Add files using upload-large-folder tool
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- logs/eval_IdkDPO_lr5e-05_beta0.5_alpha2_epoch5.log +78 -0
- logs/eval_IdkDPO_lr5e-05_beta0.5_alpha5_epoch10.log +76 -0
- logs/eval_IdkDPO_lr5e-05_beta0.5_alpha5_epoch5.log +78 -0
- logs/eval_IdkNLL_lr1e-05_alpha10_epoch10.log +76 -0
- logs/eval_IdkNLL_lr1e-05_alpha10_epoch5.log +78 -0
- logs/eval_IdkNLL_lr1e-05_alpha1_epoch10.log +78 -0
- logs/eval_IdkNLL_lr1e-05_alpha1_epoch5.log +78 -0
- logs/eval_IdkNLL_lr1e-05_alpha2_epoch10.log +78 -0
- logs/eval_IdkNLL_lr1e-05_alpha2_epoch5.log +78 -0
- logs/eval_IdkNLL_lr1e-05_alpha5_epoch10.log +78 -0
- logs/eval_IdkNLL_lr1e-05_alpha5_epoch5.log +78 -0
- logs/eval_IdkNLL_lr2e-05_alpha10_epoch10.log +76 -0
- logs/eval_IdkNLL_lr2e-05_alpha10_epoch5.log +78 -0
- logs/eval_IdkNLL_lr2e-05_alpha1_epoch10.log +78 -0
- logs/eval_IdkNLL_lr2e-05_alpha1_epoch5.log +78 -0
- logs/eval_IdkNLL_lr2e-05_alpha2_epoch10.log +78 -0
- logs/eval_IdkNLL_lr2e-05_alpha2_epoch5.log +78 -0
- logs/eval_IdkNLL_lr2e-05_alpha5_epoch10.log +78 -0
- logs/eval_IdkNLL_lr2e-05_alpha5_epoch5.log +78 -0
- logs/eval_IdkNLL_lr3e-05_alpha10_epoch10.log +76 -0
- logs/eval_IdkNLL_lr3e-05_alpha10_epoch5.log +78 -0
- logs/eval_IdkNLL_lr3e-05_alpha1_epoch10.log +78 -0
- logs/eval_IdkNLL_lr3e-05_alpha1_epoch5.log +78 -0
- logs/eval_IdkNLL_lr3e-05_alpha2_epoch10.log +78 -0
- logs/eval_IdkNLL_lr3e-05_alpha2_epoch5.log +78 -0
- logs/eval_IdkNLL_lr3e-05_alpha5_epoch10.log +78 -0
- logs/eval_IdkNLL_lr3e-05_alpha5_epoch5.log +78 -0
- logs/eval_IdkNLL_lr4e-05_alpha10_epoch10.log +76 -0
- logs/eval_IdkNLL_lr4e-05_alpha10_epoch5.log +78 -0
- logs/eval_IdkNLL_lr4e-05_alpha1_epoch10.log +78 -0
- logs/eval_IdkNLL_lr4e-05_alpha1_epoch5.log +78 -0
- logs/eval_IdkNLL_lr4e-05_alpha2_epoch10.log +76 -0
- logs/eval_IdkNLL_lr4e-05_alpha2_epoch5.log +78 -0
- logs/eval_IdkNLL_lr4e-05_alpha5_epoch10.log +78 -0
- logs/eval_IdkNLL_lr4e-05_alpha5_epoch5.log +78 -0
- logs/eval_IdkNLL_lr5e-05_alpha10_epoch10.log +78 -0
- logs/eval_IdkNLL_lr5e-05_alpha10_epoch5.log +78 -0
- logs/eval_IdkNLL_lr5e-05_alpha1_epoch10.log +78 -0
- logs/eval_IdkNLL_lr5e-05_alpha1_epoch5.log +78 -0
- logs/eval_IdkNLL_lr5e-05_alpha2_epoch10.log +76 -0
- logs/eval_IdkNLL_lr5e-05_alpha2_epoch5.log +78 -0
- logs/eval_IdkNLL_lr5e-05_alpha5_epoch10.log +78 -0
- logs/eval_IdkNLL_lr5e-05_alpha5_epoch5.log +78 -0
- logs/eval_NPO_lr1e-05_beta0.05_alpha1_epoch10.log +124 -0
- logs/eval_NPO_lr1e-05_beta0.05_alpha1_epoch5.log +78 -0
- logs/eval_NPO_lr1e-05_beta0.05_alpha2_epoch10.log +78 -0
- logs/eval_NPO_lr1e-05_beta0.05_alpha2_epoch5.log +78 -0
- logs/eval_NPO_lr1e-05_beta0.05_alpha5_epoch10.log +78 -0
- logs/eval_NPO_lr1e-05_beta0.05_alpha5_epoch5.log +78 -0
- logs/eval_NPO_lr1e-05_beta0.1_alpha1_epoch10.log +78 -0
logs/eval_IdkDPO_lr5e-05_beta0.5_alpha2_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 14.21it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.90it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.33it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.77it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.98it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.70it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.49it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.32it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.70it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.73it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.82it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.92it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:09:58,388[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkDPO_lr5e-05_beta0.5_alpha2_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:09:58,389[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:09:58,389[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkDPO_lr5e-05_beta0.5_alpha2_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:09:58,389[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkDPO_lr5e-05_beta0.5_alpha2_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:09:59,059[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:10:02,628[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:10:17,050[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:10:17,051[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5424492803784302[0m
|
| 15 |
+
[[36m2026-07-22 17:10:17,055[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:10:17,055[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:10:17,062[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:10:17,062[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 1.0229984068189572e-13[0m
|
| 19 |
+
[[36m2026-07-22 17:10:17,410[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:10:20,284[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.62306640625[0m
|
| 23 |
+
[[36m2026-07-22 17:10:20,531[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:10:25,150[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.4722967646500736[0m
|
| 27 |
+
[[36m2026-07-22 17:10:25,393[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:10:28,408[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:10:33,302[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:10:36,510[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:10:50,506[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:10:50,839[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:10:51,698[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:10:53,311[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:10:53,617[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:10:54,449[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:10:54,449[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:10:54,449[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:10:54,699[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:10:55,455[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:10:57,007[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:10:57,307[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:10:58,199[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:10:58,199[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:10:58,199[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:10:58,199[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:10:58,199[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5444555848679563[0m
|
| 68 |
+
[[36m2026-07-22 17:10:58,733[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:10:58,740[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 14.21it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.90it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.33it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.77it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.98it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.70it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.49it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.32it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.70it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.73it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.82it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.92it/s]
|
| 86 |
+
[[36m2026-07-22 17:11:00,568[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:11:00,575[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:11:00,575[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -97.77995989588713[0m
|
| 89 |
+
[[36m2026-07-22 17:11:00,839[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:11:01,943[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.24642572663532428[0m
|
logs/eval_IdkDPO_lr5e-05_beta0.5_alpha5_epoch10.log
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.05it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 11.27it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 11.55it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 11.70it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 11.95it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:01<00:00, 11.69it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 11.51it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 11.37it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 11.69it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 11.74it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 11.74it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:01<00:00, 11.85it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[[36m2026-07-21 07:49:18,817[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkDPO_lr5e-05_beta0.5_alpha5_epoch10[0m
|
| 2 |
+
[[36m2026-07-21 07:49:18,819[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 3 |
+
[[36m2026-07-21 07:49:18,819[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkDPO_lr5e-05_beta0.5_alpha5_epoch10/TOFU_EVAL.json[0m
|
| 4 |
+
[[36m2026-07-21 07:49:18,819[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkDPO_lr5e-05_beta0.5_alpha5_epoch10/TOFU_SUMMARY.json[0m
|
| 5 |
+
[[36m2026-07-21 07:49:19,632[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 6 |
+
|
| 7 |
+
Evaluated 400 examples
|
| 8 |
+
[[36m2026-07-21 07:49:23,178[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 9 |
+
|
| 10 |
+
Evaluated 400 examples
|
| 11 |
+
[[36m2026-07-21 07:49:36,315[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 12 |
+
[[36m2026-07-21 07:49:36,315[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.521592377972619[0m
|
| 13 |
+
[[36m2026-07-21 07:49:36,323[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 14 |
+
[[36m2026-07-21 07:49:36,325[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 15 |
+
[[36m2026-07-21 07:49:36,340[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 16 |
+
[[36m2026-07-21 07:49:36,341[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 2.776495661774257e-16[0m
|
| 17 |
+
[[36m2026-07-21 07:49:36,854[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 18 |
+
|
| 19 |
+
Evaluated 400 examples
|
| 20 |
+
[[36m2026-07-21 07:49:39,529[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.6661669921875[0m
|
| 21 |
+
[[36m2026-07-21 07:49:39,827[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 22 |
+
|
| 23 |
+
Evaluated 400 examples
|
| 24 |
+
[[36m2026-07-21 07:49:48,649[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.5204367519573109[0m
|
| 25 |
+
[[36m2026-07-21 07:49:49,021[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 26 |
+
|
| 27 |
+
Evaluated 400 examples
|
| 28 |
+
[[36m2026-07-21 07:49:51,857[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 29 |
+
|
| 30 |
+
Evaluated 400 examples
|
| 31 |
+
[[36m2026-07-21 07:50:00,188[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 32 |
+
|
| 33 |
+
Evaluated 400 examples
|
| 34 |
+
[[36m2026-07-21 07:50:03,087[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 35 |
+
|
| 36 |
+
Evaluated 400 examples
|
| 37 |
+
[[36m2026-07-21 07:50:15,929[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 38 |
+
[[36m2026-07-21 07:50:16,284[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 39 |
+
|
| 40 |
+
Evaluated 100 examples
|
| 41 |
+
[[36m2026-07-21 07:50:17,038[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 42 |
+
|
| 43 |
+
Evaluated 100 examples
|
| 44 |
+
[[36m2026-07-21 07:50:18,537[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 45 |
+
[[36m2026-07-21 07:50:18,811[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 46 |
+
|
| 47 |
+
Evaluated 100 examples
|
| 48 |
+
[[36m2026-07-21 07:50:20,154[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 49 |
+
[[36m2026-07-21 07:50:20,154[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 50 |
+
[[36m2026-07-21 07:50:20,154[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 51 |
+
[[36m2026-07-21 07:50:20,417[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 52 |
+
|
| 53 |
+
Evaluated 117 examples
|
| 54 |
+
[[36m2026-07-21 07:50:21,130[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 55 |
+
|
| 56 |
+
Evaluated 117 examples
|
| 57 |
+
[[36m2026-07-21 07:50:22,608[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 58 |
+
[[36m2026-07-21 07:50:22,884[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 59 |
+
|
| 60 |
+
Evaluated 117 examples
|
| 61 |
+
[[36m2026-07-21 07:50:24,847[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 62 |
+
[[36m2026-07-21 07:50:24,847[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 63 |
+
[[36m2026-07-21 07:50:24,847[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 64 |
+
[[36m2026-07-21 07:50:24,848[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 65 |
+
[[36m2026-07-21 07:50:24,849[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5546318636120038[0m
|
| 66 |
+
[[36m2026-07-21 07:50:25,425[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 67 |
+
[[36m2026-07-21 07:50:25,441[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 68 |
+
|
| 69 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 70 |
15%|█▌ | 2/13 [00:00<00:00, 13.05it/s]
|
| 71 |
31%|███ | 4/13 [00:00<00:00, 11.27it/s]
|
| 72 |
46%|████▌ | 6/13 [00:00<00:00, 11.55it/s]
|
| 73 |
62%|██████▏ | 8/13 [00:00<00:00, 11.70it/s]
|
| 74 |
77%|███████▋ | 10/13 [00:00<00:00, 11.95it/s]
|
| 75 |
92%|█████████▏| 12/13 [00:01<00:00, 11.69it/s]
|
| 76 |
+
|
| 77 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 78 |
15%|█▌ | 2/13 [00:00<00:00, 11.51it/s]
|
| 79 |
31%|███ | 4/13 [00:00<00:00, 11.37it/s]
|
| 80 |
46%|████▌ | 6/13 [00:00<00:00, 11.69it/s]
|
| 81 |
62%|██████▏ | 8/13 [00:00<00:00, 11.74it/s]
|
| 82 |
77%|███████▋ | 10/13 [00:00<00:00, 11.74it/s]
|
| 83 |
92%|█████████▏| 12/13 [00:01<00:00, 11.85it/s]
|
| 84 |
+
[[36m2026-07-21 07:50:27,581[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 85 |
+
[[36m2026-07-21 07:50:27,596[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 86 |
+
[[36m2026-07-21 07:50:27,596[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.11504118594642[0m
|
| 87 |
+
[[36m2026-07-21 07:50:27,882[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 88 |
+
|
| 89 |
+
Evaluated 400 examples
|
| 90 |
+
[[36m2026-07-21 07:50:29,123[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.2952828489835585[0m
|
logs/eval_IdkDPO_lr5e-05_beta0.5_alpha5_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 14.04it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.63it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.19it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.50it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.83it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.54it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.20it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.02it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.45it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.44it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.52it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.64it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:00:21,742[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkDPO_lr5e-05_beta0.5_alpha5_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:00:21,743[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:00:21,743[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkDPO_lr5e-05_beta0.5_alpha5_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:00:21,743[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkDPO_lr5e-05_beta0.5_alpha5_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:00:22,484[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:00:25,906[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:00:40,312[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:00:40,313[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5391518666563003[0m
|
| 15 |
+
[[36m2026-07-22 17:00:40,317[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:00:40,317[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:00:40,324[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:00:40,324[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 3.1508505165833036e-15[0m
|
| 19 |
+
[[36m2026-07-22 17:00:40,815[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:00:43,681[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.6280224609375[0m
|
| 23 |
+
[[36m2026-07-22 17:00:43,924[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:00:48,781[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.4796897890575585[0m
|
| 27 |
+
[[36m2026-07-22 17:00:49,027[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:00:52,076[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:00:56,617[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:00:59,670[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:01:13,744[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:01:14,091[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:01:14,870[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:01:16,462[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:01:16,763[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:01:17,525[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:01:17,525[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:01:17,525[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:01:17,770[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:01:18,498[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:01:20,076[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:01:20,382[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:01:21,376[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:01:21,377[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:01:21,377[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:01:21,377[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:01:21,377[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.543886049725611[0m
|
| 68 |
+
[[36m2026-07-22 17:01:21,890[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:01:21,899[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 14.04it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.63it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.19it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.50it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.83it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.54it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.20it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.02it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.45it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.44it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.52it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.64it/s]
|
| 86 |
+
[[36m2026-07-22 17:01:23,759[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:01:23,766[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:01:23,766[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -97.90852582893103[0m
|
| 89 |
+
[[36m2026-07-22 17:01:24,030[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:01:25,122[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.250188864524075[0m
|
logs/eval_IdkNLL_lr1e-05_alpha10_epoch10.log
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.54it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 11.58it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 11.85it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 12.01it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 12.24it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 12.00it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 11.81it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 11.66it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 11.95it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 11.98it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 11.95it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:01<00:00, 12.08it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[[36m2026-07-21 07:49:13,805[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr1e-05_alpha10_epoch10[0m
|
| 2 |
+
[[36m2026-07-21 07:49:13,806[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 3 |
+
[[36m2026-07-21 07:49:13,806[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr1e-05_alpha10_epoch10/TOFU_EVAL.json[0m
|
| 4 |
+
[[36m2026-07-21 07:49:13,806[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr1e-05_alpha10_epoch10/TOFU_SUMMARY.json[0m
|
| 5 |
+
[[36m2026-07-21 07:49:14,461[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 6 |
+
|
| 7 |
+
Evaluated 400 examples
|
| 8 |
+
[[36m2026-07-21 07:49:17,978[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 9 |
+
|
| 10 |
+
Evaluated 400 examples
|
| 11 |
+
[[36m2026-07-21 07:49:31,092[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 12 |
+
[[36m2026-07-21 07:49:31,093[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.47608075658606114[0m
|
| 13 |
+
[[36m2026-07-21 07:49:31,101[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 14 |
+
[[36m2026-07-21 07:49:31,102[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 15 |
+
[[36m2026-07-21 07:49:31,118[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 16 |
+
[[36m2026-07-21 07:49:31,119[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 6.916893189690245e-21[0m
|
| 17 |
+
[[36m2026-07-21 07:49:31,575[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 18 |
+
|
| 19 |
+
Evaluated 400 examples
|
| 20 |
+
[[36m2026-07-21 07:49:34,241[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.842978515625[0m
|
| 21 |
+
[[36m2026-07-21 07:49:34,509[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 22 |
+
|
| 23 |
+
Evaluated 400 examples
|
| 24 |
+
[[36m2026-07-21 07:49:42,477[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.43768437843662367[0m
|
| 25 |
+
[[36m2026-07-21 07:49:42,835[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 26 |
+
|
| 27 |
+
Evaluated 400 examples
|
| 28 |
+
[[36m2026-07-21 07:49:45,737[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 29 |
+
|
| 30 |
+
Evaluated 400 examples
|
| 31 |
+
[[36m2026-07-21 07:49:53,259[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 32 |
+
|
| 33 |
+
Evaluated 400 examples
|
| 34 |
+
[[36m2026-07-21 07:49:56,166[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 35 |
+
|
| 36 |
+
Evaluated 400 examples
|
| 37 |
+
[[36m2026-07-21 07:50:08,980[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 38 |
+
[[36m2026-07-21 07:50:09,342[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 39 |
+
|
| 40 |
+
Evaluated 100 examples
|
| 41 |
+
[[36m2026-07-21 07:50:10,143[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 42 |
+
|
| 43 |
+
Evaluated 100 examples
|
| 44 |
+
[[36m2026-07-21 07:50:11,640[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 45 |
+
[[36m2026-07-21 07:50:11,903[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 46 |
+
|
| 47 |
+
Evaluated 100 examples
|
| 48 |
+
[[36m2026-07-21 07:50:13,266[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 49 |
+
[[36m2026-07-21 07:50:13,266[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 50 |
+
[[36m2026-07-21 07:50:13,266[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 51 |
+
[[36m2026-07-21 07:50:13,517[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 52 |
+
|
| 53 |
+
Evaluated 117 examples
|
| 54 |
+
[[36m2026-07-21 07:50:14,241[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 55 |
+
|
| 56 |
+
Evaluated 117 examples
|
| 57 |
+
[[36m2026-07-21 07:50:15,711[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 58 |
+
[[36m2026-07-21 07:50:15,977[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 59 |
+
|
| 60 |
+
Evaluated 117 examples
|
| 61 |
+
[[36m2026-07-21 07:50:17,650[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 62 |
+
[[36m2026-07-21 07:50:17,651[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 63 |
+
[[36m2026-07-21 07:50:17,651[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 64 |
+
[[36m2026-07-21 07:50:17,651[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 65 |
+
[[36m2026-07-21 07:50:17,652[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5770912163051024[0m
|
| 66 |
+
[[36m2026-07-21 07:50:18,238[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 67 |
+
[[36m2026-07-21 07:50:18,254[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 68 |
+
|
| 69 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 70 |
15%|█▌ | 2/13 [00:00<00:00, 13.54it/s]
|
| 71 |
31%|███ | 4/13 [00:00<00:00, 11.58it/s]
|
| 72 |
46%|████▌ | 6/13 [00:00<00:00, 11.85it/s]
|
| 73 |
62%|██████▏ | 8/13 [00:00<00:00, 12.01it/s]
|
| 74 |
77%|███████▋ | 10/13 [00:00<00:00, 12.24it/s]
|
| 75 |
92%|█████████▏| 12/13 [00:00<00:00, 12.00it/s]
|
| 76 |
+
|
| 77 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 78 |
15%|█▌ | 2/13 [00:00<00:00, 11.81it/s]
|
| 79 |
31%|███ | 4/13 [00:00<00:00, 11.66it/s]
|
| 80 |
46%|████▌ | 6/13 [00:00<00:00, 11.95it/s]
|
| 81 |
62%|██████▏ | 8/13 [00:00<00:00, 11.98it/s]
|
| 82 |
77%|███████▋ | 10/13 [00:00<00:00, 11.95it/s]
|
| 83 |
92%|█████████▏| 12/13 [00:01<00:00, 12.08it/s]
|
| 84 |
+
[[36m2026-07-21 07:50:20,342[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 85 |
+
[[36m2026-07-21 07:50:20,357[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 86 |
+
[[36m2026-07-21 07:50:20,357[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.44929236509499[0m
|
| 87 |
+
[[36m2026-07-21 07:50:20,639[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 88 |
+
|
| 89 |
+
Evaluated 400 examples
|
| 90 |
+
[[36m2026-07-21 07:50:21,772[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.6415381995913018[0m
|
logs/eval_IdkNLL_lr1e-05_alpha10_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.75it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.47it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.81it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.01it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.50it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.26it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.41it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.07it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.60it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.64it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.45it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.73it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:11:33,008[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr1e-05_alpha10_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:11:33,009[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:11:33,009[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr1e-05_alpha10_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:11:33,009[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr1e-05_alpha10_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:11:34,408[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:11:37,919[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:11:52,284[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:11:52,284[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4756530184094565[0m
|
| 15 |
+
[[36m2026-07-22 17:11:52,289[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:11:52,289[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:11:52,296[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:11:52,297[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 6.916893189690245e-21[0m
|
| 19 |
+
[[36m2026-07-22 17:11:52,663[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:11:55,559[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.8422119140625[0m
|
| 23 |
+
[[36m2026-07-22 17:11:56,043[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:12:00,136[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.44177991317558574[0m
|
| 27 |
+
[[36m2026-07-22 17:12:00,384[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:12:03,424[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:12:07,380[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:12:10,466[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:12:24,469[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:12:24,956[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:12:25,769[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:12:27,570[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:12:27,809[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:12:28,528[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:12:28,528[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:12:28,528[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:12:28,766[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:12:29,520[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:12:31,071[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:12:31,316[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:12:32,285[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:12:32,285[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:12:32,285[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:12:32,285[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:12:32,286[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5760213222711852[0m
|
| 68 |
+
[[36m2026-07-22 17:12:32,780[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:12:32,787[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.75it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.47it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.81it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.01it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.50it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.26it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.41it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.07it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.60it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.64it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.45it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.73it/s]
|
| 86 |
+
[[36m2026-07-22 17:12:34,661[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:12:34,669[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:12:34,669[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.47561294981261[0m
|
| 89 |
+
[[36m2026-07-22 17:12:35,386[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:12:36,386[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.634425163480305[0m
|
logs/eval_IdkNLL_lr1e-05_alpha1_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
8%|▊ | 1/13 [00:00<00:01, 9.72it/s]
|
| 2 |
15%|█▌ | 2/13 [00:00<00:01, 8.29it/s]
|
| 3 |
23%|██▎ | 3/13 [00:00<00:01, 7.12it/s]
|
| 4 |
31%|███ | 4/13 [00:00<00:01, 6.95it/s]
|
| 5 |
38%|███▊ | 5/13 [00:00<00:01, 7.52it/s]
|
| 6 |
54%|█████▍ | 7/13 [00:00<00:00, 9.10it/s]
|
| 7 |
69%|██████▉ | 9/13 [00:00<00:00, 10.40it/s]
|
| 8 |
85%|████████▍ | 11/13 [00:01<00:00, 5.98it/s]
|
| 9 |
92%|█████████▏| 12/13 [00:01<00:00, 5.27it/s]
|
|
|
|
| 10 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 11 |
8%|▊ | 1/13 [00:00<00:01, 7.76it/s]
|
| 12 |
15%|█▌ | 2/13 [00:00<00:02, 4.63it/s]
|
| 13 |
23%|██▎ | 3/13 [00:00<00:02, 4.05it/s]
|
| 14 |
31%|███ | 4/13 [00:00<00:02, 3.85it/s]
|
| 15 |
38%|███▊ | 5/13 [00:01<00:02, 3.73it/s]
|
| 16 |
46%|████▌ | 6/13 [00:01<00:01, 3.75it/s]
|
| 17 |
54%|█████▍ | 7/13 [00:01<00:01, 3.73it/s]
|
| 18 |
62%|██████▏ | 8/13 [00:02<00:01, 3.59it/s]
|
| 19 |
69%|██████▉ | 9/13 [00:02<00:01, 3.61it/s]
|
| 20 |
77%|███████▋ | 10/13 [00:02<00:00, 3.80it/s]
|
| 21 |
92%|█████████▏| 12/13 [00:02<00:00, 5.76it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-20 14:43:52,451[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr1e-05_alpha1_epoch10[0m
|
| 4 |
+
[[36m2026-07-20 14:43:52,453[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-20 14:43:52,453[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr1e-05_alpha1_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-20 14:43:52,453[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr1e-05_alpha1_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-20 14:43:53,109[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-20 14:43:59,268[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-20 14:44:22,620[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-20 14:44:22,621[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.48020106904725524[0m
|
| 15 |
+
[[36m2026-07-20 14:44:22,629[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-20 14:44:22,630[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-20 14:44:22,646[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-20 14:44:22,647[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 4.353260441808186e-19[0m
|
| 19 |
+
[[36m2026-07-20 14:44:22,997[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-20 14:44:27,365[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.79087890625[0m
|
| 23 |
+
[[36m2026-07-20 14:44:27,677[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-20 14:44:32,138[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.025900074852939204[0m
|
| 27 |
+
[[36m2026-07-20 14:44:32,477[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-20 14:44:35,960[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-20 14:44:46,560[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-20 14:44:51,527[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-20 14:45:16,246[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-20 14:45:16,621[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-20 14:45:17,988[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-20 14:45:20,296[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-20 14:45:20,567[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-20 14:45:22,167[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-20 14:45:22,167[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-20 14:45:22,167[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-20 14:45:22,568[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-20 14:45:23,473[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-20 14:45:25,568[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-20 14:45:25,856[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-20 14:45:27,828[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-20 14:45:27,828[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-20 14:45:27,828[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-20 14:45:27,829[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-20 14:45:27,830[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.4584928946618976[0m
|
| 68 |
+
[[36m2026-07-20 14:45:28,384[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-20 14:45:28,399[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
8%|▊ | 1/13 [00:00<00:01, 9.72it/s]
|
| 73 |
15%|█▌ | 2/13 [00:00<00:01, 8.29it/s]
|
| 74 |
23%|██▎ | 3/13 [00:00<00:01, 7.12it/s]
|
| 75 |
31%|███ | 4/13 [00:00<00:01, 6.95it/s]
|
| 76 |
38%|███▊ | 5/13 [00:00<00:01, 7.52it/s]
|
| 77 |
54%|█████▍ | 7/13 [00:00<00:00, 9.10it/s]
|
| 78 |
69%|██████▉ | 9/13 [00:00<00:00, 10.40it/s]
|
| 79 |
85%|████████▍ | 11/13 [00:01<00:00, 5.98it/s]
|
| 80 |
92%|█████████▏| 12/13 [00:01<00:00, 5.27it/s]
|
| 81 |
+
|
| 82 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 83 |
8%|▊ | 1/13 [00:00<00:01, 7.76it/s]
|
| 84 |
15%|█▌ | 2/13 [00:00<00:02, 4.63it/s]
|
| 85 |
23%|██▎ | 3/13 [00:00<00:02, 4.05it/s]
|
| 86 |
31%|███ | 4/13 [00:00<00:02, 3.85it/s]
|
| 87 |
38%|███▊ | 5/13 [00:01<00:02, 3.73it/s]
|
| 88 |
46%|████▌ | 6/13 [00:01<00:01, 3.75it/s]
|
| 89 |
54%|█████▍ | 7/13 [00:01<00:01, 3.73it/s]
|
| 90 |
62%|██████▏ | 8/13 [00:02<00:01, 3.59it/s]
|
| 91 |
69%|██████▉ | 9/13 [00:02<00:01, 3.61it/s]
|
| 92 |
77%|███████▋ | 10/13 [00:02<00:00, 3.80it/s]
|
| 93 |
92%|█████████▏| 12/13 [00:02<00:00, 5.76it/s]
|
| 94 |
+
[[36m2026-07-20 14:45:33,173[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 95 |
+
[[36m2026-07-20 14:45:33,187[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 96 |
+
[[36m2026-07-20 14:45:33,188[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.44929236509498[0m
|
| 97 |
+
[[36m2026-07-20 14:45:33,474[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 98 |
+
|
| 99 |
+
Evaluated 400 examples
|
| 100 |
+
[[36m2026-07-20 14:45:36,179[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.6068313205387513[0m
|
logs/eval_IdkNLL_lr1e-05_alpha1_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.47it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.14it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.56it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 12.82it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.20it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 12.82it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 12.65it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.45it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 12.78it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 12.71it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 12.86it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 12.87it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:08:48,772[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr1e-05_alpha1_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:08:48,773[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:08:48,773[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr1e-05_alpha1_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:08:48,773[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr1e-05_alpha1_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:08:49,420[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:08:52,865[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:09:07,215[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:09:07,215[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4817956462414547[0m
|
| 15 |
+
[[36m2026-07-22 17:09:07,219[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:09:07,219[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:09:07,226[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:09:07,227[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 1.11924339764853e-19[0m
|
| 19 |
+
[[36m2026-07-22 17:09:07,644[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:09:10,491[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.792119140625[0m
|
| 23 |
+
[[36m2026-07-22 17:09:10,737[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:09:12,724[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.02485689709593503[0m
|
| 27 |
+
[[36m2026-07-22 17:09:12,975[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:09:16,017[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:09:19,932[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:09:23,019[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:09:37,012[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:09:37,443[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:09:38,322[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:09:40,095[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:09:40,428[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:09:41,097[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:09:41,097[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:09:41,097[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:09:41,431[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:09:42,171[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:09:43,721[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:09:44,071[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:09:45,016[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:09:45,016[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:09:45,016[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:09:45,017[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:09:45,017[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.44971118166631735[0m
|
| 68 |
+
[[36m2026-07-22 17:09:45,560[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:09:45,568[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.47it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.14it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.56it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 12.82it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.20it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 12.82it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 12.65it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.45it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 12.78it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 12.71it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 12.86it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 12.87it/s]
|
| 86 |
+
[[36m2026-07-22 17:09:47,533[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:09:47,541[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:09:47,541[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.48877324217145[0m
|
| 89 |
+
[[36m2026-07-22 17:09:48,020[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:09:49,095[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.5879578746962085[0m
|
logs/eval_IdkNLL_lr1e-05_alpha2_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 12.73it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 11.46it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.17it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 12.62it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.08it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 12.57it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 12.83it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 11.95it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 12.05it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 11.25it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 11.35it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:01<00:00, 11.71it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:02:01,512[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr1e-05_alpha2_epoch10[0m
|
| 4 |
+
[[36m2026-07-22 17:02:01,513[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:02:01,513[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr1e-05_alpha2_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:02:01,513[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr1e-05_alpha2_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:02:02,162[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:02:05,615[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:02:20,008[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:02:20,009[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4754792238777897[0m
|
| 15 |
+
[[36m2026-07-22 17:02:20,013[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:02:20,013[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:02:20,020[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:02:20,020[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 2.8138576976529647e-20[0m
|
| 19 |
+
[[36m2026-07-22 17:02:20,582[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:02:23,442[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.80720703125[0m
|
| 23 |
+
[[36m2026-07-22 17:02:23,684[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:02:26,731[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.07287559625204562[0m
|
| 27 |
+
[[36m2026-07-22 17:02:26,997[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:02:30,060[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:02:34,127[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:02:37,275[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:02:51,287[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:02:51,636[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:02:52,412[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:02:53,988[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:02:54,233[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:02:54,968[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:02:54,968[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:02:54,968[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:02:55,226[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:02:55,945[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:02:57,500[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:02:57,755[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:02:58,712[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:02:58,712[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:02:58,712[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:02:58,712[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:02:58,713[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.542546453505407[0m
|
| 68 |
+
[[36m2026-07-22 17:02:59,304[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:02:59,316[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 12.73it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 11.46it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.17it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 12.62it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.08it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 12.57it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 12.83it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 11.95it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 12.05it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 11.25it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 11.35it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:01<00:00, 11.71it/s]
|
| 86 |
+
[[36m2026-07-22 17:03:01,385[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:03:01,392[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:03:01,392[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.36628129021624[0m
|
| 89 |
+
[[36m2026-07-22 17:03:01,699[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:03:02,724[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.6247114783743656[0m
|
logs/eval_IdkNLL_lr1e-05_alpha2_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 14.21it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.95it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.58it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.95it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 14.33it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 14.04it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.73it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.47it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.92it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.94it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 14.00it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 14.08it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:13:05,961[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr1e-05_alpha2_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:13:05,962[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:13:05,962[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr1e-05_alpha2_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:13:05,962[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr1e-05_alpha2_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:13:11,704[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:13:15,136[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:13:29,499[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:13:29,499[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4763998624487176[0m
|
| 15 |
+
[[36m2026-07-22 17:13:29,503[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:13:29,504[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:13:29,511[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:13:29,511[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 1.11924339764853e-19[0m
|
| 19 |
+
[[36m2026-07-22 17:13:29,859[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:13:32,736[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.8068798828125[0m
|
| 23 |
+
[[36m2026-07-22 17:13:32,981[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:13:35,821[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.06959552198172891[0m
|
| 27 |
+
[[36m2026-07-22 17:13:36,101[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:13:39,095[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:13:43,280[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:13:46,374[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:14:00,402[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:14:00,786[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:14:01,639[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:14:03,216[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:14:03,484[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:14:04,296[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:14:04,296[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:14:04,296[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:14:04,527[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:14:05,251[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:14:06,813[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:14:07,131[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:14:08,039[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:14:08,039[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:14:08,039[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:14:08,039[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:14:08,039[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.529183527132918[0m
|
| 68 |
+
[[36m2026-07-22 17:14:08,559[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:14:08,566[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 14.21it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.95it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.58it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.95it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 14.33it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 14.04it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.73it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.47it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.92it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.94it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 14.00it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 14.08it/s]
|
| 86 |
+
[[36m2026-07-22 17:14:10,365[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:14:10,371[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:14:10,372[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.41588546910718[0m
|
| 89 |
+
[[36m2026-07-22 17:14:10,623[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:14:11,605[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.6153080531513968[0m
|
logs/eval_IdkNLL_lr1e-05_alpha5_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.90it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.49it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.06it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.35it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.62it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.28it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 12.89it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.76it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.06it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.14it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.13it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.28it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:10:14,173[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr1e-05_alpha5_epoch10[0m
|
| 4 |
+
[[36m2026-07-22 17:10:14,174[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:10:14,174[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr1e-05_alpha5_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:10:14,174[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr1e-05_alpha5_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:10:14,977[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:10:18,373[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:10:32,731[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:10:32,732[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.47389271319887827[0m
|
| 15 |
+
[[36m2026-07-22 17:10:32,736[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:10:32,736[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:10:32,743[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:10:32,743[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 1.3990395281788573e-20[0m
|
| 19 |
+
[[36m2026-07-22 17:10:33,189[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:10:36,053[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.82615234375[0m
|
| 23 |
+
[[36m2026-07-22 17:10:36,306[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:10:40,048[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.2513266852163511[0m
|
| 27 |
+
[[36m2026-07-22 17:10:40,360[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:10:43,393[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:10:47,817[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:10:50,875[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:11:04,894[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:11:05,261[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:11:06,027[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:11:07,602[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:11:07,964[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:11:08,782[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:11:08,782[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:11:08,782[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:11:09,060[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:11:10,126[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:11:11,697[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:11:11,952[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:11:12,985[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:11:12,985[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:11:12,986[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:11:12,986[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:11:12,986[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.570631823217008[0m
|
| 68 |
+
[[36m2026-07-22 17:11:13,513[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:11:13,520[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.90it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.49it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.06it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.35it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.62it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.28it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 12.89it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.76it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.06it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.14it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.13it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.28it/s]
|
| 86 |
+
[[36m2026-07-22 17:11:15,419[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:11:15,425[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:11:15,425[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.4158854691072[0m
|
| 89 |
+
[[36m2026-07-22 17:11:15,682[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:11:16,709[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.6355469275996612[0m
|
logs/eval_IdkNLL_lr1e-05_alpha5_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.22it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.04it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.62it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.02it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.46it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.31it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 12.91it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 11.36it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 11.52it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 11.80it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 12.14it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 12.48it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:03:40,894[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr1e-05_alpha5_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:03:40,895[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:03:40,895[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr1e-05_alpha5_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:03:40,895[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr1e-05_alpha5_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:03:41,626[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:03:45,117[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:03:59,518[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:03:59,518[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.47482963606711215[0m
|
| 15 |
+
[[36m2026-07-22 17:03:59,523[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:03:59,523[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:03:59,530[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:03:59,531[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 2.8138576976529647e-20[0m
|
| 19 |
+
[[36m2026-07-22 17:03:59,935[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:04:02,791[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.8255029296875[0m
|
| 23 |
+
[[36m2026-07-22 17:04:03,026[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:04:06,873[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.23682434548046824[0m
|
| 27 |
+
[[36m2026-07-22 17:04:07,156[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:04:10,220[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:04:14,499[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:04:17,653[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:04:31,693[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:04:32,039[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:04:32,825[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:04:34,395[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:04:34,636[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:04:35,378[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:04:35,378[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:04:35,378[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:04:35,630[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:04:36,348[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:04:37,911[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:04:38,159[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:04:39,162[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:04:39,163[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:04:39,163[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:04:39,163[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:04:39,163[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.567320633002962[0m
|
| 68 |
+
[[36m2026-07-22 17:04:39,647[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:04:39,654[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.22it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.04it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.62it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.02it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.46it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.31it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 12.91it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 11.36it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 11.52it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 11.80it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 12.14it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 12.48it/s]
|
| 86 |
+
[[36m2026-07-22 17:04:41,641[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:04:41,648[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:04:41,648[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.40474983711124[0m
|
| 89 |
+
[[36m2026-07-22 17:04:41,897[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:04:43,051[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.6241283959217184[0m
|
logs/eval_IdkNLL_lr2e-05_alpha10_epoch10.log
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 12.98it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 11.02it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 11.29it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 11.46it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 11.72it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:01<00:00, 11.47it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 11.25it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 11.11it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 11.41it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 11.44it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 11.45it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:01<00:00, 11.57it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[[36m2026-07-21 07:49:18,023[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr2e-05_alpha10_epoch10[0m
|
| 2 |
+
[[36m2026-07-21 07:49:18,024[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 3 |
+
[[36m2026-07-21 07:49:18,025[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr2e-05_alpha10_epoch10/TOFU_EVAL.json[0m
|
| 4 |
+
[[36m2026-07-21 07:49:18,025[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr2e-05_alpha10_epoch10/TOFU_SUMMARY.json[0m
|
| 5 |
+
[[36m2026-07-21 07:49:18,749[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 6 |
+
|
| 7 |
+
Evaluated 400 examples
|
| 8 |
+
[[36m2026-07-21 07:49:22,356[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 9 |
+
|
| 10 |
+
Evaluated 400 examples
|
| 11 |
+
[[36m2026-07-21 07:49:35,564[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 12 |
+
[[36m2026-07-21 07:49:35,565[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4807089697996208[0m
|
| 13 |
+
[[36m2026-07-21 07:49:35,573[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 14 |
+
[[36m2026-07-21 07:49:35,574[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 15 |
+
[[36m2026-07-21 07:49:35,589[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 16 |
+
[[36m2026-07-21 07:49:35,591[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 4.353260441808186e-19[0m
|
| 17 |
+
[[36m2026-07-21 07:49:35,981[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 18 |
+
|
| 19 |
+
Evaluated 400 examples
|
| 20 |
+
[[36m2026-07-21 07:49:38,665[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.797080078125[0m
|
| 21 |
+
[[36m2026-07-21 07:49:38,948[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 22 |
+
|
| 23 |
+
Evaluated 400 examples
|
| 24 |
+
[[36m2026-07-21 07:49:45,262[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.1882429679505736[0m
|
| 25 |
+
[[36m2026-07-21 07:49:45,686[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 26 |
+
|
| 27 |
+
Evaluated 400 examples
|
| 28 |
+
[[36m2026-07-21 07:49:48,569[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 29 |
+
|
| 30 |
+
Evaluated 400 examples
|
| 31 |
+
[[36m2026-07-21 07:49:56,565[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 32 |
+
|
| 33 |
+
Evaluated 400 examples
|
| 34 |
+
[[36m2026-07-21 07:49:59,497[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 35 |
+
|
| 36 |
+
Evaluated 400 examples
|
| 37 |
+
[[36m2026-07-21 07:50:12,391[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 38 |
+
[[36m2026-07-21 07:50:12,867[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 39 |
+
|
| 40 |
+
Evaluated 100 examples
|
| 41 |
+
[[36m2026-07-21 07:50:13,686[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 42 |
+
|
| 43 |
+
Evaluated 100 examples
|
| 44 |
+
[[36m2026-07-21 07:50:15,191[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 45 |
+
[[36m2026-07-21 07:50:15,456[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 46 |
+
|
| 47 |
+
Evaluated 100 examples
|
| 48 |
+
[[36m2026-07-21 07:50:16,916[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 49 |
+
[[36m2026-07-21 07:50:16,916[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 50 |
+
[[36m2026-07-21 07:50:16,916[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 51 |
+
[[36m2026-07-21 07:50:17,219[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 52 |
+
|
| 53 |
+
Evaluated 117 examples
|
| 54 |
+
[[36m2026-07-21 07:50:17,996[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 55 |
+
|
| 56 |
+
Evaluated 117 examples
|
| 57 |
+
[[36m2026-07-21 07:50:19,476[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 58 |
+
[[36m2026-07-21 07:50:19,733[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 59 |
+
|
| 60 |
+
Evaluated 117 examples
|
| 61 |
+
[[36m2026-07-21 07:50:21,448[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 62 |
+
[[36m2026-07-21 07:50:21,448[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 63 |
+
[[36m2026-07-21 07:50:21,448[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 64 |
+
[[36m2026-07-21 07:50:21,448[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 65 |
+
[[36m2026-07-21 07:50:21,449[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.57547938415897[0m
|
| 66 |
+
[[36m2026-07-21 07:50:22,046[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 67 |
+
[[36m2026-07-21 07:50:22,061[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 68 |
+
|
| 69 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 70 |
15%|█▌ | 2/13 [00:00<00:00, 12.98it/s]
|
| 71 |
31%|███ | 4/13 [00:00<00:00, 11.02it/s]
|
| 72 |
46%|████▌ | 6/13 [00:00<00:00, 11.29it/s]
|
| 73 |
62%|██████▏ | 8/13 [00:00<00:00, 11.46it/s]
|
| 74 |
77%|███████▋ | 10/13 [00:00<00:00, 11.72it/s]
|
| 75 |
92%|█████████▏| 12/13 [00:01<00:00, 11.47it/s]
|
| 76 |
+
|
| 77 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 78 |
15%|█▌ | 2/13 [00:00<00:00, 11.25it/s]
|
| 79 |
31%|███ | 4/13 [00:00<00:00, 11.11it/s]
|
| 80 |
46%|████▌ | 6/13 [00:00<00:00, 11.41it/s]
|
| 81 |
62%|██████▏ | 8/13 [00:00<00:00, 11.44it/s]
|
| 82 |
77%|███████▋ | 10/13 [00:00<00:00, 11.45it/s]
|
| 83 |
92%|█████████▏| 12/13 [00:01<00:00, 11.57it/s]
|
| 84 |
+
[[36m2026-07-21 07:50:24,247[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 85 |
+
[[36m2026-07-21 07:50:24,261[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 86 |
+
[[36m2026-07-21 07:50:24,261[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.17090156519676[0m
|
| 87 |
+
[[36m2026-07-21 07:50:24,565[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 88 |
+
|
| 89 |
+
Evaluated 400 examples
|
| 90 |
+
[[36m2026-07-21 07:50:25,754[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.5633441753353804[0m
|
logs/eval_IdkNLL_lr2e-05_alpha10_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.74it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.30it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.58it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 12.79it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.16it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 12.88it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 12.58it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.49it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 12.96it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.02it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.08it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.28it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:11:44,125[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr2e-05_alpha10_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:11:44,126[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:11:44,126[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr2e-05_alpha10_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:11:44,126[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr2e-05_alpha10_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:11:44,903[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:11:48,373[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:12:02,724[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:12:02,725[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.48126292587804714[0m
|
| 15 |
+
[[36m2026-07-22 17:12:02,729[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:12:02,730[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:12:02,737[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:12:02,737[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 4.353260441808186e-19[0m
|
| 19 |
+
[[36m2026-07-22 17:12:03,083[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:12:05,932[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.7937744140625[0m
|
| 23 |
+
[[36m2026-07-22 17:12:06,171[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:12:09,766[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.21226291796148353[0m
|
| 27 |
+
[[36m2026-07-22 17:12:10,136[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:12:13,138[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:12:17,049[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:12:20,148[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:12:34,148[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:12:35,333[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:12:36,260[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:12:37,835[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:12:38,127[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:12:38,952[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:12:38,952[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:12:38,953[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:12:39,186[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:12:39,912[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:12:41,462[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:12:41,731[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:12:42,712[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:12:42,712[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:12:42,712[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:12:42,712[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:12:42,713[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5712395990961581[0m
|
| 68 |
+
[[36m2026-07-22 17:12:43,297[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:12:43,305[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.74it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.30it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.58it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 12.79it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.16it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 12.88it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 12.58it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.49it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 12.96it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.02it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.08it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.28it/s]
|
| 86 |
+
[[36m2026-07-22 17:12:45,238[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:12:45,245[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:12:45,245[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.18709884809991[0m
|
| 89 |
+
[[36m2026-07-22 17:12:45,523[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:12:46,542[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.5530273167686603[0m
|
logs/eval_IdkNLL_lr2e-05_alpha1_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
8%|▊ | 1/13 [00:00<00:01, 8.31it/s]
|
| 2 |
15%|█▌ | 2/13 [00:00<00:01, 7.55it/s]
|
| 3 |
23%|██▎ | 3/13 [00:00<00:01, 6.36it/s]
|
| 4 |
31%|███ | 4/13 [00:00<00:01, 7.13it/s]
|
| 5 |
38%|███▊ | 5/13 [00:00<00:01, 6.92it/s]
|
| 6 |
46%|████▌ | 6/13 [00:00<00:00, 7.05it/s]
|
| 7 |
54%|█████▍ | 7/13 [00:01<00:00, 6.80it/s]
|
| 8 |
62%|██████▏ | 8/13 [00:01<00:00, 7.04it/s]
|
| 9 |
69%|██████▉ | 9/13 [00:01<00:00, 7.12it/s]
|
| 10 |
77%|███████▋ | 10/13 [00:01<00:00, 7.05it/s]
|
| 11 |
85%|████████▍ | 11/13 [00:01<00:00, 7.62it/s]
|
| 12 |
92%|█████████▏| 12/13 [00:01<00:00, 7.13it/s]
|
|
|
|
| 13 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 14 |
8%|▊ | 1/13 [00:00<00:01, 6.75it/s]
|
| 15 |
15%|█▌ | 2/13 [00:00<00:01, 6.72it/s]
|
| 16 |
23%|██▎ | 3/13 [00:00<00:01, 6.64it/s]
|
| 17 |
31%|███ | 4/13 [00:00<00:01, 7.07it/s]
|
| 18 |
38%|███▊ | 5/13 [00:00<00:01, 7.01it/s]
|
| 19 |
46%|████▌ | 6/13 [00:00<00:00, 7.22it/s]
|
| 20 |
54%|█████▍ | 7/13 [00:00<00:00, 7.31it/s]
|
| 21 |
62%|██████▏ | 8/13 [00:01<00:00, 7.38it/s]
|
| 22 |
77%|███████▋ | 10/13 [00:01<00:00, 8.93it/s]
|
| 23 |
92%|█████████▏| 12/13 [00:01<00:00, 9.98it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-20 14:45:55,988[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr2e-05_alpha1_epoch10[0m
|
| 4 |
+
[[36m2026-07-20 14:45:55,989[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-20 14:45:55,989[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr2e-05_alpha1_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-20 14:45:55,989[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr2e-05_alpha1_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-20 14:45:56,677[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-20 14:46:02,078[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-20 14:46:20,861[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-20 14:46:20,862[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4792364398260141[0m
|
| 15 |
+
[[36m2026-07-20 14:46:20,870[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-20 14:46:20,871[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-20 14:46:20,886[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-20 14:46:20,888[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 2.213522258546535e-19[0m
|
| 19 |
+
[[36m2026-07-20 14:46:21,255[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-20 14:46:24,621[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.74496826171875[0m
|
| 23 |
+
[[36m2026-07-20 14:46:24,981[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-20 14:46:27,461[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.009941070158404315[0m
|
| 27 |
+
[[36m2026-07-20 14:46:27,762[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-20 14:46:32,692[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-20 14:46:45,227[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-20 14:46:50,494[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-20 14:47:10,872[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-20 14:47:11,281[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-20 14:47:12,475[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-20 14:47:15,465[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-20 14:47:15,825[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-20 14:47:17,627[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-20 14:47:17,627[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-20 14:47:17,627[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-20 14:47:17,973[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-20 14:47:18,928[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-20 14:47:21,019[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-20 14:47:21,275[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-20 14:47:23,250[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-20 14:47:23,250[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-20 14:47:23,250[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-20 14:47:23,251[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-20 14:47:23,252[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.4919533605980839[0m
|
| 68 |
+
[[36m2026-07-20 14:47:23,921[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-20 14:47:23,936[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
8%|▊ | 1/13 [00:00<00:01, 8.31it/s]
|
| 73 |
15%|█▌ | 2/13 [00:00<00:01, 7.55it/s]
|
| 74 |
23%|██▎ | 3/13 [00:00<00:01, 6.36it/s]
|
| 75 |
31%|███ | 4/13 [00:00<00:01, 7.13it/s]
|
| 76 |
38%|███▊ | 5/13 [00:00<00:01, 6.92it/s]
|
| 77 |
46%|████▌ | 6/13 [00:00<00:00, 7.05it/s]
|
| 78 |
54%|█████▍ | 7/13 [00:01<00:00, 6.80it/s]
|
| 79 |
62%|██████▏ | 8/13 [00:01<00:00, 7.04it/s]
|
| 80 |
69%|██████▉ | 9/13 [00:01<00:00, 7.12it/s]
|
| 81 |
77%|███████▋ | 10/13 [00:01<00:00, 7.05it/s]
|
| 82 |
85%|████████▍ | 11/13 [00:01<00:00, 7.62it/s]
|
| 83 |
92%|█████████▏| 12/13 [00:01<00:00, 7.13it/s]
|
| 84 |
+
|
| 85 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 86 |
8%|▊ | 1/13 [00:00<00:01, 6.75it/s]
|
| 87 |
15%|█▌ | 2/13 [00:00<00:01, 6.72it/s]
|
| 88 |
23%|██▎ | 3/13 [00:00<00:01, 6.64it/s]
|
| 89 |
31%|███ | 4/13 [00:00<00:01, 7.07it/s]
|
| 90 |
38%|███▊ | 5/13 [00:00<00:01, 7.01it/s]
|
| 91 |
46%|████▌ | 6/13 [00:00<00:00, 7.22it/s]
|
| 92 |
54%|█████▍ | 7/13 [00:00<00:00, 7.31it/s]
|
| 93 |
62%|██████▏ | 8/13 [00:01<00:00, 7.38it/s]
|
| 94 |
77%|███████▋ | 10/13 [00:01<00:00, 8.93it/s]
|
| 95 |
92%|█████████▏| 12/13 [00:01<00:00, 9.98it/s]
|
| 96 |
+
[[36m2026-07-20 14:47:27,207[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 97 |
+
[[36m2026-07-20 14:47:27,221[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 98 |
+
[[36m2026-07-20 14:47:27,221[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.94110261400806[0m
|
| 99 |
+
[[36m2026-07-20 14:47:27,588[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 100 |
+
|
| 101 |
+
Evaluated 400 examples
|
| 102 |
+
[[36m2026-07-20 14:47:30,857[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.5464293553274634[0m
|
logs/eval_IdkNLL_lr2e-05_alpha1_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.24it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.24it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.61it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 12.69it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.05it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 12.68it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 12.75it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.78it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.34it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.45it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.45it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.34it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:05:22,869[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr2e-05_alpha1_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:05:22,870[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:05:22,870[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr2e-05_alpha1_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:05:22,870[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr2e-05_alpha1_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:05:23,525[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:05:27,086[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:05:41,485[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:05:41,486[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4781734859796574[0m
|
| 15 |
+
[[36m2026-07-22 17:05:41,490[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:05:41,490[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:05:41,498[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:05:41,498[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 1.11924339764853e-19[0m
|
| 19 |
+
[[36m2026-07-22 17:05:41,839[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:05:44,712[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.74934814453125[0m
|
| 23 |
+
[[36m2026-07-22 17:05:45,018[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:05:46,429[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.006836801297299867[0m
|
| 27 |
+
[[36m2026-07-22 17:05:46,678[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:05:49,693[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:05:53,746[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:05:56,839[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:06:10,849[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:06:11,184[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:06:11,988[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:06:13,552[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:06:13,806[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:06:14,357[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:06:14,357[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:06:14,357[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:06:14,616[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:06:15,327[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:06:16,881[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:06:17,124[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:06:18,070[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:06:18,070[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:06:18,070[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:06:18,071[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:06:18,071[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.4494807077086565[0m
|
| 68 |
+
[[36m2026-07-22 17:06:18,729[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:06:18,736[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.24it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.24it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.61it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 12.69it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.05it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 12.68it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 12.75it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.78it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.34it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.45it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.45it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.34it/s]
|
| 86 |
+
[[36m2026-07-22 17:06:20,666[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:06:20,672[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:06:20,673[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.03423699070129[0m
|
| 89 |
+
[[36m2026-07-22 17:06:20,945[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:06:21,951[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.536815840914721[0m
|
logs/eval_IdkNLL_lr2e-05_alpha2_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.26it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.24it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.95it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.33it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.74it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.44it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.78it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.62it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 14.05it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 14.07it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 14.13it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 14.24it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:14:50,221[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr2e-05_alpha2_epoch10[0m
|
| 4 |
+
[[36m2026-07-22 17:14:50,222[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:14:50,222[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr2e-05_alpha2_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:14:50,222[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr2e-05_alpha2_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:14:50,922[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:14:54,364[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:15:08,758[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:15:08,758[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4787008684405464[0m
|
| 15 |
+
[[36m2026-07-22 17:15:08,762[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:15:08,763[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:15:08,769[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:15:08,770[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 1.11924339764853e-19[0m
|
| 19 |
+
[[36m2026-07-22 17:15:09,179[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:15:12,055[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.75279541015625[0m
|
| 23 |
+
[[36m2026-07-22 17:15:12,389[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:15:14,049[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.019885179327573877[0m
|
| 27 |
+
[[36m2026-07-22 17:15:14,590[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:15:17,619[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:15:21,801[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:15:24,882[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:15:38,894[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:15:39,283[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:15:40,096[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:15:41,667[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:15:41,933[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:15:42,622[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:15:42,622[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:15:42,622[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:15:42,856[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:15:43,580[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:15:45,142[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:15:45,388[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:15:46,378[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:15:46,378[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:15:46,378[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:15:46,378[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:15:46,379[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5483017364641685[0m
|
| 68 |
+
[[36m2026-07-22 17:15:46,897[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:15:46,905[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.26it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.24it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.95it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.33it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.74it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.44it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.78it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.62it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 14.05it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 14.07it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 14.13it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 14.24it/s]
|
| 86 |
+
[[36m2026-07-22 17:15:48,735[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:15:48,741[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:15:48,741[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.96033688745557[0m
|
| 89 |
+
[[36m2026-07-22 17:15:49,108[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:15:50,094[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.5469145524773721[0m
|
logs/eval_IdkNLL_lr2e-05_alpha2_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.55it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.20it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.66it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 12.75it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.20it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 12.80it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 12.85it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.39it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 12.93it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 12.99it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 12.78it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 12.94it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:13:16,964[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr2e-05_alpha2_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:13:16,965[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:13:16,965[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr2e-05_alpha2_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:13:16,965[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr2e-05_alpha2_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:13:17,655[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:13:21,411[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:13:35,758[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:13:35,759[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4780185789707035[0m
|
| 15 |
+
[[36m2026-07-22 17:13:35,763[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:13:35,763[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:13:35,771[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:13:35,771[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 1.3990395281788573e-20[0m
|
| 19 |
+
[[36m2026-07-22 17:13:36,132[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:13:38,987[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.75410888671875[0m
|
| 23 |
+
[[36m2026-07-22 17:13:39,231[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:13:41,076[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.02179578252579108[0m
|
| 27 |
+
[[36m2026-07-22 17:13:41,403[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:13:44,433[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:13:48,501[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:13:51,591[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:14:05,583[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:14:05,948[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:14:06,760[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:14:08,358[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:14:08,590[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:14:09,328[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:14:09,328[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:14:09,328[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:14:09,619[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:14:10,332[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:14:11,885[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:14:12,133[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:14:13,054[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:14:13,054[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:14:13,054[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:14:13,055[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:14:13,055[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5339999423611568[0m
|
| 68 |
+
[[36m2026-07-22 17:14:13,634[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:14:13,641[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.55it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.20it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.66it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 12.75it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.20it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 12.80it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 12.85it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.39it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 12.93it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 12.99it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 12.78it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 12.94it/s]
|
| 86 |
+
[[36m2026-07-22 17:14:15,596[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:14:15,605[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:14:15,605[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.03929864160851[0m
|
| 89 |
+
[[36m2026-07-22 17:14:15,873[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:14:16,918[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.5295679579660575[0m
|
logs/eval_IdkNLL_lr2e-05_alpha5_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
8%|▊ | 1/13 [00:00<00:01, 8.22it/s]
|
| 2 |
15%|█▌ | 2/13 [00:00<00:01, 7.41it/s]
|
| 3 |
23%|██▎ | 3/13 [00:00<00:01, 6.29it/s]
|
| 4 |
31%|███ | 4/13 [00:00<00:01, 6.17it/s]
|
| 5 |
38%|███▊ | 5/13 [00:00<00:01, 6.20it/s]
|
| 6 |
46%|████▌ | 6/13 [00:00<00:01, 6.48it/s]
|
| 7 |
54%|█████▍ | 7/13 [00:01<00:00, 6.27it/s]
|
| 8 |
62%|██████▏ | 8/13 [00:01<00:00, 6.13it/s]
|
| 9 |
69%|██████▉ | 9/13 [00:01<00:00, 6.47it/s]
|
| 10 |
77%|███████▋ | 10/13 [00:01<00:00, 6.56it/s]
|
| 11 |
85%|████████▍ | 11/13 [00:01<00:00, 6.48it/s]
|
| 12 |
92%|█████████▏| 12/13 [00:01<00:00, 6.38it/s]
|
|
|
|
| 13 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 14 |
8%|▊ | 1/13 [00:00<00:01, 6.01it/s]
|
| 15 |
15%|█▌ | 2/13 [00:00<00:01, 6.09it/s]
|
| 16 |
31%|███ | 4/13 [00:00<00:01, 7.95it/s]
|
| 17 |
38%|███▊ | 5/13 [00:00<00:01, 7.47it/s]
|
| 18 |
46%|████▌ | 6/13 [00:00<00:00, 7.49it/s]
|
| 19 |
54%|█████▍ | 7/13 [00:00<00:00, 7.41it/s]
|
| 20 |
62%|██████▏ | 8/13 [00:01<00:00, 7.09it/s]
|
| 21 |
69%|██████▉ | 9/13 [00:01<00:00, 6.89it/s]
|
| 22 |
77%|███████▋ | 10/13 [00:01<00:00, 6.99it/s]
|
| 23 |
85%|████████▍ | 11/13 [00:01<00:00, 6.96it/s]
|
| 24 |
92%|█████████▏| 12/13 [00:01<00:00, 6.87it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-20 14:49:45,603[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr2e-05_alpha5_epoch10[0m
|
| 4 |
+
[[36m2026-07-20 14:49:45,604[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-20 14:49:45,604[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr2e-05_alpha5_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-20 14:49:45,605[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr2e-05_alpha5_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-20 14:49:46,497[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-20 14:49:50,979[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-20 14:50:05,866[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-20 14:50:05,867[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.48091575785024465[0m
|
| 15 |
+
[[36m2026-07-20 14:50:05,875[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-20 14:50:05,877[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-20 14:50:05,892[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-20 14:50:05,894[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 4.353260441808186e-19[0m
|
| 19 |
+
[[36m2026-07-20 14:50:06,428[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-20 14:50:09,117[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.775908203125[0m
|
| 23 |
+
[[36m2026-07-20 14:50:09,601[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-20 14:50:14,209[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.06299447888449944[0m
|
| 27 |
+
[[36m2026-07-20 14:50:14,612[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-20 14:50:18,417[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-20 14:50:36,612[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-20 14:50:41,610[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-20 14:51:00,901[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-20 14:51:01,479[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-20 14:51:02,780[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-20 14:51:05,078[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-20 14:51:05,392[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-20 14:51:07,037[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-20 14:51:07,037[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-20 14:51:07,037[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-20 14:51:07,322[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-20 14:51:08,373[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-20 14:51:10,575[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-20 14:51:10,985[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-20 14:51:13,572[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-20 14:51:13,572[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-20 14:51:13,572[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-20 14:51:13,573[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-20 14:51:13,573[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5701453352863971[0m
|
| 68 |
+
[[36m2026-07-20 14:51:14,178[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-20 14:51:14,195[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
8%|▊ | 1/13 [00:00<00:01, 8.22it/s]
|
| 73 |
15%|█▌ | 2/13 [00:00<00:01, 7.41it/s]
|
| 74 |
23%|██▎ | 3/13 [00:00<00:01, 6.29it/s]
|
| 75 |
31%|███ | 4/13 [00:00<00:01, 6.17it/s]
|
| 76 |
38%|███▊ | 5/13 [00:00<00:01, 6.20it/s]
|
| 77 |
46%|████▌ | 6/13 [00:00<00:01, 6.48it/s]
|
| 78 |
54%|█████▍ | 7/13 [00:01<00:00, 6.27it/s]
|
| 79 |
62%|██████▏ | 8/13 [00:01<00:00, 6.13it/s]
|
| 80 |
69%|██████▉ | 9/13 [00:01<00:00, 6.47it/s]
|
| 81 |
77%|███████▋ | 10/13 [00:01<00:00, 6.56it/s]
|
| 82 |
85%|████████▍ | 11/13 [00:01<00:00, 6.48it/s]
|
| 83 |
92%|█████████▏| 12/13 [00:01<00:00, 6.38it/s]
|
| 84 |
+
|
| 85 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 86 |
8%|▊ | 1/13 [00:00<00:01, 6.01it/s]
|
| 87 |
15%|█▌ | 2/13 [00:00<00:01, 6.09it/s]
|
| 88 |
31%|███ | 4/13 [00:00<00:01, 7.95it/s]
|
| 89 |
38%|███▊ | 5/13 [00:00<00:01, 7.47it/s]
|
| 90 |
46%|████▌ | 6/13 [00:00<00:00, 7.49it/s]
|
| 91 |
54%|█████▍ | 7/13 [00:00<00:00, 7.41it/s]
|
| 92 |
62%|██████▏ | 8/13 [00:01<00:00, 7.09it/s]
|
| 93 |
69%|██████▉ | 9/13 [00:01<00:00, 6.89it/s]
|
| 94 |
77%|███████▋ | 10/13 [00:01<00:00, 6.99it/s]
|
| 95 |
85%|████████▍ | 11/13 [00:01<00:00, 6.96it/s]
|
| 96 |
92%|█████████▏| 12/13 [00:01<00:00, 6.87it/s]
|
| 97 |
+
[[36m2026-07-20 14:51:17,931[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 98 |
+
[[36m2026-07-20 14:51:17,946[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 99 |
+
[[36m2026-07-20 14:51:17,946[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.08687816013655[0m
|
| 100 |
+
[[36m2026-07-20 14:51:18,239[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 101 |
+
|
| 102 |
+
Evaluated 400 examples
|
| 103 |
+
[[36m2026-07-20 14:51:21,683[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.5407521460937741[0m
|
logs/eval_IdkNLL_lr2e-05_alpha5_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.63it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.29it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.95it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 12.99it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.54it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.32it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.24it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.80it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.12it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.12it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.18it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.31it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:06:51,713[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr2e-05_alpha5_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:06:51,714[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:06:51,714[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr2e-05_alpha5_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:06:51,714[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr2e-05_alpha5_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:06:52,346[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:06:55,832[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:07:10,263[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:07:10,263[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.478923026212248[0m
|
| 15 |
+
[[36m2026-07-22 17:07:10,268[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:07:10,269[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:07:10,277[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:07:10,278[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 5.627715395305929e-20[0m
|
| 19 |
+
[[36m2026-07-22 17:07:10,666[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:07:13,524[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.7750537109375[0m
|
| 23 |
+
[[36m2026-07-22 17:07:13,767[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:07:16,727[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.09126959265861279[0m
|
| 27 |
+
[[36m2026-07-22 17:07:17,012[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:07:20,042[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:07:24,211[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:07:27,304[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:07:41,313[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:07:41,686[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:07:42,457[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:07:44,042[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:07:44,353[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:07:45,157[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:07:45,157[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:07:45,157[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:07:45,423[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:07:46,193[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:07:47,749[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:07:48,000[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:07:48,959[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:07:48,959[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:07:48,959[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:07:48,959[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:07:48,959[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5644533474892907[0m
|
| 68 |
+
[[36m2026-07-22 17:07:49,480[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:07:49,491[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.63it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.29it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.95it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 12.99it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.54it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.32it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.24it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.80it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.12it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.12it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.18it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.31it/s]
|
| 86 |
+
[[36m2026-07-22 17:07:51,391[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:07:51,398[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:07:51,398[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.10914942412843[0m
|
| 89 |
+
[[36m2026-07-22 17:07:51,646[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:07:52,691[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.545986699317964[0m
|
logs/eval_IdkNLL_lr3e-05_alpha10_epoch10.log
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.28it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 11.48it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 11.80it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 11.93it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 12.19it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:01<00:00, 11.90it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 11.73it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 11.58it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 11.86it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 11.91it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 11.93it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:01<00:00, 12.04it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[[36m2026-07-21 07:50:40,346[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr3e-05_alpha10_epoch10[0m
|
| 2 |
+
[[36m2026-07-21 07:50:40,348[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 3 |
+
[[36m2026-07-21 07:50:40,348[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr3e-05_alpha10_epoch10/TOFU_EVAL.json[0m
|
| 4 |
+
[[36m2026-07-21 07:50:40,348[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr3e-05_alpha10_epoch10/TOFU_SUMMARY.json[0m
|
| 5 |
+
[[36m2026-07-21 07:50:41,087[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 6 |
+
|
| 7 |
+
Evaluated 400 examples
|
| 8 |
+
[[36m2026-07-21 07:50:44,609[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 9 |
+
|
| 10 |
+
Evaluated 400 examples
|
| 11 |
+
[[36m2026-07-21 07:50:57,712[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 12 |
+
[[36m2026-07-21 07:50:57,713[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4951789649259015[0m
|
| 13 |
+
[[36m2026-07-21 07:50:57,721[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 14 |
+
[[36m2026-07-21 07:50:57,722[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 15 |
+
[[36m2026-07-21 07:50:57,738[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 16 |
+
[[36m2026-07-21 07:50:57,739[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 2.241003819205479e-17[0m
|
| 17 |
+
[[36m2026-07-21 07:50:58,119[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 18 |
+
|
| 19 |
+
Evaluated 400 examples
|
| 20 |
+
[[36m2026-07-21 07:51:00,785[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.744580078125[0m
|
| 21 |
+
[[36m2026-07-21 07:51:01,061[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 22 |
+
|
| 23 |
+
Evaluated 400 examples
|
| 24 |
+
[[36m2026-07-21 07:51:06,608[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.10923903386871697[0m
|
| 25 |
+
[[36m2026-07-21 07:51:06,933[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 26 |
+
|
| 27 |
+
Evaluated 400 examples
|
| 28 |
+
[[36m2026-07-21 07:51:09,781[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 29 |
+
|
| 30 |
+
Evaluated 400 examples
|
| 31 |
+
[[36m2026-07-21 07:51:17,890[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 32 |
+
|
| 33 |
+
Evaluated 400 examples
|
| 34 |
+
[[36m2026-07-21 07:51:20,800[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 35 |
+
|
| 36 |
+
Evaluated 400 examples
|
| 37 |
+
[[36m2026-07-21 07:51:33,601[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 38 |
+
[[36m2026-07-21 07:51:33,969[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 39 |
+
|
| 40 |
+
Evaluated 100 examples
|
| 41 |
+
[[36m2026-07-21 07:51:34,754[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 42 |
+
|
| 43 |
+
Evaluated 100 examples
|
| 44 |
+
[[36m2026-07-21 07:51:36,251[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 45 |
+
[[36m2026-07-21 07:51:36,518[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 46 |
+
|
| 47 |
+
Evaluated 100 examples
|
| 48 |
+
[[36m2026-07-21 07:51:38,045[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 49 |
+
[[36m2026-07-21 07:51:38,046[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 50 |
+
[[36m2026-07-21 07:51:38,046[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 51 |
+
[[36m2026-07-21 07:51:38,313[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 52 |
+
|
| 53 |
+
Evaluated 117 examples
|
| 54 |
+
[[36m2026-07-21 07:51:39,031[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 55 |
+
|
| 56 |
+
Evaluated 117 examples
|
| 57 |
+
[[36m2026-07-21 07:51:40,502[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 58 |
+
[[36m2026-07-21 07:51:40,771[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 59 |
+
|
| 60 |
+
Evaluated 117 examples
|
| 61 |
+
[[36m2026-07-21 07:51:42,512[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 62 |
+
[[36m2026-07-21 07:51:42,512[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 63 |
+
[[36m2026-07-21 07:51:42,513[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 64 |
+
[[36m2026-07-21 07:51:42,513[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 65 |
+
[[36m2026-07-21 07:51:42,514[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5677726043908246[0m
|
| 66 |
+
[[36m2026-07-21 07:51:43,074[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 67 |
+
[[36m2026-07-21 07:51:43,089[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 68 |
+
|
| 69 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 70 |
15%|█▌ | 2/13 [00:00<00:00, 13.28it/s]
|
| 71 |
31%|███ | 4/13 [00:00<00:00, 11.48it/s]
|
| 72 |
46%|████▌ | 6/13 [00:00<00:00, 11.80it/s]
|
| 73 |
62%|██████▏ | 8/13 [00:00<00:00, 11.93it/s]
|
| 74 |
77%|███████▋ | 10/13 [00:00<00:00, 12.19it/s]
|
| 75 |
92%|█████████▏| 12/13 [00:01<00:00, 11.90it/s]
|
| 76 |
+
|
| 77 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 78 |
15%|█▌ | 2/13 [00:00<00:00, 11.73it/s]
|
| 79 |
31%|███ | 4/13 [00:00<00:00, 11.58it/s]
|
| 80 |
46%|████▌ | 6/13 [00:00<00:00, 11.86it/s]
|
| 81 |
62%|██████▏ | 8/13 [00:00<00:00, 11.91it/s]
|
| 82 |
77%|███████▋ | 10/13 [00:00<00:00, 11.93it/s]
|
| 83 |
92%|█████████▏| 12/13 [00:01<00:00, 12.04it/s]
|
| 84 |
+
[[36m2026-07-21 07:51:45,191[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 85 |
+
[[36m2026-07-21 07:51:45,205[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 86 |
+
[[36m2026-07-21 07:51:45,205[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.99475611362479[0m
|
| 87 |
+
[[36m2026-07-21 07:51:45,496[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 88 |
+
|
| 89 |
+
Evaluated 400 examples
|
| 90 |
+
[[36m2026-07-21 07:51:46,668[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.4922378536512631[0m
|
logs/eval_IdkNLL_lr3e-05_alpha10_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.22it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.09it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.95it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.47it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.93it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.74it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.56it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.42it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.91it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.93it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 14.03it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 14.13it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:16:24,031[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr3e-05_alpha10_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:16:24,032[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:16:24,032[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr3e-05_alpha10_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:16:24,032[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr3e-05_alpha10_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:16:24,649[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:16:28,617[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:16:42,972[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:16:42,973[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.501259620393524[0m
|
| 15 |
+
[[36m2026-07-22 17:16:42,977[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:16:42,977[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:16:42,984[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:16:42,985[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 4.2392996435511593e-17[0m
|
| 19 |
+
[[36m2026-07-22 17:16:43,732[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:16:46,629[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.7376806640625[0m
|
| 23 |
+
[[36m2026-07-22 17:16:46,885[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:16:50,367[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.14538918415014265[0m
|
| 27 |
+
[[36m2026-07-22 17:16:50,699[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:16:53,716[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:16:57,806[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:17:01,300[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:17:15,261[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:17:16,504[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:17:17,279[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:17:18,887[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:17:19,122[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:17:19,843[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:17:19,844[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:17:19,844[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:17:20,415[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:17:21,185[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:17:22,737[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:17:22,982[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:17:23,967[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:17:23,967[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:17:23,967[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:17:23,967[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:17:23,968[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5610351502264007[0m
|
| 68 |
+
[[36m2026-07-22 17:17:25,769[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:17:25,779[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.22it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.09it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.95it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.47it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.93it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.74it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.56it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.42it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.91it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.93it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 14.03it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 14.13it/s]
|
| 86 |
+
[[36m2026-07-22 17:17:27,607[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:17:27,614[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:17:27,614[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -99.02613834924968[0m
|
| 89 |
+
[[36m2026-07-22 17:17:28,178[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:17:29,185[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.46257153663969325[0m
|
logs/eval_IdkNLL_lr3e-05_alpha1_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.23it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 11.88it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.55it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 12.97it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.41it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.21it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.05it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.92it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.35it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.35it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.40it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.50it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:14:51,160[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr3e-05_alpha1_epoch10[0m
|
| 4 |
+
[[36m2026-07-22 17:14:51,161[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:14:51,161[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr3e-05_alpha1_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:14:51,161[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr3e-05_alpha1_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:14:51,829[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:14:55,377[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:15:09,730[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:15:09,731[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4949914861684177[0m
|
| 15 |
+
[[36m2026-07-22 17:15:09,735[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:15:09,735[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:15:09,742[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:15:09,742[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 3.2024272501633618e-18[0m
|
| 19 |
+
[[36m2026-07-22 17:15:10,101[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:15:12,984[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.67515380859375[0m
|
| 23 |
+
[[36m2026-07-22 17:15:13,235[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:15:14,604[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.009139949321256842[0m
|
| 27 |
+
[[36m2026-07-22 17:15:14,906[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:15:17,887[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:15:22,075[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:15:25,150[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:15:39,148[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:15:39,488[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:15:40,274[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:15:41,846[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:15:42,152[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:15:42,860[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:15:42,860[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:15:42,860[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:15:43,110[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:15:43,857[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:15:45,418[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:15:45,664[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:15:46,652[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:15:46,653[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:15:46,653[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:15:46,653[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:15:46,653[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.4917842779052955[0m
|
| 68 |
+
[[36m2026-07-22 17:15:47,155[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:15:47,166[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.23it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 11.88it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.55it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 12.97it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.41it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.21it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.05it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.92it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.35it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.35it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.40it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.50it/s]
|
| 86 |
+
[[36m2026-07-22 17:15:49,066[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:15:49,073[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:15:49,073[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.5989350126786[0m
|
| 89 |
+
[[36m2026-07-22 17:15:49,326[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:15:50,368[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.41322452248359304[0m
|
logs/eval_IdkNLL_lr3e-05_alpha1_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.04it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.24it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.90it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.21it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.62it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.36it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.18it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.95it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.34it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.35it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.35it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.44it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:08:22,993[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr3e-05_alpha1_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:08:22,993[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:08:22,993[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr3e-05_alpha1_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:08:22,994[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr3e-05_alpha1_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:08:23,802[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:08:27,465[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:08:41,905[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:08:41,905[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4978633349891568[0m
|
| 15 |
+
[[36m2026-07-22 17:08:41,909[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:08:41,910[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:08:41,916[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:08:41,917[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 8.513756583610105e-19[0m
|
| 19 |
+
[[36m2026-07-22 17:08:42,259[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:08:45,113[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.67819091796875[0m
|
| 23 |
+
[[36m2026-07-22 17:08:45,420[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:08:46,720[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.008150290861223104[0m
|
| 27 |
+
[[36m2026-07-22 17:08:47,190[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:08:50,251[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:08:54,012[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:08:57,087[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:09:11,108[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:09:11,462[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:09:12,257[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:09:13,854[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:09:14,107[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:09:14,755[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:09:14,755[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:09:14,755[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:09:14,992[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:09:15,708[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:09:17,250[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:09:17,546[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:09:18,351[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:09:18,352[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:09:18,352[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:09:18,352[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:09:18,352[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.44566065985412656[0m
|
| 68 |
+
[[36m2026-07-22 17:09:18,942[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:09:18,949[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.04it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.24it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.90it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.21it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.62it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.36it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.18it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.95it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.34it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.35it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.35it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.44it/s]
|
| 86 |
+
[[36m2026-07-22 17:09:20,839[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:09:20,846[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:09:20,846[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.70016803082339[0m
|
| 89 |
+
[[36m2026-07-22 17:09:21,132[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:09:22,248[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.4052569846595127[0m
|
logs/eval_IdkNLL_lr3e-05_alpha2_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
8%|▊ | 1/13 [00:00<00:01, 6.66it/s]
|
| 2 |
15%|█▌ | 2/13 [00:00<00:01, 7.82it/s]
|
| 3 |
31%|███ | 4/13 [00:00<00:00, 9.40it/s]
|
| 4 |
46%|████▌ | 6/13 [00:00<00:00, 10.47it/s]
|
| 5 |
62%|██████▏ | 8/13 [00:00<00:00, 10.98it/s]
|
| 6 |
77%|███████▋ | 10/13 [00:00<00:00, 11.46it/s]
|
| 7 |
92%|█████████▏| 12/13 [00:01<00:00, 11.35it/s]
|
|
|
|
| 8 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 9 |
15%|█▌ | 2/13 [00:00<00:00, 11.43it/s]
|
| 10 |
31%|███ | 4/13 [00:00<00:00, 11.27it/s]
|
| 11 |
46%|████▌ | 6/13 [00:00<00:00, 11.63it/s]
|
| 12 |
62%|██████▏ | 8/13 [00:00<00:00, 11.69it/s]
|
| 13 |
77%|███████▋ | 10/13 [00:00<00:00, 11.72it/s]
|
| 14 |
92%|█████████▏| 12/13 [00:01<00:00, 6.36it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-20 14:54:04,206[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr3e-05_alpha2_epoch10[0m
|
| 4 |
+
[[36m2026-07-20 14:54:04,207[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-20 14:54:04,207[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr3e-05_alpha2_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-20 14:54:04,207[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr3e-05_alpha2_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-20 14:54:04,932[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-20 14:54:09,818[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-20 14:54:28,938[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-20 14:54:28,939[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4993750635808588[0m
|
| 15 |
+
[[36m2026-07-20 14:54:28,947[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-20 14:54:28,948[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-20 14:54:28,964[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-20 14:54:28,966[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 3.2024272501633618e-18[0m
|
| 19 |
+
[[36m2026-07-20 14:54:29,472[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-20 14:54:34,117[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.6839794921875[0m
|
| 23 |
+
[[36m2026-07-20 14:54:34,550[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-20 14:54:39,888[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.01505161696737646[0m
|
| 27 |
+
[[36m2026-07-20 14:54:40,288[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-20 14:54:45,615[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-20 14:55:00,609[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-20 14:55:05,074[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-20 14:55:24,256[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-20 14:55:24,881[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-20 14:55:26,179[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-20 14:55:28,198[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-20 14:55:28,569[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-20 14:55:31,010[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-20 14:55:31,010[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-20 14:55:31,010[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-20 14:55:31,452[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-20 14:55:32,687[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-20 14:55:34,566[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-20 14:55:34,956[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-20 14:55:38,219[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-20 14:55:38,220[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-20 14:55:38,220[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-20 14:55:38,220[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-20 14:55:38,221[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5450860673113619[0m
|
| 68 |
+
[[36m2026-07-20 14:55:39,086[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-20 14:55:39,175[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
8%|▊ | 1/13 [00:00<00:01, 6.66it/s]
|
| 73 |
15%|█▌ | 2/13 [00:00<00:01, 7.82it/s]
|
| 74 |
31%|███ | 4/13 [00:00<00:00, 9.40it/s]
|
| 75 |
46%|████▌ | 6/13 [00:00<00:00, 10.47it/s]
|
| 76 |
62%|██████▏ | 8/13 [00:00<00:00, 10.98it/s]
|
| 77 |
77%|███████▋ | 10/13 [00:00<00:00, 11.46it/s]
|
| 78 |
92%|█████████▏| 12/13 [00:01<00:00, 11.35it/s]
|
| 79 |
+
|
| 80 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 81 |
15%|█▌ | 2/13 [00:00<00:00, 11.43it/s]
|
| 82 |
31%|███ | 4/13 [00:00<00:00, 11.27it/s]
|
| 83 |
46%|████▌ | 6/13 [00:00<00:00, 11.63it/s]
|
| 84 |
62%|██████▏ | 8/13 [00:00<00:00, 11.69it/s]
|
| 85 |
77%|███████▋ | 10/13 [00:00<00:00, 11.72it/s]
|
| 86 |
92%|█████████▏| 12/13 [00:01<00:00, 6.36it/s]
|
| 87 |
+
[[36m2026-07-20 14:55:41,960[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 88 |
+
[[36m2026-07-20 14:55:41,975[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 89 |
+
[[36m2026-07-20 14:55:41,975[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.66068715374693[0m
|
| 90 |
+
[[36m2026-07-20 14:55:42,404[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 91 |
+
|
| 92 |
+
Evaluated 400 examples
|
| 93 |
+
[[36m2026-07-20 14:55:47,850[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.4109878851926348[0m
|
logs/eval_IdkNLL_lr3e-05_alpha2_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.68it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.53it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.15it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.63it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 14.04it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.83it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.73it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.61it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 14.01it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 14.05it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 14.09it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 14.21it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:17:54,650[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr3e-05_alpha2_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:17:54,651[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:17:54,651[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr3e-05_alpha2_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:17:54,651[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr3e-05_alpha2_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:18:00,407[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:18:04,333[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:18:18,714[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:18:18,715[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5028844227401504[0m
|
| 15 |
+
[[36m2026-07-22 17:18:18,719[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:18:18,719[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:18:18,726[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:18:18,726[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 6.159412995752159e-18[0m
|
| 19 |
+
[[36m2026-07-22 17:18:19,097[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:18:21,965[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.68916015625[0m
|
| 23 |
+
[[36m2026-07-22 17:18:22,232[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:18:23,973[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.016458309696891105[0m
|
| 27 |
+
[[36m2026-07-22 17:18:24,251[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:18:27,246[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:18:31,343[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:18:34,545[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:18:48,545[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:18:48,957[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:18:49,846[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:18:51,429[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:18:51,732[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:18:52,414[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:18:52,414[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:18:52,414[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:18:52,709[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:18:53,480[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:18:55,055[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:18:55,580[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:18:56,730[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:18:56,730[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:18:56,730[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:18:56,730[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:18:56,731[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5224317236980688[0m
|
| 68 |
+
[[36m2026-07-22 17:18:57,399[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:18:57,406[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.68it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.53it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.15it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.63it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 14.04it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.83it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.73it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.61it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 14.01it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 14.05it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 14.09it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 14.21it/s]
|
| 86 |
+
[[36m2026-07-22 17:18:59,217[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:18:59,223[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:18:59,223[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.77305580388764[0m
|
| 89 |
+
[[36m2026-07-22 17:18:59,487[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:19:00,510[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.4085078851067708[0m
|
logs/eval_IdkNLL_lr3e-05_alpha5_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.60it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.40it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.99it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.31it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.56it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.23it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.01it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.87it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.23it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.21it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.27it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.13it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:16:20,718[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr3e-05_alpha5_epoch10[0m
|
| 4 |
+
[[36m2026-07-22 17:16:20,719[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:16:20,719[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr3e-05_alpha5_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:16:20,719[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr3e-05_alpha5_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:16:21,393[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:16:24,860[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:16:39,222[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:16:39,222[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.4968766294464798[0m
|
| 15 |
+
[[36m2026-07-22 17:16:39,226[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:16:39,227[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:16:39,233[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:16:39,234[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 6.159412995752159e-18[0m
|
| 19 |
+
[[36m2026-07-22 17:16:39,578[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:16:42,434[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.71641357421875[0m
|
| 23 |
+
[[36m2026-07-22 17:16:42,727[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:16:44,853[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.037324577993518855[0m
|
| 27 |
+
[[36m2026-07-22 17:16:45,115[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:16:53,393[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:16:58,424[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:17:02,501[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:17:16,442[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:17:16,790[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:17:17,631[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:17:19,211[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:17:21,022[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:17:21,771[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:17:21,771[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:17:21,771[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:17:22,008[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:17:24,084[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:17:25,659[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:17:25,908[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:17:26,989[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:17:26,989[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:17:26,989[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:17:26,989[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:17:26,989[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5687425441916057[0m
|
| 68 |
+
[[36m2026-07-22 17:17:28,560[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:17:28,568[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.60it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.40it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.99it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.31it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.56it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.23it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.01it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.87it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.23it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.21it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.27it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.13it/s]
|
| 86 |
+
[[36m2026-07-22 17:17:30,475[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:17:30,482[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:17:30,482[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.86214085985509[0m
|
| 89 |
+
[[36m2026-07-22 17:17:33,634[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:17:34,718[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.4545462262677516[0m
|
logs/eval_IdkNLL_lr3e-05_alpha5_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.99it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.68it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.29it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.64it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.98it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.70it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.40it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.13it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.51it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.53it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.42it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 12.88it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:09:46,598[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr3e-05_alpha5_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:09:46,599[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:09:46,599[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr3e-05_alpha5_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:09:46,599[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr3e-05_alpha5_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:09:52,693[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:09:56,400[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:10:10,830[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:10:10,830[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5024821186592119[0m
|
| 15 |
+
[[36m2026-07-22 17:10:10,835[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:10:10,835[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:10:10,843[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:10:10,844[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 6.159412995752159e-18[0m
|
| 19 |
+
[[36m2026-07-22 17:10:11,265[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:10:14,174[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.7125732421875[0m
|
| 23 |
+
[[36m2026-07-22 17:10:14,551[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:10:16,490[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.03528257578687019[0m
|
| 27 |
+
[[36m2026-07-22 17:10:16,741[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:10:19,804[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:10:24,244[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:10:27,324[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:10:41,352[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:10:41,698[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:10:42,458[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:10:44,073[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:10:44,334[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:10:45,045[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:10:45,045[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:10:45,045[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:10:45,288[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:10:46,045[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:10:47,624[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:10:47,891[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:10:48,803[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:10:48,804[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:10:48,804[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:10:48,804[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:10:48,804[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5514159224481929[0m
|
| 68 |
+
[[36m2026-07-22 17:10:49,396[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:10:49,403[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.99it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.68it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.29it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.64it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.98it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.70it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.40it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.13it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.51it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.53it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.42it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 12.88it/s]
|
| 86 |
+
[[36m2026-07-22 17:10:51,287[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:10:51,294[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:10:51,295[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.89048610493563[0m
|
| 89 |
+
[[36m2026-07-22 17:10:51,557[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:10:52,664[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.43525861530382004[0m
|
logs/eval_IdkNLL_lr4e-05_alpha10_epoch10.log
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.07it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 11.25it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 11.55it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 11.73it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 11.97it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:01<00:00, 11.71it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 11.49it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 11.33it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 11.62it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 11.69it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 11.71it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:01<00:00, 11.84it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[[36m2026-07-21 07:50:36,858[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr4e-05_alpha10_epoch10[0m
|
| 2 |
+
[[36m2026-07-21 07:50:36,860[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 3 |
+
[[36m2026-07-21 07:50:36,860[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr4e-05_alpha10_epoch10/TOFU_EVAL.json[0m
|
| 4 |
+
[[36m2026-07-21 07:50:36,860[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr4e-05_alpha10_epoch10/TOFU_SUMMARY.json[0m
|
| 5 |
+
[[36m2026-07-21 07:50:37,515[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 6 |
+
|
| 7 |
+
Evaluated 400 examples
|
| 8 |
+
[[36m2026-07-21 07:50:41,048[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 9 |
+
|
| 10 |
+
Evaluated 400 examples
|
| 11 |
+
[[36m2026-07-21 07:50:54,200[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 12 |
+
[[36m2026-07-21 07:50:54,201[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5113293245568581[0m
|
| 13 |
+
[[36m2026-07-21 07:50:54,209[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 14 |
+
[[36m2026-07-21 07:50:54,210[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 15 |
+
[[36m2026-07-21 07:50:54,225[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 16 |
+
[[36m2026-07-21 07:50:54,227[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 1.1781277220965948e-17[0m
|
| 17 |
+
[[36m2026-07-21 07:50:54,604[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 18 |
+
|
| 19 |
+
Evaluated 400 examples
|
| 20 |
+
[[36m2026-07-21 07:50:57,279[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.6776708984375[0m
|
| 21 |
+
[[36m2026-07-21 07:50:57,752[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 22 |
+
|
| 23 |
+
Evaluated 400 examples
|
| 24 |
+
[[36m2026-07-21 07:51:02,383[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.04984328424422209[0m
|
| 25 |
+
[[36m2026-07-21 07:51:02,666[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 26 |
+
|
| 27 |
+
Evaluated 400 examples
|
| 28 |
+
[[36m2026-07-21 07:51:05,505[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 29 |
+
|
| 30 |
+
Evaluated 400 examples
|
| 31 |
+
[[36m2026-07-21 07:51:13,675[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 32 |
+
|
| 33 |
+
Evaluated 400 examples
|
| 34 |
+
[[36m2026-07-21 07:51:16,596[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 35 |
+
|
| 36 |
+
Evaluated 400 examples
|
| 37 |
+
[[36m2026-07-21 07:51:29,441[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 38 |
+
[[36m2026-07-21 07:51:30,018[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 39 |
+
|
| 40 |
+
Evaluated 100 examples
|
| 41 |
+
[[36m2026-07-21 07:51:30,782[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 42 |
+
|
| 43 |
+
Evaluated 100 examples
|
| 44 |
+
[[36m2026-07-21 07:51:32,283[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 45 |
+
[[36m2026-07-21 07:51:32,543[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 46 |
+
|
| 47 |
+
Evaluated 100 examples
|
| 48 |
+
[[36m2026-07-21 07:51:33,776[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 49 |
+
[[36m2026-07-21 07:51:33,776[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 50 |
+
[[36m2026-07-21 07:51:33,776[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 51 |
+
[[36m2026-07-21 07:51:34,039[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 52 |
+
|
| 53 |
+
Evaluated 117 examples
|
| 54 |
+
[[36m2026-07-21 07:51:34,800[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 55 |
+
|
| 56 |
+
Evaluated 117 examples
|
| 57 |
+
[[36m2026-07-21 07:51:36,277[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 58 |
+
[[36m2026-07-21 07:51:36,535[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 59 |
+
|
| 60 |
+
Evaluated 117 examples
|
| 61 |
+
[[36m2026-07-21 07:51:38,263[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 62 |
+
[[36m2026-07-21 07:51:38,263[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 63 |
+
[[36m2026-07-21 07:51:38,263[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 64 |
+
[[36m2026-07-21 07:51:38,263[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 65 |
+
[[36m2026-07-21 07:51:38,264[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5621805155060373[0m
|
| 66 |
+
[[36m2026-07-21 07:51:38,799[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 67 |
+
[[36m2026-07-21 07:51:38,814[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 68 |
+
|
| 69 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 70 |
15%|█▌ | 2/13 [00:00<00:00, 13.07it/s]
|
| 71 |
31%|███ | 4/13 [00:00<00:00, 11.25it/s]
|
| 72 |
46%|████▌ | 6/13 [00:00<00:00, 11.55it/s]
|
| 73 |
62%|██████▏ | 8/13 [00:00<00:00, 11.73it/s]
|
| 74 |
77%|███████▋ | 10/13 [00:00<00:00, 11.97it/s]
|
| 75 |
92%|█████████▏| 12/13 [00:01<00:00, 11.71it/s]
|
| 76 |
+
|
| 77 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 78 |
15%|█▌ | 2/13 [00:00<00:00, 11.49it/s]
|
| 79 |
31%|███ | 4/13 [00:00<00:00, 11.33it/s]
|
| 80 |
46%|████▌ | 6/13 [00:00<00:00, 11.62it/s]
|
| 81 |
62%|██████▏ | 8/13 [00:00<00:00, 11.69it/s]
|
| 82 |
77%|███████▋ | 10/13 [00:00<00:00, 11.71it/s]
|
| 83 |
92%|█████████▏| 12/13 [00:01<00:00, 11.84it/s]
|
| 84 |
+
[[36m2026-07-21 07:51:40,954[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 85 |
+
[[36m2026-07-21 07:51:40,968[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 86 |
+
[[36m2026-07-21 07:51:40,969[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.67081045556141[0m
|
| 87 |
+
[[36m2026-07-21 07:51:41,263[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 88 |
+
|
| 89 |
+
Evaluated 400 examples
|
| 90 |
+
[[36m2026-07-21 07:51:42,482[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.3781519412748065[0m
|
logs/eval_IdkNLL_lr4e-05_alpha10_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.67it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.23it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.77it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.11it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.44it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.14it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 12.65it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.60it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.05it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.08it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.17it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.11it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:18:02,596[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr4e-05_alpha10_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:18:02,596[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:18:02,596[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr4e-05_alpha10_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:18:02,597[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr4e-05_alpha10_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:18:04,362[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:18:08,168[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:18:22,511[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:18:22,512[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5229820334714074[0m
|
| 15 |
+
[[36m2026-07-22 17:18:22,516[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:18:22,516[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:18:22,524[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:18:22,525[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 7.975373070421785e-17[0m
|
| 19 |
+
[[36m2026-07-22 17:18:22,882[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:18:25,768[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.6660107421875[0m
|
| 23 |
+
[[36m2026-07-22 17:18:26,079[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:18:29,177[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.08163743550245471[0m
|
| 27 |
+
[[36m2026-07-22 17:18:29,433[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:18:32,521[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:18:36,610[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:18:39,732[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:18:53,769[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:18:54,165[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:18:55,096[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:18:56,697[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:18:57,023[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:18:57,800[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:18:57,800[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:18:57,800[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:18:58,036[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:18:58,830[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:19:00,373[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:19:00,638[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:19:01,631[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:19:01,631[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:19:01,631[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:19:01,631[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:19:01,632[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5490092971263351[0m
|
| 68 |
+
[[36m2026-07-22 17:19:02,234[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:19:02,242[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.67it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.23it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.77it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.11it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.44it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.14it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 12.65it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.60it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.05it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.08it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.17it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.11it/s]
|
| 86 |
+
[[36m2026-07-22 17:19:04,164[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:19:04,171[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:19:04,171[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.57868840904965[0m
|
| 89 |
+
[[36m2026-07-22 17:19:04,556[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:19:05,690[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.3448394188954561[0m
|
logs/eval_IdkNLL_lr4e-05_alpha1_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.67it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.16it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.85it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.23it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.51it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.30it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.12it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.92it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.33it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.33it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.35it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.11it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:11:18,673[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr4e-05_alpha1_epoch10[0m
|
| 4 |
+
[[36m2026-07-22 17:11:18,674[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:11:18,674[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr4e-05_alpha1_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:11:18,674[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr4e-05_alpha1_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:11:19,386[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:11:22,928[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:11:37,368[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:11:37,368[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5215733989666866[0m
|
| 15 |
+
[[36m2026-07-22 17:11:37,373[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:11:37,374[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:11:37,381[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:11:37,382[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 7.975373070421785e-17[0m
|
| 19 |
+
[[36m2026-07-22 17:11:37,739[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:11:40,613[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.587607421875[0m
|
| 23 |
+
[[36m2026-07-22 17:11:40,923[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:11:42,129[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.0071564874892455456[0m
|
| 27 |
+
[[36m2026-07-22 17:11:42,380[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:11:45,392[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:11:49,551[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:11:52,631[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:12:06,651[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:12:07,128[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:12:07,914[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:12:09,519[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:12:09,747[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:12:10,432[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:12:10,433[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:12:10,433[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:12:10,706[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:12:11,472[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:12:13,017[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:12:13,262[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:12:14,129[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:12:14,129[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:12:14,129[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:12:14,129[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:12:14,129[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.508652343560777[0m
|
| 68 |
+
[[36m2026-07-22 17:12:14,623[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:12:14,630[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.67it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.16it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.85it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.23it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.51it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.30it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.12it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.92it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.33it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.33it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.35it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.11it/s]
|
| 86 |
+
[[36m2026-07-22 17:12:16,537[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:12:16,544[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:12:16,544[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -97.12599459867167[0m
|
| 89 |
+
[[36m2026-07-22 17:12:16,797[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:12:18,021[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.29804959133370734[0m
|
logs/eval_IdkNLL_lr4e-05_alpha1_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 14.22it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.93it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.59it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.93it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 14.27it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.98it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.20it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.26it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.72it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.72it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.81it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.91it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:19:36,919[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr4e-05_alpha1_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:19:36,920[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:19:36,920[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr4e-05_alpha1_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:19:36,920[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr4e-05_alpha1_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:19:37,585[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:19:41,065[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:19:55,423[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:19:55,424[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5261941641418377[0m
|
| 15 |
+
[[36m2026-07-22 17:19:55,428[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:19:55,428[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:19:55,436[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:19:55,436[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 5.137999089404319e-16[0m
|
| 19 |
+
[[36m2026-07-22 17:19:55,792[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:19:58,673[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.59710693359375[0m
|
| 23 |
+
[[36m2026-07-22 17:19:58,927[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:20:00,104[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.004809031889425527[0m
|
| 27 |
+
[[36m2026-07-22 17:20:00,391[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:20:03,434[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:20:07,516[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:20:10,603[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:20:24,644[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:20:25,166[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:20:25,953[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:20:27,562[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:20:27,808[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:20:28,430[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:20:28,430[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:20:28,430[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:20:28,699[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:20:29,429[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:20:30,982[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:20:31,211[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:20:32,139[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:20:32,139[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:20:32,139[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:20:32,139[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:20:32,140[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.4534620375113887[0m
|
| 68 |
+
[[36m2026-07-22 17:20:32,692[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:20:32,700[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 14.22it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.93it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.59it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.93it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 14.27it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.98it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.20it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.26it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.72it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.72it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.81it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.91it/s]
|
| 86 |
+
[[36m2026-07-22 17:20:34,514[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:20:34,521[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:20:34,522[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -97.43273064365043[0m
|
| 89 |
+
[[36m2026-07-22 17:20:34,869[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:20:35,961[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.29292188014521775[0m
|
logs/eval_IdkNLL_lr4e-05_alpha2_epoch10.log
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.46it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 11.58it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 11.87it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 12.01it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 12.23it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 11.96it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 11.72it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 11.59it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 11.88it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 11.92it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 11.96it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:01<00:00, 12.07it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[[36m2026-07-21 07:50:29,562[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr4e-05_alpha2_epoch10[0m
|
| 2 |
+
[[36m2026-07-21 07:50:29,563[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 3 |
+
[[36m2026-07-21 07:50:29,564[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr4e-05_alpha2_epoch10/TOFU_EVAL.json[0m
|
| 4 |
+
[[36m2026-07-21 07:50:29,564[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr4e-05_alpha2_epoch10/TOFU_SUMMARY.json[0m
|
| 5 |
+
[[36m2026-07-21 07:50:30,301[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 6 |
+
|
| 7 |
+
Evaluated 400 examples
|
| 8 |
+
[[36m2026-07-21 07:50:33,815[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 9 |
+
|
| 10 |
+
Evaluated 400 examples
|
| 11 |
+
[[36m2026-07-21 07:50:46,933[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 12 |
+
[[36m2026-07-21 07:50:46,933[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5161614943622759[0m
|
| 13 |
+
[[36m2026-07-21 07:50:46,941[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 14 |
+
[[36m2026-07-21 07:50:46,942[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 15 |
+
[[36m2026-07-21 07:50:46,958[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 16 |
+
[[36m2026-07-21 07:50:46,960[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 7.975373070421785e-17[0m
|
| 17 |
+
[[36m2026-07-21 07:50:47,384[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 18 |
+
|
| 19 |
+
Evaluated 400 examples
|
| 20 |
+
[[36m2026-07-21 07:50:50,051[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.60478271484375[0m
|
| 21 |
+
[[36m2026-07-21 07:50:50,320[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 22 |
+
|
| 23 |
+
Evaluated 400 examples
|
| 24 |
+
[[36m2026-07-21 07:50:52,646[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.011540083002540306[0m
|
| 25 |
+
[[36m2026-07-21 07:50:52,921[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 26 |
+
|
| 27 |
+
Evaluated 400 examples
|
| 28 |
+
[[36m2026-07-21 07:50:55,768[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 29 |
+
|
| 30 |
+
Evaluated 400 examples
|
| 31 |
+
[[36m2026-07-21 07:51:03,352[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 32 |
+
|
| 33 |
+
Evaluated 400 examples
|
| 34 |
+
[[36m2026-07-21 07:51:06,243[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 35 |
+
|
| 36 |
+
Evaluated 400 examples
|
| 37 |
+
[[36m2026-07-21 07:51:19,046[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 38 |
+
[[36m2026-07-21 07:51:19,411[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 39 |
+
|
| 40 |
+
Evaluated 100 examples
|
| 41 |
+
[[36m2026-07-21 07:51:20,232[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 42 |
+
|
| 43 |
+
Evaluated 100 examples
|
| 44 |
+
[[36m2026-07-21 07:51:21,730[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 45 |
+
[[36m2026-07-21 07:51:22,002[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 46 |
+
|
| 47 |
+
Evaluated 100 examples
|
| 48 |
+
[[36m2026-07-21 07:51:23,537[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 49 |
+
[[36m2026-07-21 07:51:23,537[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 50 |
+
[[36m2026-07-21 07:51:23,537[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 51 |
+
[[36m2026-07-21 07:51:23,816[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 52 |
+
|
| 53 |
+
Evaluated 117 examples
|
| 54 |
+
[[36m2026-07-21 07:51:24,530[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 55 |
+
|
| 56 |
+
Evaluated 117 examples
|
| 57 |
+
[[36m2026-07-21 07:51:25,999[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 58 |
+
[[36m2026-07-21 07:51:26,277[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 59 |
+
|
| 60 |
+
Evaluated 117 examples
|
| 61 |
+
[[36m2026-07-21 07:51:28,159[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 62 |
+
[[36m2026-07-21 07:51:28,159[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 63 |
+
[[36m2026-07-21 07:51:28,159[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 64 |
+
[[36m2026-07-21 07:51:28,160[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 65 |
+
[[36m2026-07-21 07:51:28,161[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5427105442935864[0m
|
| 66 |
+
[[36m2026-07-21 07:51:28,731[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 67 |
+
[[36m2026-07-21 07:51:28,746[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 68 |
+
|
| 69 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 70 |
15%|█▌ | 2/13 [00:00<00:00, 13.46it/s]
|
| 71 |
31%|███ | 4/13 [00:00<00:00, 11.58it/s]
|
| 72 |
46%|████▌ | 6/13 [00:00<00:00, 11.87it/s]
|
| 73 |
62%|██████▏ | 8/13 [00:00<00:00, 12.01it/s]
|
| 74 |
77%|███████▋ | 10/13 [00:00<00:00, 12.23it/s]
|
| 75 |
92%|█████████▏| 12/13 [00:00<00:00, 11.96it/s]
|
| 76 |
+
|
| 77 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 78 |
15%|█▌ | 2/13 [00:00<00:00, 11.72it/s]
|
| 79 |
31%|███ | 4/13 [00:00<00:00, 11.59it/s]
|
| 80 |
46%|████▌ | 6/13 [00:00<00:00, 11.88it/s]
|
| 81 |
62%|██████▏ | 8/13 [00:00<00:00, 11.92it/s]
|
| 82 |
77%|███████▋ | 10/13 [00:00<00:00, 11.96it/s]
|
| 83 |
92%|█████████▏| 12/13 [00:01<00:00, 12.07it/s]
|
| 84 |
+
[[36m2026-07-21 07:51:30,839[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 85 |
+
[[36m2026-07-21 07:51:30,853[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 86 |
+
[[36m2026-07-21 07:51:30,853[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -97.62304871776266[0m
|
| 87 |
+
[[36m2026-07-21 07:51:31,136[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 88 |
+
|
| 89 |
+
Evaluated 400 examples
|
| 90 |
+
[[36m2026-07-21 07:51:32,345[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.31056017504923966[0m
|
logs/eval_IdkNLL_lr4e-05_alpha2_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.73it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.47it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.86it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 12.91it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.44it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.21it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 12.65it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.58it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 12.97it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 12.99it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.15it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.37it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:12:44,681[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr4e-05_alpha2_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:12:44,682[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:12:44,682[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr4e-05_alpha2_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:12:44,682[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr4e-05_alpha2_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:12:45,331[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:12:48,779[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:13:03,234[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:13:03,235[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5212455071811525[0m
|
| 15 |
+
[[36m2026-07-22 17:13:03,239[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:13:03,239[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:13:03,246[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:13:03,247[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 2.776495661774257e-16[0m
|
| 19 |
+
[[36m2026-07-22 17:13:03,668[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:13:06,556[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.6036669921875[0m
|
| 23 |
+
[[36m2026-07-22 17:13:06,861[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:13:08,626[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.013236566883255176[0m
|
| 27 |
+
[[36m2026-07-22 17:13:08,879[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:13:11,954[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:13:16,013[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:13:19,113[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:13:33,128[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:13:33,495[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:13:34,429[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:13:36,009[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:13:36,383[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:13:36,970[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:13:36,970[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:13:36,970[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:13:37,209[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:13:37,924[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:13:39,480[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:13:39,716[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:13:40,602[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:13:40,602[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:13:40,602[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:13:40,602[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:13:40,603[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5046823269661426[0m
|
| 68 |
+
[[36m2026-07-22 17:13:41,181[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:13:41,191[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.73it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.47it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.86it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 12.91it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.44it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.21it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 12.65it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.58it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 12.97it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 12.99it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.15it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.37it/s]
|
| 86 |
+
[[36m2026-07-22 17:13:43,103[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:13:43,109[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:13:43,109[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -97.46411287927532[0m
|
| 89 |
+
[[36m2026-07-22 17:13:43,360[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:13:44,419[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.3025280251386857[0m
|
logs/eval_IdkNLL_lr4e-05_alpha5_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.42it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.49it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.14it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.47it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.84it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.54it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 12.84it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.02it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.50it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.58it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.66it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.60it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:21:08,971[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr4e-05_alpha5_epoch10[0m
|
| 4 |
+
[[36m2026-07-22 17:21:08,972[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:21:08,972[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr4e-05_alpha5_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:21:08,972[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr4e-05_alpha5_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:21:09,603[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:21:13,107[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:21:27,448[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:21:27,448[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5148526803425937[0m
|
| 15 |
+
[[36m2026-07-22 17:21:27,453[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:21:27,453[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:21:27,461[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:21:27,462[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 7.975373070421785e-17[0m
|
| 19 |
+
[[36m2026-07-22 17:21:27,830[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:21:30,701[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.64310546875[0m
|
| 23 |
+
[[36m2026-07-22 17:21:30,948[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:21:32,628[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.021201092885887098[0m
|
| 27 |
+
[[36m2026-07-22 17:21:32,891[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:21:36,057[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:21:40,127[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:21:43,243[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:21:57,248[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:21:57,591[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:21:58,446[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:22:00,051[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:22:00,291[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:22:01,076[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:22:01,076[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:22:01,076[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:22:01,317[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:22:02,102[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:22:03,652[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:22:04,203[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:22:05,072[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:22:05,072[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:22:05,072[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:22:05,072[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:22:05,072[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5564634064186886[0m
|
| 68 |
+
[[36m2026-07-22 17:22:05,745[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:22:05,753[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.42it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.49it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.14it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.47it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.84it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.54it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 12.84it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.02it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.50it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.58it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.66it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.60it/s]
|
| 86 |
+
[[36m2026-07-22 17:22:07,615[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:22:07,622[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:22:07,622[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.18894128919214[0m
|
| 89 |
+
[[36m2026-07-22 17:22:07,887[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:22:09,012[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.35020620206680336[0m
|
logs/eval_IdkNLL_lr4e-05_alpha5_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.85it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.38it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.86it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.19it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.52it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.23it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 12.97it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.79it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.17it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.20it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.26it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.33it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:19:41,479[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr4e-05_alpha5_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:19:41,480[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:19:41,480[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr4e-05_alpha5_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:19:41,480[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr4e-05_alpha5_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:19:42,115[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:19:45,580[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:19:59,946[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:19:59,946[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5219291566667374[0m
|
| 15 |
+
[[36m2026-07-22 17:19:59,950[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:19:59,951[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:19:59,958[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:19:59,958[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 2.776495661774257e-16[0m
|
| 19 |
+
[[36m2026-07-22 17:20:00,401[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:20:03,273[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.6360546875[0m
|
| 23 |
+
[[36m2026-07-22 17:20:03,601[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:20:05,715[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.027647814287907826[0m
|
| 27 |
+
[[36m2026-07-22 17:20:06,032[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:20:09,059[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:20:13,505[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:20:16,564[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:20:30,617[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:20:31,001[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:20:31,764[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:20:33,331[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:20:33,568[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:20:34,185[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:20:34,185[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:20:34,185[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:20:34,454[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:20:35,184[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:20:36,738[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:20:36,984[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:20:37,988[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:20:37,988[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:20:37,989[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:20:37,989[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:20:37,989[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5306669591063944[0m
|
| 68 |
+
[[36m2026-07-22 17:20:38,496[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:20:38,503[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.85it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.38it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.86it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.19it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.52it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.23it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 12.97it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.79it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.17it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.20it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.26it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.33it/s]
|
| 86 |
+
[[36m2026-07-22 17:20:40,406[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:20:40,413[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:20:40,413[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -98.06948632778128[0m
|
| 89 |
+
[[36m2026-07-22 17:20:40,679[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:20:41,736[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.32577915312947625[0m
|
logs/eval_IdkNLL_lr5e-05_alpha10_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.98it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.66it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.29it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.61it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.98it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.69it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.36it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.24it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.64it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.60it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.69it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.74it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:14:14,089[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr5e-05_alpha10_epoch10[0m
|
| 4 |
+
[[36m2026-07-22 17:14:14,090[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:14:14,090[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr5e-05_alpha10_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:14:14,090[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr5e-05_alpha10_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:14:15,996[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:14:19,560[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:14:33,989[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:14:33,990[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5273756074800511[0m
|
| 15 |
+
[[36m2026-07-22 17:14:33,994[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:14:33,994[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:14:34,001[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:14:34,002[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 3.1508505165833036e-15[0m
|
| 19 |
+
[[36m2026-07-22 17:14:35,705[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:14:38,591[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.60293212890625[0m
|
| 23 |
+
[[36m2026-07-22 17:14:38,962[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:14:41,685[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.03730176358467679[0m
|
| 27 |
+
[[36m2026-07-22 17:14:41,973[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:14:44,992[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:14:49,438[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:14:52,601[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:15:06,648[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:15:06,995[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:15:07,776[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:15:09,342[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:15:09,646[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:15:10,304[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:15:10,305[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:15:10,305[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:15:10,673[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:15:11,397[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:15:12,968[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:15:13,205[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:15:14,155[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:15:14,156[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:15:14,156[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:15:14,156[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:15:14,156[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5491056613804656[0m
|
| 68 |
+
[[36m2026-07-22 17:15:14,780[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:15:14,786[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.98it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.66it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.29it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.61it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.98it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.69it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.36it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.24it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.64it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.60it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.69it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.74it/s]
|
| 86 |
+
[[36m2026-07-22 17:15:16,630[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:15:16,637[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:15:16,637[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -97.63620901012148[0m
|
| 89 |
+
[[36m2026-07-22 17:15:16,891[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:15:17,939[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.28194443899865135[0m
|
logs/eval_IdkNLL_lr5e-05_alpha10_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 14.15it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.86it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.37it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.82it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 14.19it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.90it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.12it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.21it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.67it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.71it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.77it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.88it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:22:37,716[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr5e-05_alpha10_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:22:37,717[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:22:37,717[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr5e-05_alpha10_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:22:37,717[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr5e-05_alpha10_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:22:38,404[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:22:41,911[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:22:56,284[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:22:56,284[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.547315924175031[0m
|
| 15 |
+
[[36m2026-07-22 17:22:56,288[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:22:56,289[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:22:56,296[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:22:56,297[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 5.805666105234786e-14[0m
|
| 19 |
+
[[36m2026-07-22 17:22:56,672[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:22:59,564[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.57861572265625[0m
|
| 23 |
+
[[36m2026-07-22 17:22:59,835[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:23:02,919[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.058624543990867[0m
|
| 27 |
+
[[36m2026-07-22 17:23:03,203[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:23:06,321[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:23:10,719[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:23:13,836[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:23:27,830[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:23:28,169[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:23:28,979[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:23:30,601[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:23:30,907[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:23:31,669[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:23:31,669[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:23:31,669[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:23:31,948[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:23:32,697[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:23:34,259[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:23:34,507[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:23:35,483[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:23:35,483[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:23:35,483[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:23:35,483[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:23:35,483[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5337837919890431[0m
|
| 68 |
+
[[36m2026-07-22 17:23:36,017[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:23:36,023[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 14.15it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.86it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.37it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.82it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 14.19it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.90it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.12it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.21it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.67it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.71it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.77it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.88it/s]
|
| 86 |
+
[[36m2026-07-22 17:23:37,848[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:23:37,854[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:23:37,854[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -97.20647484809678[0m
|
| 89 |
+
[[36m2026-07-22 17:23:38,127[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:23:39,237[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.24044956048845798[0m
|
logs/eval_IdkNLL_lr5e-05_alpha1_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:01, 8.36it/s]
|
| 2 |
23%|██▎ | 3/13 [00:00<00:01, 7.01it/s]
|
| 3 |
31%|███ | 4/13 [00:00<00:01, 6.76it/s]
|
| 4 |
38%|███▊ | 5/13 [00:00<00:01, 6.93it/s]
|
| 5 |
46%|████▌ | 6/13 [00:00<00:01, 6.94it/s]
|
| 6 |
54%|█████▍ | 7/13 [00:00<00:00, 7.24it/s]
|
| 7 |
62%|██████▏ | 8/13 [00:01<00:00, 7.32it/s]
|
| 8 |
69%|██████▉ | 9/13 [00:01<00:00, 7.38it/s]
|
| 9 |
77%|███████▋ | 10/13 [00:01<00:00, 7.31it/s]
|
| 10 |
85%|████████▍ | 11/13 [00:01<00:00, 7.10it/s]
|
| 11 |
92%|█████████▏| 12/13 [00:01<00:00, 6.93it/s]
|
|
|
|
| 12 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 13 |
8%|▊ | 1/13 [00:00<00:01, 6.33it/s]
|
| 14 |
15%|█▌ | 2/13 [00:00<00:01, 6.63it/s]
|
| 15 |
23%|██▎ | 3/13 [00:00<00:01, 6.62it/s]
|
| 16 |
31%|███ | 4/13 [00:00<00:01, 6.95it/s]
|
| 17 |
38%|███▊ | 5/13 [00:00<00:01, 6.75it/s]
|
| 18 |
46%|████▌ | 6/13 [00:00<00:00, 7.25it/s]
|
| 19 |
54%|█████▍ | 7/13 [00:00<00:00, 7.66it/s]
|
| 20 |
62%|██████▏ | 8/13 [00:01<00:00, 7.20it/s]
|
| 21 |
69%|██████▉ | 9/13 [00:01<00:00, 6.96it/s]
|
| 22 |
77%|███████▋ | 10/13 [00:01<00:00, 7.09it/s]
|
| 23 |
85%|████████▍ | 11/13 [00:01<00:00, 7.07it/s]
|
| 24 |
92%|█████████▏| 12/13 [00:01<00:00, 7.05it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-20 14:47:52,739[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr5e-05_alpha1_epoch10[0m
|
| 4 |
+
[[36m2026-07-20 14:47:52,741[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-20 14:47:52,741[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr5e-05_alpha1_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-20 14:47:52,741[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr5e-05_alpha1_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-20 14:47:53,675[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-20 14:47:58,571[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-20 14:48:12,969[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-20 14:48:12,971[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5344372644932901[0m
|
| 15 |
+
[[36m2026-07-20 14:48:12,982[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-20 14:48:12,983[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-20 14:48:13,003[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-20 14:48:13,005[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 5.805666105234786e-14[0m
|
| 19 |
+
[[36m2026-07-20 14:48:13,443[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-20 14:48:16,226[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.52810302734375[0m
|
| 23 |
+
[[36m2026-07-20 14:48:16,508[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-20 14:48:18,593[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.007173440310293194[0m
|
| 27 |
+
[[36m2026-07-20 14:48:18,875[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-20 14:48:22,002[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-20 14:48:34,885[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-20 14:48:40,133[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-20 14:49:01,380[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-20 14:49:01,879[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-20 14:49:03,105[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-20 14:49:05,511[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-20 14:49:05,938[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-20 14:49:07,786[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-20 14:49:07,787[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-20 14:49:07,787[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-20 14:49:08,175[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-20 14:49:09,421[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-20 14:49:11,610[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-20 14:49:12,016[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-20 14:49:13,997[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-20 14:49:13,997[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-20 14:49:13,997[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-20 14:49:13,997[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-20 14:49:13,998[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.4970728934935518[0m
|
| 68 |
+
[[36m2026-07-20 14:49:14,788[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-20 14:49:14,803[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:01, 8.36it/s]
|
| 73 |
23%|██▎ | 3/13 [00:00<00:01, 7.01it/s]
|
| 74 |
31%|███ | 4/13 [00:00<00:01, 6.76it/s]
|
| 75 |
38%|███▊ | 5/13 [00:00<00:01, 6.93it/s]
|
| 76 |
46%|████▌ | 6/13 [00:00<00:01, 6.94it/s]
|
| 77 |
54%|█████▍ | 7/13 [00:00<00:00, 7.24it/s]
|
| 78 |
62%|██████▏ | 8/13 [00:01<00:00, 7.32it/s]
|
| 79 |
69%|██████▉ | 9/13 [00:01<00:00, 7.38it/s]
|
| 80 |
77%|███████▋ | 10/13 [00:01<00:00, 7.31it/s]
|
| 81 |
85%|████████▍ | 11/13 [00:01<00:00, 7.10it/s]
|
| 82 |
92%|█████████▏| 12/13 [00:01<00:00, 6.93it/s]
|
| 83 |
+
|
| 84 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 85 |
8%|▊ | 1/13 [00:00<00:01, 6.33it/s]
|
| 86 |
15%|█▌ | 2/13 [00:00<00:01, 6.63it/s]
|
| 87 |
23%|██▎ | 3/13 [00:00<00:01, 6.62it/s]
|
| 88 |
31%|███ | 4/13 [00:00<00:01, 6.95it/s]
|
| 89 |
38%|███▊ | 5/13 [00:00<00:01, 6.75it/s]
|
| 90 |
46%|████▌ | 6/13 [00:00<00:00, 7.25it/s]
|
| 91 |
54%|█████▍ | 7/13 [00:00<00:00, 7.66it/s]
|
| 92 |
62%|██████▏ | 8/13 [00:01<00:00, 7.20it/s]
|
| 93 |
69%|██████▉ | 9/13 [00:01<00:00, 6.96it/s]
|
| 94 |
77%|███████▋ | 10/13 [00:01<00:00, 7.09it/s]
|
| 95 |
85%|████████▍ | 11/13 [00:01<00:00, 7.07it/s]
|
| 96 |
92%|█████████▏| 12/13 [00:01<00:00, 7.05it/s]
|
| 97 |
+
[[36m2026-07-20 14:49:18,308[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 98 |
+
[[36m2026-07-20 14:49:18,324[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 99 |
+
[[36m2026-07-20 14:49:18,325[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -96.03369033288922[0m
|
| 100 |
+
[[36m2026-07-20 14:49:18,756[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 101 |
+
|
| 102 |
+
Evaluated 400 examples
|
| 103 |
+
[[36m2026-07-20 14:49:21,771[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.2364795768799366[0m
|
logs/eval_IdkNLL_lr5e-05_alpha1_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.16it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.08it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.67it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 12.85it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.16it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 12.98it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 12.64it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.31it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 12.64it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 12.70it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 12.80it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 12.90it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:21:11,260[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr5e-05_alpha1_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:21:11,261[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:21:11,261[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr5e-05_alpha1_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:21:11,261[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr5e-05_alpha1_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:21:11,958[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:21:15,651[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:21:30,016[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:21:30,016[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5453775468072528[0m
|
| 15 |
+
[[36m2026-07-22 17:21:30,020[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:21:30,021[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:21:30,028[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:21:30,028[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 1.02724452984523e-14[0m
|
| 19 |
+
[[36m2026-07-22 17:21:30,372[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:21:33,232[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.52435302734375[0m
|
| 23 |
+
[[36m2026-07-22 17:21:33,483[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:21:34,694[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.005011347532056894[0m
|
| 27 |
+
[[36m2026-07-22 17:21:34,949[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:21:38,069[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:21:42,050[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:21:45,166[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:21:59,124[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:21:59,478[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:22:00,293[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:22:01,894[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:22:02,370[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:22:02,962[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:22:02,962[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:22:02,962[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:22:03,262[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:22:04,244[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:22:05,797[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:22:06,106[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:22:07,135[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:22:07,136[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:22:07,136[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:22:07,136[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:22:07,136[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.4558601581687516[0m
|
| 68 |
+
[[36m2026-07-22 17:22:07,675[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:22:07,686[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.16it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.08it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.67it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 12.85it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.16it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 12.98it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 12.64it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.31it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 12.64it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 12.70it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 12.80it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 12.90it/s]
|
| 86 |
+
[[36m2026-07-22 17:22:09,647[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:22:09,654[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:22:09,654[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -95.85248323040999[0m
|
| 89 |
+
[[36m2026-07-22 17:22:10,017[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:22:11,096[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.2176950903433096[0m
|
logs/eval_IdkNLL_lr5e-05_alpha2_epoch10.log
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 12.94it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 11.05it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 11.31it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 11.44it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 11.70it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:01<00:00, 11.44it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 11.21it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 11.06it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 11.36it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 11.42it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 11.42it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:01<00:00, 11.54it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[[36m2026-07-21 07:50:33,476[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr5e-05_alpha2_epoch10[0m
|
| 2 |
+
[[36m2026-07-21 07:50:33,477[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 3 |
+
[[36m2026-07-21 07:50:33,477[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr5e-05_alpha2_epoch10/TOFU_EVAL.json[0m
|
| 4 |
+
[[36m2026-07-21 07:50:33,477[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr5e-05_alpha2_epoch10/TOFU_SUMMARY.json[0m
|
| 5 |
+
[[36m2026-07-21 07:50:34,111[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 6 |
+
|
| 7 |
+
Evaluated 400 examples
|
| 8 |
+
[[36m2026-07-21 07:50:37,648[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 9 |
+
|
| 10 |
+
Evaluated 400 examples
|
| 11 |
+
[[36m2026-07-21 07:50:50,850[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 12 |
+
[[36m2026-07-21 07:50:50,851[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5321104903862071[0m
|
| 13 |
+
[[36m2026-07-21 07:50:50,859[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 14 |
+
[[36m2026-07-21 07:50:50,860[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 15 |
+
[[36m2026-07-21 07:50:50,875[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 16 |
+
[[36m2026-07-21 07:50:50,877[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 5.7046977773929285e-15[0m
|
| 17 |
+
[[36m2026-07-21 07:50:51,263[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 18 |
+
|
| 19 |
+
Evaluated 400 examples
|
| 20 |
+
[[36m2026-07-21 07:50:53,950[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.53903564453125[0m
|
| 21 |
+
[[36m2026-07-21 07:50:54,250[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 22 |
+
|
| 23 |
+
Evaluated 400 examples
|
| 24 |
+
[[36m2026-07-21 07:50:56,929[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.014969829593524582[0m
|
| 25 |
+
[[36m2026-07-21 07:50:57,323[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 26 |
+
|
| 27 |
+
Evaluated 400 examples
|
| 28 |
+
[[36m2026-07-21 07:51:00,223[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 29 |
+
|
| 30 |
+
Evaluated 400 examples
|
| 31 |
+
[[36m2026-07-21 07:51:08,233[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 32 |
+
|
| 33 |
+
Evaluated 400 examples
|
| 34 |
+
[[36m2026-07-21 07:51:11,167[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 35 |
+
|
| 36 |
+
Evaluated 400 examples
|
| 37 |
+
[[36m2026-07-21 07:51:24,076[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 38 |
+
[[36m2026-07-21 07:51:24,446[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 39 |
+
|
| 40 |
+
Evaluated 100 examples
|
| 41 |
+
[[36m2026-07-21 07:51:25,259[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 42 |
+
|
| 43 |
+
Evaluated 100 examples
|
| 44 |
+
[[36m2026-07-21 07:51:26,765[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 45 |
+
[[36m2026-07-21 07:51:27,027[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 46 |
+
|
| 47 |
+
Evaluated 100 examples
|
| 48 |
+
[[36m2026-07-21 07:51:28,381[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 49 |
+
[[36m2026-07-21 07:51:28,381[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 50 |
+
[[36m2026-07-21 07:51:28,382[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 51 |
+
[[36m2026-07-21 07:51:28,642[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 52 |
+
|
| 53 |
+
Evaluated 117 examples
|
| 54 |
+
[[36m2026-07-21 07:51:29,382[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 55 |
+
|
| 56 |
+
Evaluated 117 examples
|
| 57 |
+
[[36m2026-07-21 07:51:30,863[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 58 |
+
[[36m2026-07-21 07:51:31,124[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 59 |
+
|
| 60 |
+
Evaluated 117 examples
|
| 61 |
+
[[36m2026-07-21 07:51:32,832[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 62 |
+
[[36m2026-07-21 07:51:32,833[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 63 |
+
[[36m2026-07-21 07:51:32,833[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 64 |
+
[[36m2026-07-21 07:51:32,833[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 65 |
+
[[36m2026-07-21 07:51:32,834[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5362012798873454[0m
|
| 66 |
+
[[36m2026-07-21 07:51:33,451[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 67 |
+
[[36m2026-07-21 07:51:33,466[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 68 |
+
|
| 69 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 70 |
15%|█▌ | 2/13 [00:00<00:00, 12.94it/s]
|
| 71 |
31%|███ | 4/13 [00:00<00:00, 11.05it/s]
|
| 72 |
46%|████▌ | 6/13 [00:00<00:00, 11.31it/s]
|
| 73 |
62%|██████▏ | 8/13 [00:00<00:00, 11.44it/s]
|
| 74 |
77%|███████▋ | 10/13 [00:00<00:00, 11.70it/s]
|
| 75 |
92%|█████████▏| 12/13 [00:01<00:00, 11.44it/s]
|
| 76 |
+
|
| 77 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 78 |
15%|█▌ | 2/13 [00:00<00:00, 11.21it/s]
|
| 79 |
31%|███ | 4/13 [00:00<00:00, 11.06it/s]
|
| 80 |
46%|████▌ | 6/13 [00:00<00:00, 11.36it/s]
|
| 81 |
62%|██████▏ | 8/13 [00:00<00:00, 11.42it/s]
|
| 82 |
77%|███████▋ | 10/13 [00:00<00:00, 11.42it/s]
|
| 83 |
92%|█████████▏| 12/13 [00:01<00:00, 11.54it/s]
|
| 84 |
+
[[36m2026-07-21 07:51:35,657[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 85 |
+
[[36m2026-07-21 07:51:35,671[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 86 |
+
[[36m2026-07-21 07:51:35,671[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -96.29993317061005[0m
|
| 87 |
+
[[36m2026-07-21 07:51:35,984[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 88 |
+
|
| 89 |
+
Evaluated 400 examples
|
| 90 |
+
[[36m2026-07-21 07:51:37,245[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.2512007516134813[0m
|
logs/eval_IdkNLL_lr5e-05_alpha2_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.47it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.17it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.77it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.24it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.70it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.49it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.39it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.22it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.64it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.66it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.72it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.81it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:15:45,417[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr5e-05_alpha2_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:15:45,418[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:15:45,418[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr5e-05_alpha2_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:15:45,418[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr5e-05_alpha2_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:15:46,065[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:15:49,550[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:16:03,998[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:16:03,999[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5486427139963626[0m
|
| 15 |
+
[[36m2026-07-22 17:16:04,003[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:16:04,004[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:16:04,012[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:16:04,013[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 1.0229984068189572e-13[0m
|
| 19 |
+
[[36m2026-07-22 17:16:04,466[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:16:07,333[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.52466552734375[0m
|
| 23 |
+
[[36m2026-07-22 17:16:08,272[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:16:09,749[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.006607161672815781[0m
|
| 27 |
+
[[36m2026-07-22 17:16:10,409[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:16:13,537[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:16:18,303[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:16:22,394[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:16:36,399[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:16:36,751[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:16:37,615[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:16:39,184[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:16:39,707[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:16:40,405[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:16:40,405[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:16:40,405[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:16:40,818[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:16:41,572[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:16:43,142[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:16:43,411[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:16:44,456[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:16:44,456[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:16:44,456[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:16:44,456[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:16:44,457[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5057544347432658[0m
|
| 68 |
+
[[36m2026-07-22 17:16:44,993[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:16:45,002[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.47it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.17it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.77it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.24it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.70it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.49it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.39it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.22it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.64it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.66it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.72it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.81it/s]
|
| 86 |
+
[[36m2026-07-22 17:16:46,860[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:16:46,867[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:16:46,867[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -95.55688281742715[0m
|
| 89 |
+
[[36m2026-07-22 17:17:12,006[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:17:13,056[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.21479457793325007[0m
|
logs/eval_IdkNLL_lr5e-05_alpha5_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
8%|▊ | 1/13 [00:00<00:01, 8.11it/s]
|
| 2 |
15%|█▌ | 2/13 [00:00<00:01, 7.43it/s]
|
| 3 |
23%|██▎ | 3/13 [00:00<00:01, 6.73it/s]
|
| 4 |
31%|███ | 4/13 [00:00<00:01, 5.91it/s]
|
| 5 |
38%|███▊ | 5/13 [00:00<00:01, 6.08it/s]
|
| 6 |
46%|████▌ | 6/13 [00:00<00:01, 6.33it/s]
|
| 7 |
54%|█████▍ | 7/13 [00:01<00:01, 5.86it/s]
|
| 8 |
62%|██████▏ | 8/13 [00:01<00:00, 6.08it/s]
|
| 9 |
69%|██████▉ | 9/13 [00:01<00:00, 6.35it/s]
|
| 10 |
77%|███████▋ | 10/13 [00:01<00:00, 6.28it/s]
|
| 11 |
85%|████████▍ | 11/13 [00:01<00:00, 6.11it/s]
|
|
|
|
| 12 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 13 |
8%|▊ | 1/13 [00:00<00:01, 6.09it/s]
|
| 14 |
15%|█▌ | 2/13 [00:00<00:01, 6.87it/s]
|
| 15 |
23%|██▎ | 3/13 [00:00<00:01, 7.10it/s]
|
| 16 |
31%|███ | 4/13 [00:00<00:01, 6.11it/s]
|
| 17 |
38%|███▊ | 5/13 [00:00<00:01, 5.69it/s]
|
| 18 |
46%|████▌ | 6/13 [00:00<00:01, 6.58it/s]
|
| 19 |
54%|█████▍ | 7/13 [00:01<00:00, 6.97it/s]
|
| 20 |
62%|██████▏ | 8/13 [00:01<00:00, 6.89it/s]
|
| 21 |
69%|██████▉ | 9/13 [00:01<00:00, 6.67it/s]
|
| 22 |
77%|███████▋ | 10/13 [00:01<00:00, 6.77it/s]
|
| 23 |
85%|████████▍ | 11/13 [00:01<00:00, 6.71it/s]
|
| 24 |
92%|█████████▏| 12/13 [00:01<00:00, 6.59it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-20 14:51:46,440[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr5e-05_alpha5_epoch10[0m
|
| 4 |
+
[[36m2026-07-20 14:51:46,441[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-20 14:51:46,441[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr5e-05_alpha5_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-20 14:51:46,441[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr5e-05_alpha5_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-20 14:51:47,120[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-20 14:51:53,582[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-20 14:52:17,272[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-20 14:52:17,274[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5298605825153447[0m
|
| 15 |
+
[[36m2026-07-20 14:52:17,285[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-20 14:52:17,286[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-20 14:52:17,306[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-20 14:52:17,308[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 5.7046977773929285e-15[0m
|
| 19 |
+
[[36m2026-07-20 14:52:18,007[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-20 14:52:21,870[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.56814697265625[0m
|
| 23 |
+
[[36m2026-07-20 14:52:22,414[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-20 14:52:27,498[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.017253889918628383[0m
|
| 27 |
+
[[36m2026-07-20 14:52:28,032[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-20 14:52:31,912[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-20 14:52:45,111[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-20 14:52:49,024[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-20 14:53:13,333[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-20 14:53:13,726[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-20 14:53:15,265[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-20 14:53:18,598[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-20 14:53:19,078[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-20 14:53:22,122[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-20 14:53:22,122[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-20 14:53:22,122[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-20 14:53:22,584[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-20 14:53:23,694[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-20 14:53:26,169[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-20 14:53:26,503[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-20 14:53:29,788[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-20 14:53:29,789[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-20 14:53:29,789[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-20 14:53:29,789[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-20 14:53:29,790[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5547274672864976[0m
|
| 68 |
+
[[36m2026-07-20 14:53:30,439[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-20 14:53:30,455[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
8%|▊ | 1/13 [00:00<00:01, 8.11it/s]
|
| 73 |
15%|█▌ | 2/13 [00:00<00:01, 7.43it/s]
|
| 74 |
23%|██▎ | 3/13 [00:00<00:01, 6.73it/s]
|
| 75 |
31%|███ | 4/13 [00:00<00:01, 5.91it/s]
|
| 76 |
38%|███▊ | 5/13 [00:00<00:01, 6.08it/s]
|
| 77 |
46%|████▌ | 6/13 [00:00<00:01, 6.33it/s]
|
| 78 |
54%|█████▍ | 7/13 [00:01<00:01, 5.86it/s]
|
| 79 |
62%|██████▏ | 8/13 [00:01<00:00, 6.08it/s]
|
| 80 |
69%|██████▉ | 9/13 [00:01<00:00, 6.35it/s]
|
| 81 |
77%|███████▋ | 10/13 [00:01<00:00, 6.28it/s]
|
| 82 |
85%|████████▍ | 11/13 [00:01<00:00, 6.11it/s]
|
| 83 |
+
|
| 84 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 85 |
8%|▊ | 1/13 [00:00<00:01, 6.09it/s]
|
| 86 |
15%|█▌ | 2/13 [00:00<00:01, 6.87it/s]
|
| 87 |
23%|██▎ | 3/13 [00:00<00:01, 7.10it/s]
|
| 88 |
31%|███ | 4/13 [00:00<00:01, 6.11it/s]
|
| 89 |
38%|███▊ | 5/13 [00:00<00:01, 5.69it/s]
|
| 90 |
46%|████▌ | 6/13 [00:00<00:01, 6.58it/s]
|
| 91 |
54%|█████▍ | 7/13 [00:01<00:00, 6.97it/s]
|
| 92 |
62%|██████▏ | 8/13 [00:01<00:00, 6.89it/s]
|
| 93 |
69%|██████▉ | 9/13 [00:01<00:00, 6.67it/s]
|
| 94 |
77%|███████▋ | 10/13 [00:01<00:00, 6.77it/s]
|
| 95 |
85%|████████▍ | 11/13 [00:01<00:00, 6.71it/s]
|
| 96 |
92%|█████████▏| 12/13 [00:01<00:00, 6.59it/s]
|
| 97 |
+
[[36m2026-07-20 14:53:34,269[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 98 |
+
[[36m2026-07-20 14:53:34,287[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 99 |
+
[[36m2026-07-20 14:53:34,287[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -96.96604643000288[0m
|
| 100 |
+
[[36m2026-07-20 14:53:34,581[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 101 |
+
|
| 102 |
+
Evaluated 400 examples
|
| 103 |
+
[[36m2026-07-20 14:53:38,773[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.25835151129389794[0m
|
logs/eval_IdkNLL_lr5e-05_alpha5_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 14.16it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.88it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.50it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.88it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 14.29it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 14.01it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.74it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.58it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.98it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.97it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 14.02it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 14.09it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:24:07,257[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/IdkNLL_lr5e-05_alpha5_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:24:07,258[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:24:07,258[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/IdkNLL_lr5e-05_alpha5_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:24:07,258[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/IdkNLL_lr5e-05_alpha5_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:24:08,026[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:24:11,521[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:24:25,908[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:24:25,909[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.5429714470601403[0m
|
| 15 |
+
[[36m2026-07-22 17:24:25,913[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:24:25,913[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:24:25,920[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:24:25,921[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 1.02724452984523e-14[0m
|
| 19 |
+
[[36m2026-07-22 17:24:26,329[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:24:29,201[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.5495166015625[0m
|
| 23 |
+
[[36m2026-07-22 17:24:29,555[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:24:31,590[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.020769793233637722[0m
|
| 27 |
+
[[36m2026-07-22 17:24:31,903[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:24:35,019[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:24:39,359[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:24:42,581[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:24:56,607[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:24:57,013[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:24:57,894[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:24:59,467[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:24:59,768[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:25:00,436[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:25:00,436[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:25:00,436[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:25:00,793[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:25:01,587[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:25:03,154[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:25:03,459[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:25:04,522[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:25:04,522[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:25:04,522[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:25:04,522[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:25:04,522[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.5284420115998628[0m
|
| 68 |
+
[[36m2026-07-22 17:25:05,183[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:25:05,190[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 14.16it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.88it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.50it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.88it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 14.29it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 14.01it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.74it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.58it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.98it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.97it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 14.02it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 14.09it/s]
|
| 86 |
+
[[36m2026-07-22 17:25:06,990[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:25:06,997[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:25:06,997[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -96.58287945632476[0m
|
| 89 |
+
[[36m2026-07-22 17:25:07,344[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:25:08,393[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.23153243063984383[0m
|
logs/eval_NPO_lr1e-05_beta0.05_alpha1_epoch10.log
ADDED
|
@@ -0,0 +1,124 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.28it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 11.41it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 11.70it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 11.86it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 12.10it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:01<00:00, 11.86it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 11.50it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 11.38it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 11.70it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 11.76it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 11.77it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:01<00:00, 11.89it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[[36m2026-07-21 07:51:54,514[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/NPO_lr1e-05_beta0.05_alpha1_epoch10[0m
|
| 2 |
+
[[36m2026-07-21 07:51:54,516[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 3 |
+
[[36m2026-07-21 07:51:54,516[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/NPO_lr1e-05_beta0.05_alpha1_epoch10/TOFU_EVAL.json[0m
|
| 4 |
+
[[36m2026-07-21 07:51:54,516[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/NPO_lr1e-05_beta0.05_alpha1_epoch10/TOFU_SUMMARY.json[0m
|
| 5 |
+
[[36m2026-07-21 07:51:55,163[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 6 |
+
|
| 7 |
+
Evaluated 400 examples
|
| 8 |
+
[[36m2026-07-21 07:51:58,724[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 9 |
+
|
| 10 |
+
Evaluated 400 examples
|
| 11 |
+
[[36m2026-07-21 07:52:11,870[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 12 |
+
[[36m2026-07-21 07:52:11,871[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.6877360924998683[0m
|
| 13 |
+
[[36m2026-07-21 07:52:11,879[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 14 |
+
[[36m2026-07-21 07:52:11,881[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 15 |
+
[[36m2026-07-21 07:52:11,897[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 16 |
+
[[36m2026-07-21 07:52:11,899[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 0.0013468720321565051[0m
|
| 17 |
+
[[36m2026-07-21 07:52:12,292[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 18 |
+
|
| 19 |
+
Evaluated 400 examples
|
| 20 |
+
[[36m2026-07-21 07:52:14,965[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.0645626974105835[0m
|
| 21 |
+
[[36m2026-07-21 07:52:15,247[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 22 |
+
|
| 23 |
+
Evaluated 400 examples
|
| 24 |
+
[[36m2026-07-21 07:52:20,627[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.13478649070874613[0m
|
| 25 |
+
[[36m2026-07-21 07:52:20,930[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 26 |
+
|
| 27 |
+
Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub
|
| 28 |
+
Evaluated 400 examples
|
| 29 |
+
[[36m2026-07-21 07:52:23,559[0m][[34mdatasets.load[0m][[33mWARNING[0m] - Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub[0m
|
| 30 |
+
Found the latest cached dataset configuration 'retain_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/retain_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:22:39 2026).
|
| 31 |
+
[[36m2026-07-21 07:52:23,561[0m][[34mdatasets.packaged_modules.cache.cache[0m][[33mWARNING[0m] - Found the latest cached dataset configuration 'retain_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/retain_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:22:39 2026).[0m
|
| 32 |
+
[[36m2026-07-21 07:52:23,569[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 33 |
+
|
| 34 |
+
Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-21 07:52:26,971[0m][[34mdatasets.load[0m][[33mWARNING[0m] - Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub[0m
|
| 37 |
+
Found the latest cached dataset configuration 'retain_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/retain_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:22:39 2026).
|
| 38 |
+
[[36m2026-07-21 07:52:26,974[0m][[34mdatasets.packaged_modules.cache.cache[0m][[33mWARNING[0m] - Found the latest cached dataset configuration 'retain_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/retain_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:22:39 2026).[0m
|
| 39 |
+
[[36m2026-07-21 07:52:26,979[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 40 |
+
|
| 41 |
+
Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub
|
| 42 |
+
Evaluated 400 examples
|
| 43 |
+
[[36m2026-07-21 07:52:29,666[0m][[34mdatasets.load[0m][[33mWARNING[0m] - Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub[0m
|
| 44 |
+
Found the latest cached dataset configuration 'retain_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/retain_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:22:39 2026).
|
| 45 |
+
[[36m2026-07-21 07:52:29,668[0m][[34mdatasets.packaged_modules.cache.cache[0m][[33mWARNING[0m] - Found the latest cached dataset configuration 'retain_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/retain_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:22:39 2026).[0m
|
| 46 |
+
[[36m2026-07-21 07:52:29,673[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 47 |
+
|
| 48 |
+
Evaluated 400 examples
|
| 49 |
+
[[36m2026-07-21 07:52:42,488[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 50 |
+
Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub
|
| 51 |
+
[[36m2026-07-21 07:52:42,526[0m][[34mdatasets.load[0m][[33mWARNING[0m] - Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub[0m
|
| 52 |
+
Found the latest cached dataset configuration 'real_authors_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/real_authors_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:07 2026).
|
| 53 |
+
[[36m2026-07-21 07:52:42,527[0m][[34mdatasets.packaged_modules.cache.cache[0m][[33mWARNING[0m] - Found the latest cached dataset configuration 'real_authors_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/real_authors_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:07 2026).[0m
|
| 54 |
+
[[36m2026-07-21 07:52:42,533[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 55 |
+
|
| 56 |
+
Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub
|
| 57 |
+
Evaluated 100 examples
|
| 58 |
+
[[36m2026-07-21 07:52:43,070[0m][[34mdatasets.load[0m][[33mWARNING[0m] - Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub[0m
|
| 59 |
+
Found the latest cached dataset configuration 'real_authors_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/real_authors_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:07 2026).
|
| 60 |
+
[[36m2026-07-21 07:52:43,072[0m][[34mdatasets.packaged_modules.cache.cache[0m][[33mWARNING[0m] - Found the latest cached dataset configuration 'real_authors_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/real_authors_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:07 2026).[0m
|
| 61 |
+
[[36m2026-07-21 07:52:43,078[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 62 |
+
|
| 63 |
+
Evaluated 100 examples
|
| 64 |
+
[[36m2026-07-21 07:52:44,577[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 65 |
+
Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub
|
| 66 |
+
[[36m2026-07-21 07:52:44,612[0m][[34mdatasets.load[0m][[33mWARNING[0m] - Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub[0m
|
| 67 |
+
Found the latest cached dataset configuration 'real_authors_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/real_authors_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:07 2026).
|
| 68 |
+
[[36m2026-07-21 07:52:44,613[0m][[34mdatasets.packaged_modules.cache.cache[0m][[33mWARNING[0m] - Found the latest cached dataset configuration 'real_authors_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/real_authors_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:07 2026).[0m
|
| 69 |
+
[[36m2026-07-21 07:52:44,617[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 70 |
+
|
| 71 |
+
Evaluated 100 examples
|
| 72 |
+
[[36m2026-07-21 07:52:45,367[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 73 |
+
[[36m2026-07-21 07:52:45,368[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 74 |
+
[[36m2026-07-21 07:52:45,368[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 75 |
+
Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub
|
| 76 |
+
[[36m2026-07-21 07:52:45,407[0m][[34mdatasets.load[0m][[33mWARNING[0m] - Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub[0m
|
| 77 |
+
Found the latest cached dataset configuration 'world_facts_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/world_facts_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:12 2026).
|
| 78 |
+
[[36m2026-07-21 07:52:45,410[0m][[34mdatasets.packaged_modules.cache.cache[0m][[33mWARNING[0m] - Found the latest cached dataset configuration 'world_facts_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/world_facts_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:12 2026).[0m
|
| 79 |
+
[[36m2026-07-21 07:52:45,414[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 80 |
+
|
| 81 |
+
Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub
|
| 82 |
+
Evaluated 117 examples
|
| 83 |
+
[[36m2026-07-21 07:52:45,911[0m][[34mdatasets.load[0m][[33mWARNING[0m] - Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub[0m
|
| 84 |
+
Found the latest cached dataset configuration 'world_facts_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/world_facts_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:12 2026).
|
| 85 |
+
[[36m2026-07-21 07:52:45,912[0m][[34mdatasets.packaged_modules.cache.cache[0m][[33mWARNING[0m] - Found the latest cached dataset configuration 'world_facts_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/world_facts_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:12 2026).[0m
|
| 86 |
+
[[36m2026-07-21 07:52:45,915[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 87 |
+
|
| 88 |
+
Evaluated 117 examples
|
| 89 |
+
[[36m2026-07-21 07:52:47,388[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 90 |
+
Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub
|
| 91 |
+
[[36m2026-07-21 07:52:47,423[0m][[34mdatasets.load[0m][[33mWARNING[0m] - Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub[0m
|
| 92 |
+
Found the latest cached dataset configuration 'world_facts_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/world_facts_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:12 2026).
|
| 93 |
+
[[36m2026-07-21 07:52:47,425[0m][[34mdatasets.packaged_modules.cache.cache[0m][[33mWARNING[0m] - Found the latest cached dataset configuration 'world_facts_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/world_facts_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:12 2026).[0m
|
| 94 |
+
[[36m2026-07-21 07:52:47,429[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 95 |
+
|
| 96 |
+
Evaluated 117 examples
|
| 97 |
+
[[36m2026-07-21 07:52:48,411[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 98 |
+
[[36m2026-07-21 07:52:48,411[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 99 |
+
[[36m2026-07-21 07:52:48,411[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 100 |
+
[[36m2026-07-21 07:52:48,411[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 101 |
+
[[36m2026-07-21 07:52:48,412[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.2596283949532122[0m
|
| 102 |
+
Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub
|
| 103 |
+
[[36m2026-07-21 07:52:48,485[0m][[34mdatasets.load[0m][[33mWARNING[0m] - Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub[0m
|
| 104 |
+
Found the latest cached dataset configuration 'forget10_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/forget10_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:17:15 2026).
|
| 105 |
+
[[36m2026-07-21 07:52:48,487[0m][[34mdatasets.packaged_modules.cache.cache[0m][[33mWARNING[0m] - Found the latest cached dataset configuration 'forget10_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/forget10_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:17:15 2026).[0m
|
| 106 |
+
Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub
|
| 107 |
+
[[36m2026-07-21 07:52:48,528[0m][[34mdatasets.load[0m][[33mWARNING[0m] - Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub[0m
|
| 108 |
+
Found the latest cached dataset configuration 'holdout10' at /workspace/hf_cache_ou/locuslab___tofu/holdout10/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:17 2026).
|
| 109 |
+
[[36m2026-07-21 07:52:48,531[0m][[34mdatasets.packaged_modules.cache.cache[0m][[33mWARNING[0m] - Found the latest cached dataset configuration 'holdout10' at /workspace/hf_cache_ou/locuslab___tofu/holdout10/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:23:17 2026).[0m
|
| 110 |
+
[[36m2026-07-21 07:52:48,534[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 111 |
+
[[36m2026-07-21 07:52:48,550[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 112 |
+
|
| 113 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 114 |
15%|█▌ | 2/13 [00:00<00:00, 13.28it/s]
|
| 115 |
31%|███ | 4/13 [00:00<00:00, 11.41it/s]
|
| 116 |
46%|████▌ | 6/13 [00:00<00:00, 11.70it/s]
|
| 117 |
62%|██████▏ | 8/13 [00:00<00:00, 11.86it/s]
|
| 118 |
77%|███████▋ | 10/13 [00:00<00:00, 12.10it/s]
|
| 119 |
92%|█████████▏| 12/13 [00:01<00:00, 11.86it/s]
|
| 120 |
+
|
| 121 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 122 |
15%|█▌ | 2/13 [00:00<00:00, 11.50it/s]
|
| 123 |
31%|███ | 4/13 [00:00<00:00, 11.38it/s]
|
| 124 |
46%|████▌ | 6/13 [00:00<00:00, 11.70it/s]
|
| 125 |
62%|██████▏ | 8/13 [00:00<00:00, 11.76it/s]
|
| 126 |
77%|███████▋ | 10/13 [00:00<00:00, 11.77it/s]
|
| 127 |
92%|█████████▏| 12/13 [00:01<00:00, 11.89it/s]
|
| 128 |
+
[[36m2026-07-21 07:52:50,674[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 129 |
+
[[36m2026-07-21 07:52:50,689[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 130 |
+
[[36m2026-07-21 07:52:50,690[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: 5.302079325334236[0m
|
| 131 |
+
Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub
|
| 132 |
+
[[36m2026-07-21 07:52:50,764[0m][[34mdatasets.load[0m][[33mWARNING[0m] - Using the latest cached version of the dataset since locuslab/TOFU couldn't be found on the Hugging Face Hub[0m
|
| 133 |
+
Found the latest cached dataset configuration 'forget10_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/forget10_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:17:15 2026).
|
| 134 |
+
[[36m2026-07-21 07:52:50,766[0m][[34mdatasets.packaged_modules.cache.cache[0m][[33mWARNING[0m] - Found the latest cached dataset configuration 'forget10_perturbed' at /workspace/hf_cache_ou/locuslab___tofu/forget10_perturbed/0.0.0/324592d84ae4f482ac7249b9285c2ecdb53e3a68 (last modified on Mon Jul 20 13:17:15 2026).[0m
|
| 135 |
+
[[36m2026-07-21 07:52:50,772[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 136 |
+
|
| 137 |
+
Evaluated 400 examples
|
| 138 |
+
[[36m2026-07-21 07:52:52,060[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.06207448107554553[0m
|
logs/eval_NPO_lr1e-05_beta0.05_alpha1_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 14.06it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.71it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.31it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.65it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 14.00it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.71it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.39it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.25it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.66it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.68it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.73it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.82it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:17:36,998[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/NPO_lr1e-05_beta0.05_alpha1_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:17:36,999[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:17:36,999[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/NPO_lr1e-05_beta0.05_alpha1_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:17:36,999[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/NPO_lr1e-05_beta0.05_alpha1_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:17:39,825[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:17:43,343[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:17:57,769[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:17:57,769[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.7040837745521601[0m
|
| 15 |
+
[[36m2026-07-22 17:17:57,774[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:17:57,774[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:17:57,782[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:17:57,783[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 3.596289960343452e-05[0m
|
| 19 |
+
[[36m2026-07-22 17:17:58,166[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:18:01,066[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.07519393920898437[0m
|
| 23 |
+
[[36m2026-07-22 17:18:01,359[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:18:03,037[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.12957986687563722[0m
|
| 27 |
+
[[36m2026-07-22 17:18:03,284[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:18:06,322[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:18:08,633[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:18:11,724[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:18:25,763[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:18:26,158[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:18:26,965[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:18:28,568[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:18:28,825[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:18:29,265[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:18:29,265[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:18:29,265[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:18:29,543[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:18:30,282[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:18:31,851[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:18:32,113[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:18:32,680[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:18:32,680[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:18:32,680[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:18:32,681[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:18:32,681[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.1728273832476991[0m
|
| 68 |
+
[[36m2026-07-22 17:18:33,299[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:18:33,306[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 14.06it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.71it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.31it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.65it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 14.00it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.71it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.39it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.25it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.66it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.68it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.73it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.82it/s]
|
| 86 |
+
[[36m2026-07-22 17:18:35,141[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:18:35,147[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:18:35,147[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -30.390152047070888[0m
|
| 89 |
+
[[36m2026-07-22 17:18:35,437[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:18:36,537[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.057134989491472735[0m
|
logs/eval_NPO_lr1e-05_beta0.05_alpha2_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.68it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.45it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.15it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.34it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.52it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.08it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.39it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.20it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.27it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.21it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.28it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.35it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:25:36,309[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/NPO_lr1e-05_beta0.05_alpha2_epoch10[0m
|
| 4 |
+
[[36m2026-07-22 17:25:36,310[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:25:36,310[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/NPO_lr1e-05_beta0.05_alpha2_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:25:36,310[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/NPO_lr1e-05_beta0.05_alpha2_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:25:37,119[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:25:40,815[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:25:55,219[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:25:55,220[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.6744537705504727[0m
|
| 15 |
+
[[36m2026-07-22 17:25:55,224[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:25:55,224[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:25:55,231[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:25:55,231[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 0.0004368337391825136[0m
|
| 19 |
+
[[36m2026-07-22 17:25:55,685[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:25:58,535[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.08624565601348877[0m
|
| 23 |
+
[[36m2026-07-22 17:25:59,061[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:26:01,360[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.1558186962244796[0m
|
| 27 |
+
[[36m2026-07-22 17:26:01,699[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:26:04,923[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:26:09,171[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:26:12,320[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:26:26,321[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:26:26,774[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:26:27,698[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:26:29,291[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:26:29,600[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:26:30,065[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:26:30,066[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:26:30,066[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:26:30,396[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:26:31,170[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:26:32,724[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:26:33,098[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:26:33,730[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:26:33,730[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:26:33,730[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:26:33,731[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:26:33,731[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.32742401489794026[0m
|
| 68 |
+
[[36m2026-07-22 17:26:34,379[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:26:34,389[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.68it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.45it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.15it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.34it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.52it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.08it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.39it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.20it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.27it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.21it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.28it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.35it/s]
|
| 86 |
+
[[36m2026-07-22 17:26:36,285[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:26:36,292[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:26:36,293[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: 0.9571581865591441[0m
|
| 89 |
+
[[36m2026-07-22 17:26:36,636[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:26:37,722[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.06894511800369839[0m
|
logs/eval_NPO_lr1e-05_beta0.05_alpha2_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.09it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 11.98it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.66it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.06it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.45it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.23it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.03it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 12.85it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.24it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.23it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.23it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.37it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:22:46,720[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/NPO_lr1e-05_beta0.05_alpha2_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:22:46,721[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:22:46,721[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/NPO_lr1e-05_beta0.05_alpha2_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:22:46,721[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/NPO_lr1e-05_beta0.05_alpha2_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:22:47,457[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:22:51,117[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:23:05,434[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:23:05,435[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.6875118993975735[0m
|
| 15 |
+
[[36m2026-07-22 17:23:05,439[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:23:05,439[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:23:05,446[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:23:05,447[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 0.0001305755477065129[0m
|
| 19 |
+
[[36m2026-07-22 17:23:05,865[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:23:08,758[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.09350917816162109[0m
|
| 23 |
+
[[36m2026-07-22 17:23:09,116[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:23:10,962[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.12072863328457346[0m
|
| 27 |
+
[[36m2026-07-22 17:23:11,263[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:23:14,318[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:23:16,517[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:23:19,670[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:23:33,676[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:23:34,089[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:23:34,904[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:23:36,479[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:23:36,802[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:23:37,222[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:23:37,222[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:23:37,222[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:23:37,575[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:23:38,346[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:23:39,900[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:23:40,202[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:23:40,731[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:23:40,731[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:23:40,731[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:23:40,732[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:23:40,732[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.18557101233951934[0m
|
| 68 |
+
[[36m2026-07-22 17:23:41,460[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:23:41,470[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.09it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 11.98it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.66it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.06it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.45it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.23it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.03it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 12.85it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.24it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.23it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.23it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.37it/s]
|
| 86 |
+
[[36m2026-07-22 17:23:43,376[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:23:43,382[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:23:43,382[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -35.686663556407154[0m
|
| 89 |
+
[[36m2026-07-22 17:23:43,693[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:23:44,781[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.06249982300166865[0m
|
logs/eval_NPO_lr1e-05_beta0.05_alpha5_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.63it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.38it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 12.95it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.25it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 13.60it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.31it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.15it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.01it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.36it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 13.18it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 13.33it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 13.44it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:19:03,914[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/NPO_lr1e-05_beta0.05_alpha5_epoch10[0m
|
| 4 |
+
[[36m2026-07-22 17:19:03,915[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:19:03,915[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/NPO_lr1e-05_beta0.05_alpha5_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:19:03,915[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/NPO_lr1e-05_beta0.05_alpha5_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:19:04,622[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:19:08,125[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:19:22,565[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:19:22,565[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.6292047787397415[0m
|
| 15 |
+
[[36m2026-07-22 17:19:22,570[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:19:22,570[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:19:22,579[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:19:22,580[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 0.0003254839625295495[0m
|
| 19 |
+
[[36m2026-07-22 17:19:23,141[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:19:26,006[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.15761444091796875[0m
|
| 23 |
+
[[36m2026-07-22 17:19:26,630[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:19:29,477[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.15076141467060733[0m
|
| 27 |
+
[[36m2026-07-22 17:19:30,172[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:19:33,279[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:19:36,352[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:19:39,442[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:19:53,470[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:19:53,839[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:19:54,679[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:19:56,465[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:19:56,703[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:19:57,136[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:19:57,136[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:19:57,136[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:19:57,402[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:19:58,104[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:19:59,646[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:19:59,892[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:20:00,406[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:20:00,406[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:20:00,406[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:20:00,406[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:20:00,407[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.4501295582236843[0m
|
| 68 |
+
[[36m2026-07-22 17:20:00,963[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:20:00,970[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.63it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.38it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 12.95it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.25it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 13.60it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.31it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.15it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.01it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.36it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 13.18it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 13.33it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 13.44it/s]
|
| 86 |
+
[[36m2026-07-22 17:20:02,859[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:20:02,866[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:20:02,866[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -52.1324735191217[0m
|
| 89 |
+
[[36m2026-07-22 17:20:03,126[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:20:04,287[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.07698815361441316[0m
|
logs/eval_NPO_lr1e-05_beta0.05_alpha5_epoch5.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.85it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 12.70it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 13.36it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 13.77it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 14.13it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:00<00:00, 13.82it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 13.68it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 13.56it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 13.99it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 14.02it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 14.08it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:00<00:00, 14.18it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-22 17:27:04,381[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/NPO_lr1e-05_beta0.05_alpha5_epoch5[0m
|
| 4 |
+
[[36m2026-07-22 17:27:04,382[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-22 17:27:04,382[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/NPO_lr1e-05_beta0.05_alpha5_epoch5/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-22 17:27:04,382[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/NPO_lr1e-05_beta0.05_alpha5_epoch5/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-22 17:27:05,164[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-22 17:27:08,955[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-22 17:27:23,347[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-22 17:27:23,348[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.6853083762251453[0m
|
| 15 |
+
[[36m2026-07-22 17:27:23,352[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-22 17:27:23,352[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-22 17:27:23,359[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-22 17:27:23,360[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 0.0001305755477065129[0m
|
| 19 |
+
[[36m2026-07-22 17:27:23,786[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-22 17:27:26,696[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.1974298095703125[0m
|
| 23 |
+
[[36m2026-07-22 17:27:27,006[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-22 17:27:28,980[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.1313780968030782[0m
|
| 27 |
+
[[36m2026-07-22 17:27:29,299[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-22 17:27:32,370[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-22 17:27:35,484[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-22 17:27:38,614[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-22 17:27:52,633[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-22 17:27:53,046[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-22 17:27:53,872[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-22 17:27:55,473[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-22 17:27:55,783[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-22 17:27:56,201[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-22 17:27:56,202[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-22 17:27:56,202[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-22 17:27:56,519[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-22 17:27:57,356[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-22 17:27:58,926[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-22 17:27:59,236[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-22 17:27:59,674[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-22 17:27:59,674[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-22 17:27:59,674[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-22 17:27:59,674[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-22 17:27:59,674[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.3557828895774852[0m
|
| 68 |
+
[[36m2026-07-22 17:28:00,305[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-22 17:28:00,314[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.85it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 12.70it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 13.36it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 13.77it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 14.13it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:00<00:00, 13.82it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 13.68it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 13.56it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 13.99it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 14.02it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 14.08it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:00<00:00, 14.18it/s]
|
| 86 |
+
[[36m2026-07-22 17:28:02,123[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-22 17:28:02,129[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-22 17:28:02,129[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -68.06300741947918[0m
|
| 89 |
+
[[36m2026-07-22 17:28:02,493[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-22 17:28:03,589[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.07352854552860216[0m
|
logs/eval_NPO_lr1e-05_beta0.1_alpha1_epoch10.log
ADDED
|
@@ -0,0 +1,78 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 1 |
15%|█▌ | 2/13 [00:00<00:00, 13.22it/s]
|
| 2 |
31%|███ | 4/13 [00:00<00:00, 11.48it/s]
|
| 3 |
46%|████▌ | 6/13 [00:00<00:00, 11.81it/s]
|
| 4 |
62%|██████▏ | 8/13 [00:00<00:00, 11.91it/s]
|
| 5 |
77%|███████▋ | 10/13 [00:00<00:00, 12.19it/s]
|
| 6 |
92%|█████████▏| 12/13 [00:01<00:00, 11.89it/s]
|
|
|
|
| 7 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 8 |
15%|█▌ | 2/13 [00:00<00:00, 11.61it/s]
|
| 9 |
31%|███ | 4/13 [00:00<00:00, 11.52it/s]
|
| 10 |
46%|████▌ | 6/13 [00:00<00:00, 11.88it/s]
|
| 11 |
62%|██████▏ | 8/13 [00:00<00:00, 11.17it/s]
|
| 12 |
77%|███████▋ | 10/13 [00:00<00:00, 10.97it/s]
|
| 13 |
92%|█████████▏| 12/13 [00:01<00:00, 11.42it/s]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/opt/conda/envs/copygen/lib/python3.12/site-packages/requests/__init__.py:113: RequestsDependencyWarning: urllib3 (2.6.3) or chardet (6.0.0.post1)/charset_normalizer (3.4.4) doesn't match a supported version!
|
| 2 |
+
warnings.warn(
|
| 3 |
+
[[36m2026-07-20 13:26:02,589[0m][[34mevaluator[0m][[32mINFO[0m] - Evaluations stored in the experiment directory: ./saves/eval/NPO_lr1e-05_beta0.1_alpha1_epoch10[0m
|
| 4 |
+
[[36m2026-07-20 13:26:02,590[0m][[34mevaluator[0m][[32mINFO[0m] - ***** Running TOFU evaluation suite *****[0m
|
| 5 |
+
[[36m2026-07-20 13:26:02,591[0m][[34mevaluator[0m][[32mINFO[0m] - Fine-grained evaluations will be saved to: ./saves/eval/NPO_lr1e-05_beta0.1_alpha1_epoch10/TOFU_EVAL.json[0m
|
| 6 |
+
[[36m2026-07-20 13:26:02,591[0m][[34mevaluator[0m][[32mINFO[0m] - Aggregated evaluations will be summarised in: ./saves/eval/NPO_lr1e-05_beta0.1_alpha1_epoch10/TOFU_SUMMARY.json[0m
|
| 7 |
+
[[36m2026-07-20 13:26:03,295[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PARA_Prob[0m
|
| 8 |
+
|
| 9 |
+
Evaluated 400 examples
|
| 10 |
+
[[36m2026-07-20 13:26:07,077[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_PERT_Prob[0m
|
| 11 |
+
|
| 12 |
+
Evaluated 400 examples
|
| 13 |
+
[[36m2026-07-20 13:26:20,224[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_truth_ratio[0m
|
| 14 |
+
[[36m2026-07-20 13:26:20,225[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_truth_ratio: 0.6409193995767771[0m
|
| 15 |
+
[[36m2026-07-20 13:26:20,233[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping forget_quality's precompute forget_truth_ratio, already evaluated.[0m
|
| 16 |
+
[[36m2026-07-20 13:26:20,234[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 17 |
+
[[36m2026-07-20 13:26:20,249[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_quality[0m
|
| 18 |
+
[[36m2026-07-20 13:26:20,251[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_quality: 0.00017795251634334[0m
|
| 19 |
+
[[36m2026-07-20 13:26:20,616[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_Prob[0m
|
| 20 |
+
|
| 21 |
+
Evaluated 400 examples
|
| 22 |
+
[[36m2026-07-20 13:26:23,351[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_Prob: 0.20775527954101564[0m
|
| 23 |
+
[[36m2026-07-20 13:26:23,624[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating forget_Q_A_ROUGE[0m
|
| 24 |
+
|
| 25 |
+
Evaluated 400 examples
|
| 26 |
+
[[36m2026-07-20 13:26:30,124[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric forget_Q_A_ROUGE: 0.18619367500997983[0m
|
| 27 |
+
[[36m2026-07-20 13:26:30,535[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_Prob[0m
|
| 28 |
+
|
| 29 |
+
Evaluated 400 examples
|
| 30 |
+
[[36m2026-07-20 13:26:33,396[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_ROUGE[0m
|
| 31 |
+
|
| 32 |
+
Evaluated 400 examples
|
| 33 |
+
[[36m2026-07-20 13:26:40,673[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PARA_Prob[0m
|
| 34 |
+
|
| 35 |
+
Evaluated 400 examples
|
| 36 |
+
[[36m2026-07-20 13:26:43,686[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Q_A_PERT_Prob[0m
|
| 37 |
+
|
| 38 |
+
Evaluated 400 examples
|
| 39 |
+
[[36m2026-07-20 13:26:56,590[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating retain_Truth_Ratio[0m
|
| 40 |
+
[[36m2026-07-20 13:26:56,990[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob[0m
|
| 41 |
+
|
| 42 |
+
Evaluated 100 examples
|
| 43 |
+
[[36m2026-07-20 13:26:57,760[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_PERT_Prob[0m
|
| 44 |
+
|
| 45 |
+
Evaluated 100 examples
|
| 46 |
+
[[36m2026-07-20 13:26:59,276[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_Prob_normalised[0m
|
| 47 |
+
[[36m2026-07-20 13:26:59,741[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Q_A_ROUGE[0m
|
| 48 |
+
|
| 49 |
+
Evaluated 100 examples
|
| 50 |
+
[[36m2026-07-20 13:27:01,084[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_Prob, already evaluated.[0m
|
| 51 |
+
[[36m2026-07-20 13:27:01,085[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping ra_Truth_Ratio's precompute ra_Q_A_PERT_Prob, already evaluated.[0m
|
| 52 |
+
[[36m2026-07-20 13:27:01,085[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating ra_Truth_Ratio[0m
|
| 53 |
+
[[36m2026-07-20 13:27:01,364[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob[0m
|
| 54 |
+
|
| 55 |
+
Evaluated 117 examples
|
| 56 |
+
[[36m2026-07-20 13:27:02,086[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_PERT_Prob[0m
|
| 57 |
+
|
| 58 |
+
Evaluated 117 examples
|
| 59 |
+
[[36m2026-07-20 13:27:03,568[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_Prob_normalised[0m
|
| 60 |
+
[[36m2026-07-20 13:27:03,851[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Q_A_ROUGE[0m
|
| 61 |
+
|
| 62 |
+
Evaluated 117 examples
|
| 63 |
+
[[36m2026-07-20 13:27:05,400[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_Prob, already evaluated.[0m
|
| 64 |
+
[[36m2026-07-20 13:27:05,400[0m][[34mmetrics[0m][[32mINFO[0m] - Skipping wf_Truth_Ratio's precompute wf_Q_A_PERT_Prob, already evaluated.[0m
|
| 65 |
+
[[36m2026-07-20 13:27:05,400[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating wf_Truth_Ratio[0m
|
| 66 |
+
[[36m2026-07-20 13:27:05,401[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating model_utility[0m
|
| 67 |
+
[[36m2026-07-20 13:27:05,402[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric model_utility: 0.4312575503326488[0m
|
| 68 |
+
[[36m2026-07-20 13:27:05,951[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 69 |
+
[[36m2026-07-20 13:27:05,966[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating mia_min_k[0m
|
| 70 |
+
|
| 71 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 72 |
15%|█▌ | 2/13 [00:00<00:00, 13.22it/s]
|
| 73 |
31%|███ | 4/13 [00:00<00:00, 11.48it/s]
|
| 74 |
46%|████▌ | 6/13 [00:00<00:00, 11.81it/s]
|
| 75 |
62%|██████▏ | 8/13 [00:00<00:00, 11.91it/s]
|
| 76 |
77%|███████▋ | 10/13 [00:00<00:00, 12.19it/s]
|
| 77 |
92%|█████████▏| 12/13 [00:01<00:00, 11.89it/s]
|
| 78 |
+
|
| 79 |
0%| | 0/13 [00:00<?, ?it/s]
|
| 80 |
15%|█▌ | 2/13 [00:00<00:00, 11.61it/s]
|
| 81 |
31%|███ | 4/13 [00:00<00:00, 11.52it/s]
|
| 82 |
46%|████▌ | 6/13 [00:00<00:00, 11.88it/s]
|
| 83 |
62%|██████▏ | 8/13 [00:00<00:00, 11.17it/s]
|
| 84 |
77%|███████▋ | 10/13 [00:00<00:00, 10.97it/s]
|
| 85 |
92%|█████████▏| 12/13 [00:01<00:00, 11.42it/s]
|
| 86 |
+
[[36m2026-07-20 13:27:08,116[0m][[34mmetrics[0m][[32mINFO[0m] - Loading evaluations from /workspace/jspace-unlearning/refs/tofu_Llama-3.2-1B-Instruct_retain90__TOFU_EVAL.json[0m
|
| 87 |
+
[[36m2026-07-20 13:27:08,132[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating privleak[0m
|
| 88 |
+
[[36m2026-07-20 13:27:08,132[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric privleak: -52.45793767245727[0m
|
| 89 |
+
[[36m2026-07-20 13:27:08,516[0m][[34mmetrics[0m][[32mINFO[0m] - Evaluating extraction_strength[0m
|
| 90 |
+
|
| 91 |
+
Evaluated 400 examples
|
| 92 |
+
[[36m2026-07-20 13:27:09,803[0m][[34mevaluator[0m][[32mINFO[0m] - Result for metric extraction_strength: 0.09623633139113906[0m
|