_token_accuracy_percent": 57.50567642114022, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.906289018665996, "objective_total": 13.416195301833557, "prediction_oracle_tpot_regret_percent": 4.075136801455015, "regression_loss": 5.0330209438918745, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 47.33727611298991}, "validation": {"drafter_accepted_token_accuracy_percent": 60.65007662484244, "drafter_accepted_token_mae": 1.524460623794902, "drafter_accepted_token_normalized_mae_percent": 39.34992337515756, "drafter_accepted_token_rounded_exact_fraction": 0.41886980762849807, "drafter_accepted_token_within_one_fraction": 0.5746063346997641, "hard_routing_tpot": 11.528083862563978, "macro_source_drafter_accepted_token_accuracy_percent": 56.59836203534646, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.888880539505534, "objective_total": 13.515890114259731, "prediction_oracle_tpot_regret_percent": 5.373761992371806, "regression_loss": 5.423365249180659, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.627495819385395}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.52433146461519, "early_stop_improved": true, "early_stop_metric": 11.52433146461519, "epoch": 9, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.401051702899149, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 62.61747400639321, "drafter_accepted_token_mae": 1.512684559126986, "drafter_accepted_token_normalized_mae_percent": 37.38252599360679, "drafter_accepted_token_rounded_exact_fraction": 0.40825109394711245, "drafter_accepted_token_within_one_fraction": 0.5691418452235442, "hard_routing_tpot": 11.531751626087477, "macro_source_drafter_accepted_token_accuracy_percent": 57.315328561758136, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.900353179716916, "objective_total": 13.397334747738979, "prediction_oracle_tpot_regret_percent": 3.9650917383069917, "regression_loss": 4.989938560073543, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 45.89990726493582}, "validation": {"drafter_accepted_token_accuracy_percent": 60.49694983242998, "drafter_accepted_token_mae": 1.530392929259776, "drafter_accepted_token_normalized_mae_percent": 39.50305016757002, "drafter_accepted_token_rounded_exact_fraction": 0.4163381223726254, "drafter_accepted_token_within_one_fraction": 0.5714867968302987, "hard_routing_tpot": 11.52433146461519, "macro_source_drafter_accepted_token_accuracy_percent": 56.31566433415439, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.884929770205225, "objective_total": 13.50524528107497, "prediction_oracle_tpot_regret_percent": 5.2117078459831045, "regression_loss": 5.401051702899149, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.151767120067234}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.519614927145467, "early_stop_improved": true, "early_stop_metric": 11.519614927145467, "epoch": 10, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.446871225323897, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 62.91948425922532, "drafter_accepted_token_mae": 1.5004637090368846, "drafter_accepted_token_normalized_mae_percent": 37.08051574077468, "drafter_accepted_token_rounded_exact_fraction": 0.4125635941113991, "drafter_accepted_token_within_one_fraction": 0.5727899194904262, "hard_routing_tpot": 11.527678950517837, "macro_source_drafter_accepted_token_accuracy_percent": 57.580614578793345, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.8951456397935, "objective_total": 13.384156849156872, "prediction_oracle_tpot_regret_percent": 3.9969433572810686, "regression_loss": 4.963370697877906, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.650531635780084}, "validation": {"drafter_accepted_token_accuracy_percent": 60.63135732135475, "drafter_accepted_token_mae": 1.5251858308251622, "drafter_accepted_token_normalized_mae_percent": 39.36864267864525, "drafter_accepted_token_rounded_exact_fraction": 0.42022056230062, "drafter_accepted_token_within_one_fraction": 0.5744652500725018, "hard_routing_tpot": 11.519614927145467, "macro_source_drafter_accepted_token_accuracy_percent": 56.62279540422282, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.879828187913159, "objective_total": 13.513889555510328, "prediction_oracle_tpot_regret_percent": 5.23795763080519, "regression_loss": 5.446871225323897, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.95382450061117}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.511194765123488, "early_stop_improved": true, "early_stop_metric": 11.511194765123488, "epoch": 11, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.377294595912254, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.26631608037041, "drafter_accepted_token_mae": 1.4864291534118907, "drafter_accepted_token_normalized_mae_percent": 36.73368391962959, "drafter_accepted_token_rounded_exact_fraction": 0.4146151051141585, "drafter_accepted_token_within_one_fraction": 0.5762645070196414, "hard_routing_tpot": 11.522369572975816, "macro_source_drafter_accepted_token_accuracy_percent": 57.79500132407983, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.891482337447012, "objective_total": 13.37076365307535, "prediction_oracle_tpot_regret_percent": 4.1351553939271035, "regression_loss": 4.93093771876113, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.939675188693066}, "validation": {"drafter_accepted_token_accuracy_percent": 61.084068312010686, "drafter_accepted_token_mae": 1.5076472940246088, "drafter_accepted_token_normalized_mae_percent": 38.915931687989314, "drafter_accepted_token_rounded_exact_fraction": 0.42231854296157534, "drafter_accepted_token_within_one_fraction": 0.5776971331081204, "hard_routing_tpot": 11.511194765123488, "macro_source_drafter_accepted_token_accuracy_percent": 56.87488893051137, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.876164206957078, "objective_total": 13.489352585730755, "prediction_oracle_tpot_regret_percent": 5.366852889529983, "regression_loss": 5.377294595912254, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.245771199004956}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.513019986254047, "early_stop_improved": false, "early_stop_metric": 11.513019986254047, "epoch": 12, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.404746479164473, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 62.99978760247063, "drafter_accepted_token_mae": 1.4972142328673443, "drafter_accepted_token_normalized_mae_percent": 37.00021239752937, "drafter_accepted_token_rounded_exact_fraction": 0.41019968566144244, "drafter_accepted_token_within_one_fraction": 0.5705976779871818, "hard_routing_tpot": 11.517977280033147, "macro_source_drafter_accepted_token_accuracy_percent": 57.460703400044004, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.88436794344409, "objective_total": 13.355497839479238, "prediction_oracle_tpot_regret_percent": 4.06944256407864, "regression_loss": 4.903766320117162, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 45.76407321039433}, "validation": {"drafter_accepted_token_accuracy_percent": 60.72195015171333, "drafter_accepted_token_mae": 1.5216761619151846, "drafter_accepted_token_normalized_mae_percent": 39.27804984828667, "drafter_accepted_token_rounded_exact_fraction": 0.4182192507361221, "drafter_accepted_token_within_one_fraction": 0.5730335023736182, "hard_routing_tpot": 11.513019986254047, "macro_source_drafter_accepted_token_accuracy_percent": 56.41665437983533, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.872934628744645, "objective_total": 13.494358572493987, "prediction_oracle_tpot_regret_percent": 5.322682357988828, "regression_loss": 5.404746479164473, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.01899056207836}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.507126243308278, "early_stop_improved": true, "early_stop_metric": 11.507126243308278, "epoch": 13, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.377221107271743, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.262944418290445, "drafter_accepted_token_mae": 1.4865655877761128, "drafter_accepted_token_normalized_mae_percent": 36.737055581709555, "drafter_accepted_token_rounded_exact_fraction": 0.41201887001900467, "drafter_accepted_token_within_one_fraction": 0.5742592995867255, "hard_routing_tpot": 11.508097306762464, "macro_source_drafter_accepted_token_accuracy_percent": 57.50881682613648, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.87938631031789, "objective_total": 13.340108324773695, "prediction_oracle_tpot_regret_percent": 4.010259004951133, "regression_loss": 4.86907338151935, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.850803369360825}, "validation": {"drafter_accepted_token_accuracy_percent": 60.94287987173703, "drafter_accepted_token_mae": 1.5131170941988092, "drafter_accepted_token_normalized_mae_percent": 39.05712012826297, "drafter_accepted_token_rounded_exact_fraction": 0.4194158573895686, "drafter_accepted_token_within_one_fraction": 0.5754267156805113, "hard_routing_tpot": 11.507126243308278, "macro_source_drafter_accepted_token_accuracy_percent": 56.497241259627174, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.86884328272379, "objective_total": 13.482009614905312, "prediction_oracle_tpot_regret_percent": 5.245885750716872, "regression_loss": 5.377221107271743, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.13039714992838}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.505521375055846, "early_stop_improved": true, "early_stop_metric": 11.505521375055846, "epoch": 14, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.370889350872232, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.11822499247282, "drafter_accepted_token_mae": 1.4924216618380284, "drafter_accepted_token_normalized_mae_percent": 36.88177500752718, "drafter_accepted_token_rounded_exact_fraction": 0.41128569646331714, "drafter_accepted_token_within_one_fraction": 0.5717453594825161, "hard_routing_tpot": 11.508570153851194, "macro_source_drafter_accepted_token_accuracy_percent": 57.49075415736277, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.870758229203078, "objective_total": 13.329828392096722, "prediction_oracle_tpot_regret_percent": 3.9592190524641024, "regression_loss": 4.863567209645482, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.90253205352585}, "validation": {"drafter_accepted_token_accuracy_percent": 60.81842232677013, "drafter_accepted_token_mae": 1.517938720528989, "drafter_accepted_token_normalized_mae_percent": 39.18157767322987, "drafter_accepted_token_rounded_exact_fraction": 0.41906575849969563, "drafter_accepted_token_within_one_fraction": 0.5731798123574458, "hard_routing_tpot": 11.505521375055846, "macro_source_drafter_accepted_token_accuracy_percent": 56.48096389515098, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.861748011185487, "objective_total": 13.473014816447156, "prediction_oracle_tpot_regret_percent": 5.222174113735578, "regression_loss": 5.370889350872232, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.23311163268345}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.503199418466801, "early_stop_improved": true, "early_stop_metric": 11.503199418466801, "epoch": 15, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.366492183721825, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.221434351819454, "drafter_accepted_token_mae": 1.4882452933318513, "drafter_accepted_token_normalized_mae_percent": 36.778565648180546, "drafter_accepted_token_rounded_exact_fraction": 0.4109631890663724, "drafter_accepted_token_within_one_fraction": 0.5726532329698402, "hard_routing_tpot": 11.505517748345355, "macro_source_drafter_accepted_token_accuracy_percent": 57.46191617346177, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.8705129017346, "objective_total": 13.322694714526763, "prediction_oracle_tpot_regret_percent": 3.987030820491344, "regression_loss": 4.840606042640547, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.7417754882146}, "validation": {"drafter_accepted_token_accuracy_percent": 60.86199884000039, "drafter_accepted_token_mae": 1.5162505170245382, "drafter_accepted_token_normalized_mae_percent": 39.13800115999961, "drafter_accepted_token_rounded_exact_fraction": 0.41842042696388493, "drafter_accepted_token_within_one_fraction": 0.5743607429411964, "hard_routing_tpot": 11.503199418466801, "macro_source_drafter_accepted_token_accuracy_percent": 56.43819325444411, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.859213175282928, "objective_total": 13.469160830399476, "prediction_oracle_tpot_regret_percent": 5.256578009100532, "regression_loss": 5.366492183721825, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.066832200656464}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.506323916342694, "early_stop_improved": false, "early_stop_metric": 11.506323916342694, "epoch": 16, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.3471802731766305, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.298387033072736, "drafter_accepted_token_mae": 1.4851314017576123, "drafter_accepted_token_normalized_mae_percent": 36.701612966927264, "drafter_accepted_token_rounded_exact_fraction": 0.41176430148436305, "drafter_accepted_token_within_one_fraction": 0.5731166730662649, "hard_routing_tpot": 11.510415404107707, "macro_source_drafter_accepted_token_accuracy_percent": 57.62929000096276, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.86913237797888, "objective_total": 13.317692205660359, "prediction_oracle_tpot_regret_percent": 3.882715795507319, "regression_loss": 4.8285327589382625, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 45.71046504350471}, "validation": {"drafter_accepted_token_accuracy_percent": 61.00068938859789, "drafter_accepted_token_mae": 1.5108774880045408, "drafter_accepted_token_normalized_mae_percent": 38.99931061140211, "drafter_accepted_token_rounded_exact_fraction": 0.42046876673747025, "drafter_accepted_token_within_one_fraction": 0.5751210323214431, "hard_routing_tpot": 11.506323916342694, "macro_source_drafter_accepted_token_accuracy_percent": 56.564550499026424, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.858189039598935, "objective_total": 13.462343121551925, "prediction_oracle_tpot_regret_percent": 5.161427678321284, "regression_loss": 5.3471802731766305, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 44.99716563240125}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.503182578938809, "early_stop_improved": true, "early_stop_metric": 11.503182578938809, "epoch": 17, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.396185277200379, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.43382307808476, "drafter_accepted_token_mae": 1.4796509798601207, "drafter_accepted_token_normalized_mae_percent": 36.56617692191524, "drafter_accepted_token_rounded_exact_fraction": 0.41095590990255426, "drafter_accepted_token_within_one_fraction": 0.5731953284753003, "hard_routing_tpot": 11.507697703890138, "macro_source_drafter_accepted_token_accuracy_percent": 57.59506026724431, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.873263869340288, "objective_total": 13.326571487434533, "prediction_oracle_tpot_regret_percent": 3.7221141327706375, "regression_loss": 4.844358726980815, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.04277555112665}, "validation": {"drafter_accepted_token_accuracy_percent": 61.02053145250783, "drafter_accepted_token_mae": 1.5101087839631848, "drafter_accepted_token_normalized_mae_percent": 38.97946854749217, "drafter_accepted_token_rounded_exact_fraction": 0.4184935819557987, "drafter_accepted_token_within_one_fraction": 0.5736945099791247, "hard_routing_tpot": 11.503182578938809, "macro_source_drafter_accepted_token_accuracy_percent": 56.48778181931732, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.862605817040475, "objective_total": 13.481461400200589, "prediction_oracle_tpot_regret_percent": 4.972056411988945, "regression_loss": 5.396185277200379, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.310838075122405}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.505028252441548, "early_stop_improved": false, "early_stop_metric": 11.505028252441548, "epoch": 18, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.391518810212985, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.48847145053158, "drafter_accepted_token_mae": 1.4774396325264532, "drafter_accepted_token_normalized_mae_percent": 36.51152854946842, "drafter_accepted_token_rounded_exact_fraction": 0.4121345278441159, "drafter_accepted_token_within_one_fraction": 0.5738973633857656, "hard_routing_tpot": 11.499433612575832, "macro_source_drafter_accepted_token_accuracy_percent": 57.71355306927475, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.870453746095542, "objective_total": 13.310073012345205, "prediction_oracle_tpot_regret_percent": 3.822702040750592, "regression_loss": 4.798730887498874, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 45.72829817259812}, "validation": {"drafter_accepted_token_accuracy_percent": 61.04996532034889, "drafter_accepted_token_mae": 1.508968482567884, "drafter_accepted_token_normalized_mae_percent": 38.95003467965111, "drafter_accepted_token_rounded_exact_fraction": 0.41976073092287636, "drafter_accepted_token_within_one_fraction": 0.5753378846189017, "hard_routing_tpot": 11.505028252441548, "macro_source_drafter_accepted_token_accuracy_percent": 56.50722815296291, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.863464976560893, "objective_total": 13.480920619624788, "prediction_oracle_tpot_regret_percent": 5.092293676497639, "regression_loss": 5.391518810212985, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 44.95027326570773}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.493755229438875, "early_stop_improved": true, "early_stop_metric": 11.493755229438875, "epoch": 19, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.414985378674529, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.17671917956025, "drafter_accepted_token_mae": 1.4900546935485934, "drafter_accepted_token_normalized_mae_percent": 36.82328082043975, "drafter_accepted_token_rounded_exact_fraction": 0.40998555692579075, "drafter_accepted_token_within_one_fraction": 0.5702808321620957, "hard_routing_tpot": 11.491906150366445, "macro_source_drafter_accepted_token_accuracy_percent": 57.633808806487565, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.863626011353043, "objective_total": 13.307453343733293, "prediction_oracle_tpot_regret_percent": 3.9006668652865173, "regression_loss": 4.8127577746008345, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 45.874635510248126}, "validation": {"drafter_accepted_token_accuracy_percent": 60.738782732971046, "drafter_accepted_token_mae": 1.521024048642191, "drafter_accepted_token_normalized_mae_percent": 39.261217267028954, "drafter_accepted_token_rounded_exact_fraction": 0.417827348993727, "drafter_accepted_token_within_one_fraction": 0.572163480505501, "hard_routing_tpot": 11.493755229438875, "macro_source_drafter_accepted_token_accuracy_percent": 56.46609139364677, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.860088978159212, "objective_total": 13.48458459176157, "prediction_oracle_tpot_regret_percent": 5.187751741820579, "regression_loss": 5.414985378674529, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.182098833419076}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.494902950432268, "early_stop_improved": false, "early_stop_metric": 11.494902950432268, "epoch": 20, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.370018195658583, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.43671171618316, "drafter_accepted_token_mae": 1.4795340910696437, "drafter_accepted_token_normalized_mae_percent": 36.56328828381684, "drafter_accepted_token_rounded_exact_fraction": 0.41054726573376094, "drafter_accepted_token_within_one_fraction": 0.5712499719448895, "hard_routing_tpot": 11.494100934364083, "macro_source_drafter_accepted_token_accuracy_percent": 57.766599417518485, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.86599312346506, "objective_total": 13.29282048015517, "prediction_oracle_tpot_regret_percent": 3.847606250169533, "regression_loss": 4.756091188967032, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 47.39629784570755}, "validation": {"drafter_accepted_token_accuracy_percent": 60.959219039040995, "drafter_accepted_token_mae": 1.512484096341532, "drafter_accepted_token_normalized_mae_percent": 39.040780960959005, "drafter_accepted_token_rounded_exact_fraction": 0.41849096927751606, "drafter_accepted_token_within_one_fraction": 0.5728244881110075, "hard_routing_tpot": 11.494902950432268, "macro_source_drafter_accepted_token_accuracy_percent": 56.648149131555186, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.859987088228523, "objective_total": 13.470992546926098, "prediction_oracle_tpot_regret_percent": 5.067044834807843, "regression_loss": 5.370018195658583, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.70990671262365}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.493051827045923, "early_stop_improved": true, "early_stop_metric": 11.493051827045923, "epoch": 21, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.365520695310672, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.492010539942136, "drafter_accepted_token_mae": 1.4772964232124104, "drafter_accepted_token_normalized_mae_percent": 36.507989460057864, "drafter_accepted_token_rounded_exact_fraction": 0.4127459776048437, "drafter_accepted_token_within_one_fraction": 0.5744635205716246, "hard_routing_tpot": 11.492109065351196, "macro_source_drafter_accepted_token_accuracy_percent": 57.60281890568135, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.850972509258419, "objective_total": 13.283037060258916, "prediction_oracle_tpot_regret_percent": 3.818786728100787, "regression_loss": 4.773548503334991, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.74853222039953}, "validation": {"drafter_accepted_token_accuracy_percent": 61.00990105567087, "drafter_accepted_token_mae": 1.5105206175832595, "drafter_accepted_token_normalized_mae_percent": 38.99009894432913, "drafter_accepted_token_rounded_exact_fraction": 0.4200690269602272, "drafter_accepted_token_within_one_fraction": 0.5747186798659174, "hard_routing_tpot": 11.493051827045923, "macro_source_drafter_accepted_token_accuracy_percent": 56.45977013511629, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.844319956511361, "objective_total": 13.453976165104564, "prediction_oracle_tpot_regret_percent": 5.0523326434377624, "regression_loss": 5.365520695310672, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.00836182060465}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.495541107308576, "early_stop_improved": false, "early_stop_metric": 11.495541107308576, "epoch": 22, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.409336043192033, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.56992021962604, "drafter_accepted_token_mae": 1.4741438066801695, "drafter_accepted_token_normalized_mae_percent": 36.43007978037396, "drafter_accepted_token_rounded_exact_fraction": 0.415565642589417, "drafter_accepted_token_within_one_fraction": 0.5764082705050506, "hard_routing_tpot": 11.493099492502006, "macro_source_drafter_accepted_token_accuracy_percent": 57.852012309946645, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.848684924357727, "objective_total": 13.289695483180935, "prediction_oracle_tpot_regret_percent": 3.88073782004128, "regression_loss": 4.803368529410692, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.96920820729841}, "validation": {"drafter_accepted_token_accuracy_percent": 61.1024351393264, "drafter_accepted_token_mae": 1.5069357423206853, "drafter_accepted_token_normalized_mae_percent": 38.8975648606736, "drafter_accepted_token_rounded_exact_fraction": 0.4229246843231465, "drafter_accepted_token_within_one_fraction": 0.5772764919046164, "hard_routing_tpot": 11.495541107308576, "macro_source_drafter_accepted_token_accuracy_percent": 56.75302475058487, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.845673567225637, "objective_total": 13.468474380183247, "prediction_oracle_tpot_regret_percent": 5.140601972580461, "regression_loss": 5.409336043192033, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.26146776713377}}
+{"checkpoint_improved": true, "checkpoint_metric": 11.487410921958647, "early_stop_improved": true, "early_stop_metric": 11.487410921958647, "epoch": 23, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.346699827930165, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.650274844368774, "drafter_accepted_token_mae": 1.4708922554039514, "drafter_accepted_token_normalized_mae_percent": 36.349725155631226, "drafter_accepted_token_rounded_exact_fraction": 0.4149736039322043, "drafter_accepted_token_within_one_fraction": 0.5757537523583985, "hard_routing_tpot": 11.487431084184818, "macro_source_drafter_accepted_token_accuracy_percent": 58.046326160796625, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.854889966497192, "objective_total": 13.28002030385811, "prediction_oracle_tpot_regret_percent": 3.8925862206027118, "regression_loss": 4.75043445786973, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 47.56381211097466}, "validation": {"drafter_accepted_token_accuracy_percent": 61.1739040298743, "drafter_accepted_token_mae": 1.5041669565106637, "drafter_accepted_token_normalized_mae_percent": 38.8260959701257, "drafter_accepted_token_rounded_exact_fraction": 0.4228175645135585, "drafter_accepted_token_within_one_fraction": 0.5767095407172846, "hard_routing_tpot": 11.487410921958647, "macro_source_drafter_accepted_token_accuracy_percent": 56.96673152951396, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.848872965263448, "objective_total": 13.452882913642497, "prediction_oracle_tpot_regret_percent": 5.138989175024378, "regression_loss": 5.346699827930165, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.80040918727787}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.492592046697052, "early_stop_improved": false, "early_stop_metric": 11.492592046697052, "epoch": 24, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.36139910750867, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.5006077373236, "drafter_accepted_token_mae": 1.47694853747208, "drafter_accepted_token_normalized_mae_percent": 36.4993922626764, "drafter_accepted_token_rounded_exact_fraction": 0.41252173891944455, "drafter_accepted_token_within_one_fraction": 0.5735150050863157, "hard_routing_tpot": 11.483265822011328, "macro_source_drafter_accepted_token_accuracy_percent": 57.68805197210513, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.845944415855532, "objective_total": 13.272772101226353, "prediction_oracle_tpot_regret_percent": 3.876090753458894, "regression_loss": 4.756092284569404, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.68487763091952}, "validation": {"drafter_accepted_token_accuracy_percent": 61.028218884056855, "drafter_accepted_token_mae": 1.5098109641534008, "drafter_accepted_token_normalized_mae_percent": 38.971781115943145, "drafter_accepted_token_rounded_exact_fraction": 0.42039299906727384, "drafter_accepted_token_within_one_fraction": 0.5746585882654167, "hard_routing_tpot": 11.492592046697052, "macro_source_drafter_accepted_token_accuracy_percent": 56.50926940167087, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.847829038604253, "objective_total": 13.456248770856854, "prediction_oracle_tpot_regret_percent": 5.129389130024942, "regression_loss": 5.36139910750867, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.87653490330595}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.495920976442786, "early_stop_improved": false, "early_stop_metric": 11.495920976442786, "epoch": 25, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.350460433825786, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.549665755583256, "drafter_accepted_token_mae": 1.4749634039170216, "drafter_accepted_token_normalized_mae_percent": 36.450334244416744, "drafter_accepted_token_rounded_exact_fraction": 0.41140964444722133, "drafter_accepted_token_within_one_fraction": 0.572902948728603, "hard_routing_tpot": 11.489863256121247, "macro_source_drafter_accepted_token_accuracy_percent": 57.82612976788534, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.85608608073046, "objective_total": 13.276243448001376, "prediction_oracle_tpot_regret_percent": 3.8735443206598053, "regression_loss": 4.733857890903053, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.94520037136986}, "validation": {"drafter_accepted_token_accuracy_percent": 61.0768630984234, "drafter_accepted_token_mae": 1.5079264321640944, "drafter_accepted_token_normalized_mae_percent": 38.9231369015766, "drafter_accepted_token_rounded_exact_fraction": 0.4194210827461339, "drafter_accepted_token_within_one_fraction": 0.5737232494402337, "hard_routing_tpot": 11.495920976442786, "macro_source_drafter_accepted_token_accuracy_percent": 56.679218694809244, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.853303027170021, "objective_total": 13.458441157317758, "prediction_oracle_tpot_regret_percent": 5.091982825452468, "regression_loss": 5.350460433825786, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.23893039325678}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.490964337921197, "early_stop_improved": false, "early_stop_metric": 11.490964337921197, "epoch": 26, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.376998041503274, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.57473868211799, "drafter_accepted_token_mae": 1.4739488275123211, "drafter_accepted_token_normalized_mae_percent": 36.42526131788201, "drafter_accepted_token_rounded_exact_fraction": 0.412925732511354, "drafter_accepted_token_within_one_fraction": 0.573807182634018, "hard_routing_tpot": 11.48417325200359, "macro_source_drafter_accepted_token_accuracy_percent": 57.87272237537495, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.846148917541909, "objective_total": 13.267573109908842, "prediction_oracle_tpot_regret_percent": 3.8206246247868725, "regression_loss": 4.738080641223111, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.56577512329721}, "validation": {"drafter_accepted_token_accuracy_percent": 61.03074232893831, "drafter_accepted_token_mae": 1.5097132030390674, "drafter_accepted_token_normalized_mae_percent": 38.96925767106169, "drafter_accepted_token_rounded_exact_fraction": 0.4203825483541433, "drafter_accepted_token_within_one_fraction": 0.5745906586300683, "hard_routing_tpot": 11.490964337921197, "macro_source_drafter_accepted_token_accuracy_percent": 56.71832940471976, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.84304470963344, "objective_total": 13.456144122084423, "prediction_oracle_tpot_regret_percent": 5.116260173425746, "regression_loss": 5.376998041503274, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.80521258991644}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.495256978545338, "early_stop_improved": false, "early_stop_metric": 11.495256978545338, "epoch": 27, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.381226564718246, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.59607264144454, "drafter_accepted_token_mae": 1.4730855484801915, "drafter_accepted_token_normalized_mae_percent": 36.40392735855546, "drafter_accepted_token_rounded_exact_fraction": 0.41167290753864577, "drafter_accepted_token_within_one_fraction": 0.5731239522300831, "hard_routing_tpot": 11.47941396314605, "macro_source_drafter_accepted_token_accuracy_percent": 57.810009247839574, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.84422198614558, "objective_total": 13.26077747834173, "prediction_oracle_tpot_regret_percent": 3.7533501589513962, "regression_loss": 4.7218516406538305, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.85068493828285}, "validation": {"drafter_accepted_token_accuracy_percent": 61.02114775893858, "drafter_accepted_token_mae": 1.5100849075536449, "drafter_accepted_token_normalized_mae_percent": 38.97885224106142, "drafter_accepted_token_rounded_exact_fraction": 0.4190187302906082, "drafter_accepted_token_within_one_fraction": 0.5738173058584085, "hard_routing_tpot": 11.495256978545338, "macro_source_drafter_accepted_token_accuracy_percent": 56.60559772761323, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.846164343383865, "objective_total": 13.460532312799339, "prediction_oracle_tpot_regret_percent": 5.029452658765043, "regression_loss": 5.381226564718246, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.149295220914986}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.494154724277529, "early_stop_improved": false, "early_stop_metric": 11.494154724277529, "epoch": 28, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.370331865906436, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.635052662116934, "drafter_accepted_token_mae": 1.471508221271333, "drafter_accepted_token_normalized_mae_percent": 36.364947337883066, "drafter_accepted_token_rounded_exact_fraction": 0.41275143697770733, "drafter_accepted_token_within_one_fraction": 0.5745300440409631, "hard_routing_tpot": 11.477841784606813, "macro_source_drafter_accepted_token_accuracy_percent": 57.88033293382618, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.845137277060344, "objective_total": 13.255335353737031, "prediction_oracle_tpot_regret_percent": 3.840027746840513, "regression_loss": 4.700660255588961, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 47.172677328098914}, "validation": {"drafter_accepted_token_accuracy_percent": 61.091577661427976, "drafter_accepted_token_mae": 1.507356373318415, "drafter_accepted_token_normalized_mae_percent": 38.908422338572024, "drafter_accepted_token_rounded_exact_fraction": 0.42039038638899123, "drafter_accepted_token_within_one_fraction": 0.5757977159966453, "hard_routing_tpot": 11.494154724277529, "macro_source_drafter_accepted_token_accuracy_percent": 56.60170023138399, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.848010391111627, "objective_total": 13.459109950883558, "prediction_oracle_tpot_regret_percent": 5.114950756237513, "regression_loss": 5.370331865906436, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.405022672178696}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.493168509666257, "early_stop_improved": false, "early_stop_metric": 11.493168509666257, "epoch": 29, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.3895176774812645, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.49052847343615, "drafter_accepted_token_mae": 1.4773563950591386, "drafter_accepted_token_normalized_mae_percent": 36.50947152656385, "drafter_accepted_token_rounded_exact_fraction": 0.4150785452105832, "drafter_accepted_token_within_one_fraction": 0.5743424033736498, "hard_routing_tpot": 11.484184703891199, "macro_source_drafter_accepted_token_accuracy_percent": 57.81408973941517, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.843495864426586, "objective_total": 13.264331577767773, "prediction_oracle_tpot_regret_percent": 3.8806704218811383, "regression_loss": 4.736119044470627, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.167028348204816}, "validation": {"drafter_accepted_token_accuracy_percent": 60.96558455322755, "drafter_accepted_token_mae": 1.5122374891083974, "drafter_accepted_token_normalized_mae_percent": 39.03441544677245, "drafter_accepted_token_rounded_exact_fraction": 0.4227705363044711, "drafter_accepted_token_within_one_fraction": 0.5757898779617974, "hard_routing_tpot": 11.493168509666257, "macro_source_drafter_accepted_token_accuracy_percent": 56.55957058442689, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.843202729729455, "objective_total": 13.460058032973835, "prediction_oracle_tpot_regret_percent": 5.119217643696672, "regression_loss": 5.3895176774812645, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.473771236611654}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.49228177073944, "early_stop_improved": false, "early_stop_metric": 11.49228177073944, "epoch": 30, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.327050319468851, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.710067350242575, "drafter_accepted_token_mae": 1.4684727506224318, "drafter_accepted_token_normalized_mae_percent": 36.289932649757425, "drafter_accepted_token_rounded_exact_fraction": 0.41427298441470367, "drafter_accepted_token_within_one_fraction": 0.5752994112167467, "hard_routing_tpot": 11.477111940620903, "macro_source_drafter_accepted_token_accuracy_percent": 57.94485081486074, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.84210911574826, "objective_total": 13.250300621454954, "prediction_oracle_tpot_regret_percent": 3.918787134235324, "regression_loss": 4.693971685688981, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.364684412638255}, "validation": {"drafter_accepted_token_accuracy_percent": 61.24128572475086, "drafter_accepted_token_mae": 1.5015565132926538, "drafter_accepted_token_normalized_mae_percent": 38.75871427524914, "drafter_accepted_token_rounded_exact_fraction": 0.42228980350046635, "drafter_accepted_token_within_one_fraction": 0.5770335128243314, "hard_routing_tpot": 11.49228177073944, "macro_source_drafter_accepted_token_accuracy_percent": 56.725432079069535, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.849334960950596, "objective_total": 13.44745005679125, "prediction_oracle_tpot_regret_percent": 5.192000378992501, "regression_loss": 5.327050319468851, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.61649690854868}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.492048558585392, "early_stop_improved": false, "early_stop_metric": 11.492048558585392, "epoch": 31, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.37900834611549, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.61561472258087, "drafter_accepted_token_mae": 1.4722947778299367, "drafter_accepted_token_normalized_mae_percent": 36.38438527741913, "drafter_accepted_token_rounded_exact_fraction": 0.41101798499400377, "drafter_accepted_token_within_one_fraction": 0.5724997639326734, "hard_routing_tpot": 11.47403985151681, "macro_source_drafter_accepted_token_accuracy_percent": 57.79834008542189, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.840181306214726, "objective_total": 13.249232945038013, "prediction_oracle_tpot_regret_percent": 3.840907785272396, "regression_loss": 4.696838796077624, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.518552992701004}, "validation": {"drafter_accepted_token_accuracy_percent": 61.036762216265636, "drafter_accepted_token_mae": 1.5094799857821313, "drafter_accepted_token_normalized_mae_percent": 38.963237783734364, "drafter_accepted_token_rounded_exact_fraction": 0.41890115976788966, "drafter_accepted_token_within_one_fraction": 0.573485495716514, "hard_routing_tpot": 11.492048558585392, "macro_source_drafter_accepted_token_accuracy_percent": 56.545049664285145, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.844960165002501, "objective_total": 13.458662668837148, "prediction_oracle_tpot_regret_percent": 5.12602628774772, "regression_loss": 5.37900834611549, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.89696547133512}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.494670473093986, "early_stop_improved": false, "early_stop_metric": 11.494670473093986, "epoch": 32, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.353123018582634, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.55971235429383, "drafter_accepted_token_mae": 1.4745568681269086, "drafter_accepted_token_normalized_mae_percent": 36.44028764570617, "drafter_accepted_token_rounded_exact_fraction": 0.4117026307909034, "drafter_accepted_token_within_one_fraction": 0.5727296641899312, "hard_routing_tpot": 11.480003442722232, "macro_source_drafter_accepted_token_accuracy_percent": 57.763320236556424, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.842779964814017, "objective_total": 13.25045685328093, "prediction_oracle_tpot_regret_percent": 3.7869853321347735, "regression_loss": 4.692256294889711, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.155900989929}, "validation": {"drafter_accepted_token_accuracy_percent": 61.00711746862214, "drafter_accepted_token_mae": 1.5106284568999526, "drafter_accepted_token_normalized_mae_percent": 38.99288253137786, "drafter_accepted_token_rounded_exact_fraction": 0.4194837870249171, "drafter_accepted_token_within_one_fraction": 0.5736082915957977, "hard_routing_tpot": 11.494670473093986, "macro_source_drafter_accepted_token_accuracy_percent": 56.53410580672072, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.84287257986974, "objective_total": 13.44880948544453, "prediction_oracle_tpot_regret_percent": 5.055231278008798, "regression_loss": 5.353123018582634, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.47488392730677}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.49214543179734, "early_stop_improved": false, "early_stop_metric": 11.49214543179734, "epoch": 33, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.380286482464085, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.72633942709654, "drafter_accepted_token_mae": 1.4678143007518656, "drafter_accepted_token_normalized_mae_percent": 36.27366057290346, "drafter_accepted_token_rounded_exact_fraction": 0.41230821678077767, "drafter_accepted_token_within_one_fraction": 0.5753216531061911, "hard_routing_tpot": 11.477432297284007, "macro_source_drafter_accepted_token_accuracy_percent": 58.08162041718783, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.839452020011734, "objective_total": 13.250150952083645, "prediction_oracle_tpot_regret_percent": 3.6843441341824157, "regression_loss": 4.702329773573037, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.799046410875135}, "validation": {"drafter_accepted_token_accuracy_percent": 61.16462004986468, "drafter_accepted_token_mae": 1.5045266284170566, "drafter_accepted_token_normalized_mae_percent": 38.83537995013532, "drafter_accepted_token_rounded_exact_fraction": 0.4202728158662727, "drafter_accepted_token_within_one_fraction": 0.5767382801783937, "hard_routing_tpot": 11.49214543179734, "macro_source_drafter_accepted_token_accuracy_percent": 56.81017643781397, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.845129188632738, "objective_total": 13.459215133371963, "prediction_oracle_tpot_regret_percent": 5.027446876699337, "regression_loss": 5.380286482464085, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.14479117604031}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.497835477689033, "early_stop_improved": false, "early_stop_metric": 11.497835477689033, "epoch": 34, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.363463571813643, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.59301087400491, "drafter_accepted_token_mae": 1.4732094429524476, "drafter_accepted_token_normalized_mae_percent": 36.40698912599509, "drafter_accepted_token_rounded_exact_fraction": 0.409659005548947, "drafter_accepted_token_within_one_fraction": 0.5715259735729957, "hard_routing_tpot": 11.474221983446272, "macro_source_drafter_accepted_token_accuracy_percent": 57.77872633408102, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.836283494458845, "objective_total": 13.244927540800168, "prediction_oracle_tpot_regret_percent": 3.80424193997404, "regression_loss": 4.695480154471077, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 45.86349806607318}, "validation": {"drafter_accepted_token_accuracy_percent": 61.04211495418674, "drafter_accepted_token_mae": 1.509272614649222, "drafter_accepted_token_normalized_mae_percent": 38.95788504581326, "drafter_accepted_token_rounded_exact_fraction": 0.41764968687050785, "drafter_accepted_token_within_one_fraction": 0.5727591711539416, "hard_routing_tpot": 11.497835477689033, "macro_source_drafter_accepted_token_accuracy_percent": 56.49596892130365, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.843952872038297, "objective_total": 13.45299194358239, "prediction_oracle_tpot_regret_percent": 5.119126363022275, "regression_loss": 5.363463571813643, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.13509859298172}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.492847854435936, "early_stop_improved": false, "early_stop_metric": 11.492847854435936, "epoch": 35, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.343156325731427, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.70233233981071, "drafter_accepted_token_mae": 1.4687857479529753, "drafter_accepted_token_normalized_mae_percent": 36.29766766018929, "drafter_accepted_token_rounded_exact_fraction": 0.41187510653359544, "drafter_accepted_token_within_one_fraction": 0.5733722525958812, "hard_routing_tpot": 11.47142511402886, "macro_source_drafter_accepted_token_accuracy_percent": 57.91177807546119, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.840449235185025, "objective_total": 13.241050412640313, "prediction_oracle_tpot_regret_percent": 3.874350210682276, "regression_loss": 4.668670591517625, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.50021419809315}, "validation": {"drafter_accepted_token_accuracy_percent": 61.14140272365767, "drafter_accepted_token_mae": 1.5054260939447262, "drafter_accepted_token_normalized_mae_percent": 38.85859727634233, "drafter_accepted_token_rounded_exact_fraction": 0.41900827957747766, "drafter_accepted_token_within_one_fraction": 0.5748649898497449, "hard_routing_tpot": 11.492847854435936, "macro_source_drafter_accepted_token_accuracy_percent": 56.624556587047756, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.849149992691286, "objective_total": 13.452096890410715, "prediction_oracle_tpot_regret_percent": 5.207947710026328, "regression_loss": 5.343156325731427, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.85133713188868}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.494217795964197, "early_stop_improved": false, "early_stop_metric": 11.494217795964197, "epoch": 36, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.377442301629187, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.64100523183274, "drafter_accepted_token_mae": 1.4712673504351095, "drafter_accepted_token_normalized_mae_percent": 36.35899476816726, "drafter_accepted_token_rounded_exact_fraction": 0.411527526461277, "drafter_accepted_token_within_one_fraction": 0.5732658959245377, "hard_routing_tpot": 11.469373197838271, "macro_source_drafter_accepted_token_accuracy_percent": 57.789657090881256, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.83202419681811, "objective_total": 13.238694078951294, "prediction_oracle_tpot_regret_percent": 3.8076993883730648, "regression_loss": 4.688899607110612, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.32307627365193}, "validation": {"drafter_accepted_token_accuracy_percent": 61.044536054529466, "drafter_accepted_token_mae": 1.5091788184782977, "drafter_accepted_token_normalized_mae_percent": 38.955463945470534, "drafter_accepted_token_rounded_exact_fraction": 0.41886980762849807, "drafter_accepted_token_within_one_fraction": 0.5746821023699604, "hard_routing_tpot": 11.494217795964197, "macro_source_drafter_accepted_token_accuracy_percent": 56.4620481311776, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.841471662370592, "objective_total": 13.454704352859348, "prediction_oracle_tpot_regret_percent": 5.1292614708537965, "regression_loss": 5.377442301629187, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.58573299045559}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.4916309995258, "early_stop_improved": false, "early_stop_metric": 11.4916309995258, "epoch": 37, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.39305590922281, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.61780921591124, "drafter_accepted_token_mae": 1.4722059776194716, "drafter_accepted_token_normalized_mae_percent": 36.38219078408876, "drafter_accepted_token_rounded_exact_fraction": 0.4134049441293847, "drafter_accepted_token_within_one_fraction": 0.5755812766157065, "hard_routing_tpot": 11.468461219553381, "macro_source_drafter_accepted_token_accuracy_percent": 58.195289591538085, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.847489236399625, "objective_total": 13.257240411449041, "prediction_oracle_tpot_regret_percent": 3.8452228216126816, "regression_loss": 4.699170583498058, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 47.17739935188848}, "validation": {"drafter_accepted_token_accuracy_percent": 61.06263891986886, "drafter_accepted_token_mae": 1.508477492960465, "drafter_accepted_token_normalized_mae_percent": 38.93736108013114, "drafter_accepted_token_rounded_exact_fraction": 0.42075616134856003, "drafter_accepted_token_within_one_fraction": 0.5775325343763145, "hard_routing_tpot": 11.4916309995258, "macro_source_drafter_accepted_token_accuracy_percent": 56.87474361246344, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.857295365467097, "objective_total": 13.47521213823394, "prediction_oracle_tpot_regret_percent": 5.151642647934018, "regression_loss": 5.39305590922281, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.4407451645856}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.493105223658324, "early_stop_improved": false, "early_stop_metric": 11.493105223658324, "epoch": 38, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.344648543368292, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.941330688136006, "drafter_accepted_token_mae": 1.4591146756656088, "drafter_accepted_token_normalized_mae_percent": 36.058669311863994, "drafter_accepted_token_rounded_exact_fraction": 0.41565804753010893, "drafter_accepted_token_within_one_fraction": 0.5770953426898896, "hard_routing_tpot": 11.469779831074872, "macro_source_drafter_accepted_token_accuracy_percent": 58.12296958348506, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.8349705478238, "objective_total": 13.231741504970438, "prediction_oracle_tpot_regret_percent": 3.89497335585971, "regression_loss": 4.65590319048879, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 47.599428020577456}, "validation": {"drafter_accepted_token_accuracy_percent": 61.33507112516042, "drafter_accepted_token_mae": 1.4979231606010075, "drafter_accepted_token_normalized_mae_percent": 38.66492887483958, "drafter_accepted_token_rounded_exact_fraction": 0.42294558574940755, "drafter_accepted_token_within_one_fraction": 0.5784704858797802, "hard_routing_tpot": 11.493105223658324, "macro_source_drafter_accepted_token_accuracy_percent": 56.87080828106089, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.84391562312646, "objective_total": 13.447310186136948, "prediction_oracle_tpot_regret_percent": 5.181144964699702, "regression_loss": 5.344648543368292, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.90623777932807}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.490754915427605, "early_stop_improved": false, "early_stop_metric": 11.490754915427605, "epoch": 39, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.335998257875497, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.8939265799043, "drafter_accepted_token_mae": 1.4610328837228583, "drafter_accepted_token_normalized_mae_percent": 36.1060734200957, "drafter_accepted_token_rounded_exact_fraction": 0.41470083748801717, "drafter_accepted_token_within_one_fraction": 0.5762467135080859, "hard_routing_tpot": 11.469652916057669, "macro_source_drafter_accepted_token_accuracy_percent": 58.27508235593561, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.835517180170752, "objective_total": 13.229415482691218, "prediction_oracle_tpot_regret_percent": 3.8391714221202466, "regression_loss": 4.64632767506822, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 47.536372974806405}, "validation": {"drafter_accepted_token_accuracy_percent": 61.29266887391566, "drafter_accepted_token_mae": 1.4995658718654392, "drafter_accepted_token_normalized_mae_percent": 38.70733112608434, "drafter_accepted_token_rounded_exact_fraction": 0.42177771855707, "drafter_accepted_token_within_one_fraction": 0.5774672174192487, "hard_routing_tpot": 11.490754915427605, "macro_source_drafter_accepted_token_accuracy_percent": 57.07051402807271, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.844427007546587, "objective_total": 13.445226484909236, "prediction_oracle_tpot_regret_percent": 5.125882135748361, "regression_loss": 5.335998257875497, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.81600607728544}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.496662772959564, "early_stop_improved": false, "early_stop_metric": 11.496662772959564, "epoch": 40, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.363182417015849, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 64.02554196304746, "drafter_accepted_token_mae": 1.4557070649737909, "drafter_accepted_token_normalized_mae_percent": 35.974458036952534, "drafter_accepted_token_rounded_exact_fraction": 0.41819422952376273, "drafter_accepted_token_within_one_fraction": 0.5797615386372961, "hard_routing_tpot": 11.473133970005524, "macro_source_drafter_accepted_token_accuracy_percent": 58.28573421070917, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.838764727971046, "objective_total": 13.235098545812047, "prediction_oracle_tpot_regret_percent": 3.714830327533414, "regression_loss": 4.654446059470004, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 47.49597557438137}, "validation": {"drafter_accepted_token_accuracy_percent": 61.337989854418595, "drafter_accepted_token_mae": 1.4978100857219745, "drafter_accepted_token_normalized_mae_percent": 38.662010145581405, "drafter_accepted_token_rounded_exact_fraction": 0.42418660793365887, "drafter_accepted_token_within_one_fraction": 0.5798238532301848, "hard_routing_tpot": 11.496662772959564, "macro_source_drafter_accepted_token_accuracy_percent": 56.943731832469574, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.84852360341438, "objective_total": 13.457478328519134, "prediction_oracle_tpot_regret_percent": 5.040528951969625, "regression_loss": 5.363182417015849, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.747968561090204}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.494728921564786, "early_stop_improved": false, "early_stop_metric": 11.494728921564786, "epoch": 41, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.406093758926714, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.69576891323227, "drafter_accepted_token_mae": 1.4690513371226814, "drafter_accepted_token_normalized_mae_percent": 36.30423108676773, "drafter_accepted_token_rounded_exact_fraction": 0.41231023877072714, "drafter_accepted_token_within_one_fraction": 0.5741586044872405, "hard_routing_tpot": 11.468451723704096, "macro_source_drafter_accepted_token_accuracy_percent": 58.058253654661286, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.830852494986711, "objective_total": 13.23370400697445, "prediction_oracle_tpot_regret_percent": 3.7286170622251813, "regression_loss": 4.676171706625799, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 47.20479625213354}, "validation": {"drafter_accepted_token_accuracy_percent": 61.018683323520904, "drafter_accepted_token_mae": 1.5101803825744808, "drafter_accepted_token_normalized_mae_percent": 38.981316676479096, "drafter_accepted_token_rounded_exact_fraction": 0.4194681109552213, "drafter_accepted_token_within_one_fraction": 0.5751027435734646, "hard_routing_tpot": 11.494728921564786, "macro_source_drafter_accepted_token_accuracy_percent": 56.74636872184341, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.84218320225914, "objective_total": 13.464011329937154, "prediction_oracle_tpot_regret_percent": 5.0681063431145645, "regression_loss": 5.406093758926714, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.541321980052565}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.490739361827202, "early_stop_improved": false, "early_stop_metric": 11.490739361827202, "epoch": 42, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.372520804908836, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.92038940141928, "drafter_accepted_token_mae": 1.4599620651938088, "drafter_accepted_token_normalized_mae_percent": 36.07961059858072, "drafter_accepted_token_rounded_exact_fraction": 0.4169046043339737, "drafter_accepted_token_within_one_fraction": 0.5778281118475872, "hard_routing_tpot": 11.466565297122957, "macro_source_drafter_accepted_token_accuracy_percent": 58.09531034078521, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.826268610933266, "objective_total": 13.226286021263899, "prediction_oracle_tpot_regret_percent": 3.907990754816737, "regression_loss": 4.66672470110211, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 46.94785944300112}, "validation": {"drafter_accepted_token_accuracy_percent": 61.271556789359146, "drafter_accepted_token_mae": 1.50038377794586, "drafter_accepted_token_normalized_mae_percent": 38.728443210640854, "drafter_accepted_token_rounded_exact_fraction": 0.42358830460693564, "drafter_accepted_token_within_one_fraction": 0.5783685914267575, "hard_routing_tpot": 11.490739361827202, "macro_source_drafter_accepted_token_accuracy_percent": 56.76998658874663, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.839775076157878, "objective_total": 13.45153131763053, "prediction_oracle_tpot_regret_percent": 5.210544145650586, "regression_loss": 5.372520804908836, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 46.196183765831584}}
+{"checkpoint_improved": false, "checkpoint_metric": 11.490566220862105, "early_stop_improved": false, "early_stop_metric": 11.490566220862105, "epoch": 43, "fit_seed": 7, "learning_rate": 0.001, "regression_checkpoint_metric": 5.3399222075906945, "rollout_round": 0, "selected_by_configured_checkpoint": false, "selected_by_early_stop_metric": false, "selected_by_original_checkpoint": false, "selected_by_routing_tpot": false, "train": {"drafter_accepted_token_accuracy_percent": 63.85372760989606, "drafter_accepted_token_mae": 1.462659535737617, "drafter_accepted_token_normalized_mae_percent": 36.14627239010394, "drafter_accepted_token_rounded_exact_fraction": 0.4137887178217992, "drafter_accepted_token_within_one_fraction": 0.5759618151242017, "hard_routing_tpot": 11.468049307027396, "macro_source_drafter_accepted_token_accuracy_percent": 57.93939179151964, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.828035512744973, "objective_total": 13.226659472608908, "prediction_oracle_tpot_regret_percent": 3.818259499084564, "regression_loss": 4.6620798662131175, "rows": 1648541, "worst_source_drafter_accepted_token_accuracy_percent": 45.88351973841585}, "validation": {"drafter_accepted_token_accuracy_percent": 61.27585486483292, "drafter_accepted_token_mae": 1.5002172656313275, "drafter_accepted_token_normalized_mae_percent": 38.72414513516708, "drafter_accepted_token_rounded_exact_fraction": 0.42140410556265334, "drafter_accepted_token_within_one_fraction": 0.5772477524435073, "hard_routing_tpot": 11.490566220862105, "macro_source_drafter_accepted_token_accuracy_percent": 56.68874856262728, "mlp_objective": "hybrid", "objective_rank_loss": 0.0, "objective_routing_loss": 11.838105040875593, "objective_total": 13.440081703152801, "prediction_oracle_tpot_regret_percent": 5.1167366933048575, "regression_loss": 5.3399222075906945, "rows": 127583, "worst_source_drafter_accepted_token_accuracy_percent": 45.135184432211055}}

results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-margin-sweep/command.json

@@ -0,0 +1,86 @@
+{
+ "gpu": "1",
+ "command": [
+  "/home/heck2/sbhansali8/SpecRouter/envs/hedgespec/bin/python",
+  "/home/heck2/sbhansali8/SpecRouter/implementations/specrouter/scripts/train_mlp_screen.py",
+  "--store",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/hvd-training-10k-campaign-v2/mlp-sweeps/screening/llama31-8b/base",
+  "--output-dir",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-margin-sweep",
+  "--train-request-count",
+  "10000",
+  "--train-request-selection-mode",
+  "source-balanced",
+  "--request-selection-seed",
+  "42",
+  "--validation-fold-count",
+  "2",
+  "--validation-fold-index",
+  "0",
+  "--validation-fold-seed",
+  "42",
+  "--batch-size",
+  "8192",
+  "--max-epochs",
+  "60",
+  "--patience",
+  "20",
+  "--causal-refit-rounds",
+  "0",
+  "--device",
+  "cuda",
+  "--preload-hidden-to-device",
+  "--max-hidden-bytes",
+  "128849018880",
+  "--hidden-layer-slots",
+  "1",
+  "--hidden-sample-dim",
+  "384",
+  "--hidden-fusion",
+  "flat",
+  "--projection-dim",
+  "32",
+  "--width",
+  "64",
+  "--depth",
+  "2",
+  "--activation",
+  "gelu",
+  "--normalization",
+  "layernorm",
+  "--output-link",
+  "softplus",
+  "--loss",
+  "mse",
+  "--objective",
+  "hybrid",
+  "--routing-temperature",
+  "0.2",
+  "--regression-weight",
+  "0.3",
+  "--optimizer",
+  "adamw",
+  "--learning-rate",
+  "0.001",
+  "--weight-decay",
+  "0.001",
+  "--scalar-features",
+  "ngram_hit_len",
+  "log_ngram_hit",
+  "is_repetition",
+  "seg_pos",
+  "log_seg_pos",
+  "context_len",
+  "log_context_len",
+  "prompt_len",
+  "--causal-refit-rounds",
+  "2",
+  "--decision-margin-sweep",
+  "0.0",
+  "0.02",
+  "0.05",
+  "0.10",
+  "0.20",
+  "0.35"
+ ]
+}

results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-ngram-scalars-only/command.json

@@ -0,0 +1,71 @@
+{
+ "gpu": "3",
+ "command": [
+  "/home/heck2/sbhansali8/SpecRouter/envs/hedgespec/bin/python",
+  "/home/heck2/sbhansali8/SpecRouter/implementations/specrouter/scripts/train_mlp_screen.py",
+  "--store",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/hvd-training-10k-campaign-v2/mlp-sweeps/screening/llama31-8b/base",
+  "--output-dir",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-ngram-scalars-only",
+  "--train-request-count",
+  "10000",
+  "--train-request-selection-mode",
+  "source-balanced",
+  "--request-selection-seed",
+  "42",
+  "--validation-fold-count",
+  "2",
+  "--validation-fold-index",
+  "0",
+  "--validation-fold-seed",
+  "42",
+  "--batch-size",
+  "8192",
+  "--max-epochs",
+  "60",
+  "--patience",
+  "20",
+  "--causal-refit-rounds",
+  "0",
+  "--device",
+  "cuda",
+  "--preload-hidden-to-device",
+  "--max-hidden-bytes",
+  "68719476736",
+  "--hidden-layer-slots",
+  "1",
+  "--hidden-sample-dim",
+  "384",
+  "--hidden-fusion",
+  "flat",
+  "--projection-dim",
+  "32",
+  "--width",
+  "64",
+  "--depth",
+  "2",
+  "--activation",
+  "gelu",
+  "--normalization",
+  "layernorm",
+  "--output-link",
+  "softplus",
+  "--loss",
+  "mse",
+  "--objective",
+  "hybrid",
+  "--routing-temperature",
+  "0.2",
+  "--regression-weight",
+  "0.3",
+  "--optimizer",
+  "adamw",
+  "--learning-rate",
+  "0.001",
+  "--weight-decay",
+  "0.001",
+  "--scalar-features",
+  "ngram_hit_len",
+  "log_ngram_hit"
+ ]
+}

results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-norm-minmax/command.json

@@ -0,0 +1,70 @@
+{
+ "gpu": "1",
+ "command": [
+  "/home/heck2/sbhansali8/SpecRouter/envs/hedgespec/bin/python",
+  "/home/heck2/sbhansali8/SpecRouter/implementations/specrouter/scripts/train_mlp_screen.py",
+  "--store",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/hvd-training-10k-campaign-v2/mlp-sweeps/screening/llama31-8b/base",
+  "--output-dir",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-norm-minmax",
+  "--train-request-count",
+  "10000",
+  "--train-request-selection-mode",
+  "source-balanced",
+  "--request-selection-seed",
+  "42",
+  "--validation-fold-count",
+  "2",
+  "--validation-fold-index",
+  "0",
+  "--validation-fold-seed",
+  "42",
+  "--batch-size",
+  "8192",
+  "--max-epochs",
+  "60",
+  "--patience",
+  "20",
+  "--causal-refit-rounds",
+  "0",
+  "--device",
+  "cuda",
+  "--preload-hidden-to-device",
+  "--max-hidden-bytes",
+  "68719476736",
+  "--hidden-layer-slots",
+  "1",
+  "--hidden-sample-dim",
+  "384",
+  "--hidden-fusion",
+  "flat",
+  "--projection-dim",
+  "32",
+  "--width",
+  "64",
+  "--depth",
+  "2",
+  "--activation",
+  "gelu",
+  "--normalization",
+  "layernorm",
+  "--output-link",
+  "softplus",
+  "--loss",
+  "mse",
+  "--objective",
+  "hybrid",
+  "--routing-temperature",
+  "0.2",
+  "--regression-weight",
+  "0.3",
+  "--optimizer",
+  "adamw",
+  "--learning-rate",
+  "0.001",
+  "--weight-decay",
+  "0.001",
+  "--scalar-normalization",
+  "minmax"
+ ]
+}

results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-norm-none-seed11/command.json

@@ -0,0 +1,72 @@
+{
+ "gpu": "2",
+ "command": [
+  "/home/heck2/sbhansali8/SpecRouter/envs/hedgespec/bin/python",
+  "/home/heck2/sbhansali8/SpecRouter/implementations/specrouter/scripts/train_mlp_screen.py",
+  "--store",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/hvd-training-10k-campaign-v2/mlp-sweeps/screening/llama31-8b/base",
+  "--output-dir",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-norm-none-seed11",
+  "--train-request-count",
+  "10000",
+  "--train-request-selection-mode",
+  "source-balanced",
+  "--request-selection-seed",
+  "42",
+  "--validation-fold-count",
+  "2",
+  "--validation-fold-index",
+  "0",
+  "--validation-fold-seed",
+  "42",
+  "--batch-size",
+  "8192",
+  "--max-epochs",
+  "60",
+  "--patience",
+  "20",
+  "--causal-refit-rounds",
+  "0",
+  "--device",
+  "cuda",
+  "--preload-hidden-to-device",
+  "--max-hidden-bytes",
+  "128849018880",
+  "--hidden-layer-slots",
+  "1",
+  "--hidden-sample-dim",
+  "384",
+  "--hidden-fusion",
+  "flat",
+  "--projection-dim",
+  "32",
+  "--width",
+  "64",
+  "--depth",
+  "2",
+  "--activation",
+  "gelu",
+  "--normalization",
+  "layernorm",
+  "--output-link",
+  "softplus",
+  "--loss",
+  "mse",
+  "--objective",
+  "hybrid",
+  "--routing-temperature",
+  "0.2",
+  "--regression-weight",
+  "0.3",
+  "--optimizer",
+  "adamw",
+  "--learning-rate",
+  "0.001",
+  "--weight-decay",
+  "0.001",
+  "--scalar-normalization",
+  "none",
+  "--model-seed",
+  "11"
+ ]
+}

results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-norm-none-seed23/command.json

@@ -0,0 +1,72 @@
+{
+ "gpu": "3",
+ "command": [
+  "/home/heck2/sbhansali8/SpecRouter/envs/hedgespec/bin/python",
+  "/home/heck2/sbhansali8/SpecRouter/implementations/specrouter/scripts/train_mlp_screen.py",
+  "--store",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/hvd-training-10k-campaign-v2/mlp-sweeps/screening/llama31-8b/base",
+  "--output-dir",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-norm-none-seed23",
+  "--train-request-count",
+  "10000",
+  "--train-request-selection-mode",
+  "source-balanced",
+  "--request-selection-seed",
+  "42",
+  "--validation-fold-count",
+  "2",
+  "--validation-fold-index",
+  "0",
+  "--validation-fold-seed",
+  "42",
+  "--batch-size",
+  "8192",
+  "--max-epochs",
+  "60",
+  "--patience",
+  "20",
+  "--causal-refit-rounds",
+  "0",
+  "--device",
+  "cuda",
+  "--preload-hidden-to-device",
+  "--max-hidden-bytes",
+  "128849018880",
+  "--hidden-layer-slots",
+  "1",
+  "--hidden-sample-dim",
+  "384",
+  "--hidden-fusion",
+  "flat",
+  "--projection-dim",
+  "32",
+  "--width",
+  "64",
+  "--depth",
+  "2",
+  "--activation",
+  "gelu",
+  "--normalization",
+  "layernorm",
+  "--output-link",
+  "softplus",
+  "--loss",
+  "mse",
+  "--objective",
+  "hybrid",
+  "--routing-temperature",
+  "0.2",
+  "--regression-weight",
+  "0.3",
+  "--optimizer",
+  "adamw",
+  "--learning-rate",
+  "0.001",
+  "--weight-decay",
+  "0.001",
+  "--scalar-normalization",
+  "none",
+  "--model-seed",
+  "23"
+ ]
+}

results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-norm-none-seed7/command.json

@@ -0,0 +1,72 @@
+{
+ "gpu": "1",
+ "command": [
+  "/home/heck2/sbhansali8/SpecRouter/envs/hedgespec/bin/python",
+  "/home/heck2/sbhansali8/SpecRouter/implementations/specrouter/scripts/train_mlp_screen.py",
+  "--store",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/hvd-training-10k-campaign-v2/mlp-sweeps/screening/llama31-8b/base",
+  "--output-dir",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-norm-none-seed7",
+  "--train-request-count",
+  "10000",
+  "--train-request-selection-mode",
+  "source-balanced",
+  "--request-selection-seed",
+  "42",
+  "--validation-fold-count",
+  "2",
+  "--validation-fold-index",
+  "0",
+  "--validation-fold-seed",
+  "42",
+  "--batch-size",
+  "8192",
+  "--max-epochs",
+  "60",
+  "--patience",
+  "20",
+  "--causal-refit-rounds",
+  "0",
+  "--device",
+  "cuda",
+  "--preload-hidden-to-device",
+  "--max-hidden-bytes",
+  "128849018880",
+  "--hidden-layer-slots",
+  "1",
+  "--hidden-sample-dim",
+  "384",
+  "--hidden-fusion",
+  "flat",
+  "--projection-dim",
+  "32",
+  "--width",
+  "64",
+  "--depth",
+  "2",
+  "--activation",
+  "gelu",
+  "--normalization",
+  "layernorm",
+  "--output-link",
+  "softplus",
+  "--loss",
+  "mse",
+  "--objective",
+  "hybrid",
+  "--routing-temperature",
+  "0.2",
+  "--regression-weight",
+  "0.3",
+  "--optimizer",
+  "adamw",
+  "--learning-rate",
+  "0.001",
+  "--weight-decay",
+  "0.001",
+  "--scalar-normalization",
+  "none",
+  "--model-seed",
+  "7"
+ ]
+}

results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-norm-none/command.json

@@ -0,0 +1,70 @@
+{
+ "gpu": "3",
+ "command": [
+  "/home/heck2/sbhansali8/SpecRouter/envs/hedgespec/bin/python",
+  "/home/heck2/sbhansali8/SpecRouter/implementations/specrouter/scripts/train_mlp_screen.py",
+  "--store",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/hvd-training-10k-campaign-v2/mlp-sweeps/screening/llama31-8b/base",
+  "--output-dir",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-norm-none",
+  "--train-request-count",
+  "10000",
+  "--train-request-selection-mode",
+  "source-balanced",
+  "--request-selection-seed",
+  "42",
+  "--validation-fold-count",
+  "2",
+  "--validation-fold-index",
+  "0",
+  "--validation-fold-seed",
+  "42",
+  "--batch-size",
+  "8192",
+  "--max-epochs",
+  "60",
+  "--patience",
+  "20",
+  "--causal-refit-rounds",
+  "0",
+  "--device",
+  "cuda",
+  "--preload-hidden-to-device",
+  "--max-hidden-bytes",
+  "68719476736",
+  "--hidden-layer-slots",
+  "1",
+  "--hidden-sample-dim",
+  "384",
+  "--hidden-fusion",
+  "flat",
+  "--projection-dim",
+  "32",
+  "--width",
+  "64",
+  "--depth",
+  "2",
+  "--activation",
+  "gelu",
+  "--normalization",
+  "layernorm",
+  "--output-link",
+  "softplus",
+  "--loss",
+  "mse",
+  "--objective",
+  "hybrid",
+  "--routing-temperature",
+  "0.2",
+  "--regression-weight",
+  "0.3",
+  "--optimizer",
+  "adamw",
+  "--learning-rate",
+  "0.001",
+  "--weight-decay",
+  "0.001",
+  "--scalar-normalization",
+  "none"
+ ]
+}

results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-refit2-prod-seed11/command.json

@@ -0,0 +1,72 @@
+{
+ "gpu": "0",
+ "command": [
+  "/home/heck2/sbhansali8/SpecRouter/envs/hedgespec/bin/python",
+  "/home/heck2/sbhansali8/SpecRouter/implementations/specrouter/scripts/train_mlp_screen.py",
+  "--store",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/hvd-training-10k-campaign-v2/mlp-sweeps/screening/llama31-8b/base",
+  "--output-dir",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-refit2-prod-seed11",
+  "--train-request-count",
+  "10000",
+  "--train-request-selection-mode",
+  "source-balanced",
+  "--request-selection-seed",
+  "42",
+  "--validation-fold-count",
+  "2",
+  "--validation-fold-index",
+  "0",
+  "--validation-fold-seed",
+  "42",
+  "--batch-size",
+  "8192",
+  "--max-epochs",
+  "60",
+  "--patience",
+  "20",
+  "--causal-refit-rounds",
+  "0",
+  "--device",
+  "cuda",
+  "--preload-hidden-to-device",
+  "--max-hidden-bytes",
+  "128849018880",
+  "--hidden-layer-slots",
+  "1",
+  "--hidden-sample-dim",
+  "384",
+  "--hidden-fusion",
+  "flat",
+  "--projection-dim",
+  "32",
+  "--width",
+  "64",
+  "--depth",
+  "2",
+  "--activation",
+  "gelu",
+  "--normalization",
+  "layernorm",
+  "--output-link",
+  "softplus",
+  "--loss",
+  "mse",
+  "--objective",
+  "hybrid",
+  "--routing-temperature",
+  "0.2",
+  "--regression-weight",
+  "0.3",
+  "--optimizer",
+  "adamw",
+  "--learning-rate",
+  "0.001",
+  "--weight-decay",
+  "0.001",
+  "--causal-refit-rounds",
+  "2",
+  "--model-seed",
+  "11"
+ ]
+}

results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-refit2-prod-seed23/command.json

@@ -0,0 +1,72 @@
+{
+ "gpu": "1",
+ "command": [
+  "/home/heck2/sbhansali8/SpecRouter/envs/hedgespec/bin/python",
+  "/home/heck2/sbhansali8/SpecRouter/implementations/specrouter/scripts/train_mlp_screen.py",
+  "--store",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/hvd-training-10k-campaign-v2/mlp-sweeps/screening/llama31-8b/base",
+  "--output-dir",
+  "/home/heck2/sbhansali8/SpecRouter/results/specrouter/repr-sampling-campaign-v1/cells/llama31-8b-small-refit2-prod-seed23",
+  "--train-request-count",
+  "10000",
+  "--train-request-selection-mode",
+  "source-balanced",
+  "--request-selection-seed",
+  "42",
+  "--validation-fold-count",
+  "2",
+  "--validation-fold-index",
+  "0",
+  "--validation-fold-seed",
+  "42",
+  "--batch-size",
+  "8192",
+  "--max-epochs",
+  "60",
+  "--patience",
+  "20",
+  "--causal-refit-rounds",
+  "0",
+  "--device",
+  "cuda",
+  "--preload-hidden-to-device",
+  "--max-hidden-bytes",
+  "128849018880",
+  "--hidden-layer-slots",
+  "1",
+  "--hidden-sample-dim",
+  "384",
+  "--hidden-fusion",
+  "flat",
+  "--projection-dim",
+  "32",
+  "--width",
+  "64",
+  "--depth",
+  "2",
+  "--activation",
+  "gelu",
+  "--normalization",
+  "layernorm",
+  "--output-link",
+  "softplus",
+  "--loss",
+  "mse",
+  "--objective",
+  "hybrid",
+  "--routing-temperature",
+  "0.2",
+  "--regression-weight",
+  "0.3",
+  "--optimizer",
+  "adamw",
+  "--learning-rate",
+  "0.001",
+  "--weight-decay",
+  "0.001",
+  "--causal-refit-rounds",
+  "2",
+  "--model-seed",
+  "23"
+ ]
+}