初始化项目，由ModelHub XC社区提供模型
Model: SeongryongJung/Qwen3-8B-Chemistry-RLSD-TR Source: Original Platform
This commit is contained in:
ModelHub XC
911
artifacts/output.log
Normal file
911
artifacts/output.log
Normal file
@@ -0,0 +1,911 @@
|
||||
wandb: Detected [openai] in use.
|
||||
wandb: Use W&B Weave for improved LLM call tracing. Install Weave with `pip install weave` then add `import weave` to the top of your script.
|
||||
wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
|
||||
Checkpoint tracker file does not exist: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/latest_checkpointed_iteration.txt
|
||||
Training from scratch
|
||||
Training Progress: 9%|▉ | 9/100 [07:02<1:06:50, 44.07s/it]
|
||||
step:1 - global_seqlen/min:18655 - global_seqlen/max:26853 - global_seqlen/minmax_diff:8198 - global_seqlen/balanced_min:23692 - global_seqlen/balanced_max:23701 - global_seqlen/mean:23696.5 - actor/entropy:0.5224554538726807 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.14097940921783447 - training/rollout_probs_diff_mean:0.0045183561742305756 - training/rollout_probs_diff_std:0.009310152381658554 - training/rollout_actor_probs_pearson_corr:0.9994592070579529 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.7421875 - self_distillation/correct_sample_fraction:0.46484375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7421875 - rollout_corr/rollout_is_mean:0.9999948143959045 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.428677908028476e-05 - rollout_corr/rollout_is_max:1.4933885335922241 - rollout_corr/rollout_is_min:0.4801149070262909 - rollout_corr/rollout_is_std:0.03428589925169945 - rollout_corr/rollout_is_eff_sample_size:0.9988258574271096 - rollout_corr/rollout_is_seq_mean:1.0000158548355103 - rollout_corr/rollout_is_seq_std:0.0017025339184328914 - rollout_corr/rollout_is_seq_max:1.0078747272491455 - rollout_corr/rollout_is_seq_min:0.9956079721450806 - rollout_corr/rollout_is_seq_max_deviation:0.007874727249145508 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7518747616559267) - rollout_corr/training_log_ppl:np.float64(0.5485204759170301) - rollout_corr/kl:np.float64(0.0006133353596755597) - rollout_corr/k3_kl:np.float64(0.0006291264688229603) - rollout_corr/rollout_ppl:np.float64(1.750815692357719) - rollout_corr/rollout_log_ppl:np.float64(0.5479071354493499) - rollout_corr/log_ppl_diff:np.float64(0.0006133404676802456) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013540681102313101) - rollout_corr/log_ppl_diff_max:np.float64(0.005063414573669434) - rollout_corr/log_ppl_diff_min:np.float64(-0.006822109222412109) - rollout_corr/ppl_ratio:np.float64(1.0006149462424219) - rollout_corr/chi2_token:np.float64(0.0012849036138504744) - rollout_corr/chi2_seq:np.float64(0.5767202903516591) - actor/pg_loss:np.float64(0.0015394691145047545) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0) - actor/ppo_kl:np.float64(0.0) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.7421875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.7421875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.46484375) - self_distillation/token_reweight_w_mean:np.float64(29057.488782484084) - self_distillation/token_reweight_w_std:np.float64(551978.9261493584) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0031458756420761) - self_distillation/token_reweight_w_clip_frac:np.float64(0.14994207414565608) - self_distillation/empty_target_batch:np.float64(0.2578125) - actor/grad_norm:np.float64(0.34795472025871277) - perf/mfu/actor:np.float64(0.039808731599246736) - perf/max_memory_allocated_gb:np.float64(130.98325157165527) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(208.84796524047852) - actor/lr:np.float64(0.0) - training/global_step:1 - training/epoch:0 - critic/score/mean:0.46484375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.46484375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.004086250439286232 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.004086250439286232 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:482.515625 - response_length/max:1194.0 - response_length/min:161.0 - respo
|
||||
step:2 - global_seqlen/min:18554 - global_seqlen/max:28889 - global_seqlen/minmax_diff:10335 - global_seqlen/balanced_min:23930 - global_seqlen/balanced_max:23950 - global_seqlen/mean:23945.25 - actor/entropy:0.5649150013923645 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.21355730295181274 - training/rollout_probs_diff_mean:0.004759751260280609 - training/rollout_probs_diff_std:0.009735357947647572 - training/rollout_actor_probs_pearson_corr:0.9994669556617737 - self_distillation/success_group_fraction:0.625 - self_distillation/success_sample_fraction:0.609375 - self_distillation/correct_sample_fraction:0.3671875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.609375 - rollout_corr/rollout_is_mean:0.999841034412384 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.8218586774310097e-05 - rollout_corr/rollout_is_max:1.8565502166748047 - rollout_corr/rollout_is_min:0.4281544089317322 - rollout_corr/rollout_is_std:0.03569299355149269 - rollout_corr/rollout_is_eff_sample_size:0.9987271556828731 - rollout_corr/rollout_is_seq_mean:0.9997751712799072 - rollout_corr/rollout_is_seq_std:0.002073427429422736 - rollout_corr/rollout_is_seq_max:1.0066989660263062 - rollout_corr/rollout_is_seq_min:0.993266224861145 - rollout_corr/rollout_is_seq_max_deviation:0.0067337751388549805 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.8422705363482237) - rollout_corr/training_log_ppl:np.float64(0.5950889247469604) - rollout_corr/kl:np.float64(0.000921569124747279) - rollout_corr/k3_kl:np.float64(0.0006883058893549787) - rollout_corr/rollout_ppl:np.float64(1.8405294218100607) - rollout_corr/rollout_log_ppl:np.float64(0.5941673531197011) - rollout_corr/log_ppl_diff:np.float64(0.0009215716272592545) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017208073986694217) - rollout_corr/log_ppl_diff_max:np.float64(0.007348239421844482) - rollout_corr/log_ppl_diff_min:np.float64(-0.005036115646362305) - rollout_corr/ppl_ratio:np.float64(1.0009240037761629) - rollout_corr/chi2_token:np.float64(0.0009099591989070177) - rollout_corr/chi2_seq:np.float64(0.3998524572234601) - actor/pg_loss:np.float64(0.005818751524202526) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002621370244924037) - actor/ppo_kl:np.float64(9.864444990625998e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.609375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.609375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.3671875) - self_distillation/token_reweight_w_mean:np.float64(25264.271957140416) - self_distillation/token_reweight_w_std:np.float64(438644.3446098004) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0013297640252858) - self_distillation/token_reweight_w_clip_frac:np.float64(0.13903450994985178) - self_distillation/empty_target_batch:np.float64(0.390625) - actor/grad_norm:np.float64(0.3431096374988556) - perf/mfu/actor:np.float64(0.0564479877562713) - perf/max_memory_allocated_gb:np.float64(130.98325157165527) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(204.3056116104126) - actor/lr:np.float64(1e-07) - training/global_step:2 - training/epoch:0 - critic/score/mean:0.3671875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.3671875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0009724170668050647 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0009724170668050647 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:428.8203125 - response_length/max:1497.0 - res
|
||||
step:3 - global_seqlen/min:19604 - global_seqlen/max:28466 - global_seqlen/minmax_diff:8862 - global_seqlen/balanced_min:24149 - global_seqlen/balanced_max:24177 - global_seqlen/mean:24164.75 - actor/entropy:0.5242184400558472 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1616789698600769 - training/rollout_probs_diff_mean:0.004656430333852768 - training/rollout_probs_diff_std:0.009660309180617332 - training/rollout_actor_probs_pearson_corr:0.9994632601737976 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.765625 - self_distillation/correct_sample_fraction:0.4140625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.765625 - rollout_corr/rollout_is_mean:1.0001205205917358 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:8.629170224594418e-06 - rollout_corr/rollout_is_max:1.6356302499771118 - rollout_corr/rollout_is_min:0.49262791872024536 - rollout_corr/rollout_is_std:0.0352848656475544 - rollout_corr/rollout_is_eff_sample_size:0.9987568830709286 - rollout_corr/rollout_is_seq_mean:1.000150442123413 - rollout_corr/rollout_is_seq_std:0.0021288390271365643 - rollout_corr/rollout_is_seq_max:1.007064700126648 - rollout_corr/rollout_is_seq_min:0.9945412278175354 - rollout_corr/rollout_is_seq_max_deviation:0.007064700126647949 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7977697649039328) - rollout_corr/training_log_ppl:np.float64(0.5697851715376601) - rollout_corr/kl:np.float64(0.0006146689036183872) - rollout_corr/k3_kl:np.float64(0.0006752699637218029) - rollout_corr/rollout_ppl:np.float64(1.796580535825342) - rollout_corr/rollout_log_ppl:np.float64(0.5691705014905892) - rollout_corr/log_ppl_diff:np.float64(0.0006146700470708311) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015643324586562812) - rollout_corr/log_ppl_diff_max:np.float64(0.006385087966918945) - rollout_corr/log_ppl_diff_min:np.float64(-0.006580471992492676) - rollout_corr/ppl_ratio:np.float64(1.0006169071421027) - rollout_corr/chi2_token:np.float64(0.0014803188387304544) - rollout_corr/chi2_seq:np.float64(0.8378349014092237) - actor/pg_loss:np.float64(0.008528164820745587) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00031198100373330817) - actor/ppo_kl:np.float64(8.786334534249818e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.765625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.765625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.4140625) - self_distillation/token_reweight_w_mean:np.float64(19923.08156236657) - self_distillation/token_reweight_w_std:np.float64(356201.15468769113) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.001924671465531) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1814161114161834) - self_distillation/empty_target_batch:np.float64(0.234375) - actor/grad_norm:np.float64(0.4067366272211075) - perf/mfu/actor:np.float64(0.05678632217924573) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(202.01263046264648) - actor/lr:np.float64(2e-07) - training/global_step:3 - training/epoch:0 - critic/score/mean:0.4140625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.4140625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.002736525610089302 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.002736525610089302 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:452.6796875 - response_length/max:1521.0 - respons
|
||||
step:4 - global_seqlen/min:23313 - global_seqlen/max:27346 - global_seqlen/minmax_diff:4033 - global_seqlen/balanced_min:24709 - global_seqlen/balanced_max:24717 - global_seqlen/mean:24714.5 - actor/entropy:0.54982590675354 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7950142621994019 - training/rollout_probs_diff_mean:0.004631490912288427 - training/rollout_probs_diff_std:0.009951530024409294 - training/rollout_actor_probs_pearson_corr:0.9994604587554932 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.79296875 - self_distillation/correct_sample_fraction:0.4453125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.79296875 - rollout_corr/rollout_is_mean:1.0000749826431274 - rollout_corr/rollout_is_ratio_fraction_high:8.3895429270342e-06 - rollout_corr/rollout_is_ratio_fraction_low:1.67790858540684e-05 - rollout_corr/rollout_is_max:4.939357280731201 - rollout_corr/rollout_is_min:0.19150345027446747 - rollout_corr/rollout_is_std:0.0351443812251091 - rollout_corr/rollout_is_eff_sample_size:0.9987665151242391 - rollout_corr/rollout_is_seq_mean:1.0000817775726318 - rollout_corr/rollout_is_seq_std:0.0017939795507118106 - rollout_corr/rollout_is_seq_max:1.0054137706756592 - rollout_corr/rollout_is_seq_min:0.9938213229179382 - rollout_corr/rollout_is_seq_max_deviation:0.006178677082061768 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.8050872194580734) - rollout_corr/training_log_ppl:np.float64(0.5758670450886711) - rollout_corr/kl:np.float64(0.0006997874668996928) - rollout_corr/k3_kl:np.float64(0.0006716491621432397) - rollout_corr/rollout_ppl:np.float64(1.803772633895278) - rollout_corr/rollout_log_ppl:np.float64(0.5751672576880082) - rollout_corr/log_ppl_diff:np.float64(0.000699787400662899) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015515540726482868) - rollout_corr/log_ppl_diff_max:np.float64(0.00667417049407959) - rollout_corr/log_ppl_diff_min:np.float64(-0.004105627536773682) - rollout_corr/ppl_ratio:np.float64(1.0007018391042948) - rollout_corr/chi2_token:np.float64(0.0013059712946414948) - rollout_corr/chi2_seq:np.float64(0.8464284441433847) - actor/pg_loss:np.float64(0.0077759657287970185) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00014584792415917036) - actor/ppo_kl:np.float64(0.00010806482939074158) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.79296875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.79296875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.4453125) - self_distillation/token_reweight_w_mean:np.float64(18390.18527350994) - self_distillation/token_reweight_w_std:np.float64(278661.7640647436) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0027553322724998) - self_distillation/token_reweight_w_clip_frac:np.float64(0.18228401592932642) - self_distillation/empty_target_batch:np.float64(0.20703125) - actor/grad_norm:np.float64(0.41205956041812897) - perf/mfu/actor:np.float64(0.05901765112715433) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(199.91095733642578) - actor/lr:np.float64(3e-07) - training/global_step:4 - training/epoch:0 - critic/score/mean:0.4453125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.4453125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0022473488934338093 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0022473488934338093 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:465.609375 - response_le
|
||||
step:5 - global_seqlen/min:18324 - global_seqlen/max:26762 - global_seqlen/minmax_diff:8438 - global_seqlen/balanced_min:22550 - global_seqlen/balanced_max:22563 - global_seqlen/mean:22556.25 - actor/entropy:0.5743926167488098 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18463686108589172 - training/rollout_probs_diff_mean:0.004951741546392441 - training/rollout_probs_diff_std:0.009703434072434902 - training/rollout_actor_probs_pearson_corr:0.9994615912437439 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.42578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8125 - rollout_corr/rollout_is_mean:1.000128984451294 - rollout_corr/rollout_is_ratio_fraction_high:9.058463547262363e-06 - rollout_corr/rollout_is_ratio_fraction_low:1.8116927094524726e-05 - rollout_corr/rollout_is_max:2.5132858753204346 - rollout_corr/rollout_is_min:0.4844912588596344 - rollout_corr/rollout_is_std:0.03672673553228378 - rollout_corr/rollout_is_eff_sample_size:0.9986534393962608 - rollout_corr/rollout_is_seq_mean:1.0001883506774902 - rollout_corr/rollout_is_seq_std:0.0020744451321661472 - rollout_corr/rollout_is_seq_max:1.0107048749923706 - rollout_corr/rollout_is_seq_min:0.9943471550941467 - rollout_corr/rollout_is_seq_max_deviation:0.010704874992370605 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.8240970252081752) - rollout_corr/training_log_ppl:np.float64(0.5867100825998932) - rollout_corr/kl:np.float64(0.000413151877571849) - rollout_corr/k3_kl:np.float64(0.0007104324042757071) - rollout_corr/rollout_ppl:np.float64(1.8233414590358734) - rollout_corr/rollout_log_ppl:np.float64(0.5862969291047193) - rollout_corr/log_ppl_diff:np.float64(0.0004131534951739013) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015994933783076704) - rollout_corr/log_ppl_diff_max:np.float64(0.005790770053863525) - rollout_corr/log_ppl_diff_min:np.float64(-0.006320059299468994) - rollout_corr/ppl_ratio:np.float64(1.0004152327310294) - rollout_corr/chi2_token:np.float64(0.0020708865486085415) - rollout_corr/chi2_seq:np.float64(1.291173740522936) - actor/pg_loss:np.float64(0.00783018337097019) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00029261777604006056) - actor/ppo_kl:np.float64(-0.00011415763041355831) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.42578125) - self_distillation/token_reweight_w_mean:np.float64(7564.456899565645) - self_distillation/token_reweight_w_std:np.float64(147554.27462118154) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0038912589661777) - self_distillation/token_reweight_w_clip_frac:np.float64(0.19292002776637673) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.37297531962394714) - perf/mfu/actor:np.float64(0.05395284544548605) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(198.01452255249023) - actor/lr:np.float64(4e-07) - training/global_step:5 - training/epoch:0 - critic/score/mean:0.42578125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.42578125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0022725421003997326 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0022725421003997326 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:431.2265625 - response_lengt
|
||||
step:6 - global_seqlen/min:19330 - global_seqlen/max:29022 - global_seqlen/minmax_diff:9692 - global_seqlen/balanced_min:23612 - global_seqlen/balanced_max:23621 - global_seqlen/mean:23617.625 - actor/entropy:0.5028008818626404 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5328631401062012 - training/rollout_probs_diff_mean:0.004628530703485012 - training/rollout_probs_diff_std:0.009826785884797573 - training/rollout_actor_probs_pearson_corr:0.9994211196899414 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.8515625 - self_distillation/correct_sample_fraction:0.421875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8515625 - rollout_corr/rollout_is_mean:0.9999843239784241 - rollout_corr/rollout_is_ratio_fraction_high:8.433054063061718e-06 - rollout_corr/rollout_is_ratio_fraction_low:8.433054063061718e-06 - rollout_corr/rollout_is_max:2.1156668663024902 - rollout_corr/rollout_is_min:0.4510452151298523 - rollout_corr/rollout_is_std:0.03478982672095299 - rollout_corr/rollout_is_eff_sample_size:0.998791012294431 - rollout_corr/rollout_is_seq_mean:0.9999687671661377 - rollout_corr/rollout_is_seq_std:0.0017612202791497111 - rollout_corr/rollout_is_seq_max:1.0042917728424072 - rollout_corr/rollout_is_seq_min:0.994244396686554 - rollout_corr/rollout_is_seq_max_deviation:0.005755603313446045 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.732960639987141) - rollout_corr/training_log_ppl:np.float64(0.5360075897770002) - rollout_corr/kl:np.float64(0.000639045071211708) - rollout_corr/k3_kl:np.float64(0.0006780325196018566) - rollout_corr/rollout_ppl:np.float64(1.7318218029104173) - rollout_corr/rollout_log_ppl:np.float64(0.5353685396839865) - rollout_corr/log_ppl_diff:np.float64(0.0006390500930137932) - rollout_corr/log_ppl_abs_diff:np.float64(0.001630233891773969) - rollout_corr/log_ppl_diff_max:np.float64(0.007409811019897461) - rollout_corr/log_ppl_diff_min:np.float64(-0.0069411396980285645) - rollout_corr/ppl_ratio:np.float64(1.0006413247901946) - rollout_corr/chi2_token:np.float64(0.0014372044242918491) - rollout_corr/chi2_seq:np.float64(1.4436022932641208) - actor/pg_loss:np.float64(0.00812386441975832) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004630846383406606) - actor/ppo_kl:np.float64(-4.196475500339858e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8515625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8515625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.421875) - self_distillation/token_reweight_w_mean:np.float64(19923.261127433972) - self_distillation/token_reweight_w_std:np.float64(319160.37308162905) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0048198685981333) - self_distillation/token_reweight_w_clip_frac:np.float64(0.18613416605512612) - self_distillation/empty_target_batch:np.float64(0.1484375) - actor/grad_norm:np.float64(0.3917096257209778) - perf/mfu/actor:np.float64(0.05549581290987645) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(196.47495651245117) - actor/lr:np.float64(5e-07) - training/global_step:6 - training/epoch:0 - critic/score/mean:0.421875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.421875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.008403538726270199 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.008403538726270199 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:463.20703125 - response_length/
|
||||
step:7 - global_seqlen/min:18757 - global_seqlen/max:24212 - global_seqlen/minmax_diff:5455 - global_seqlen/balanced_min:22489 - global_seqlen/balanced_max:22504 - global_seqlen/mean:22499.75 - actor/entropy:0.5316406488418579 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.8049387335777283 - training/rollout_probs_diff_mean:0.004808515310287476 - training/rollout_probs_diff_std:0.0102939298376441 - training/rollout_actor_probs_pearson_corr:0.9993799924850464 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.69921875 - self_distillation/correct_sample_fraction:0.33203125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.69921875 - rollout_corr/rollout_is_mean:1.000180721282959 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:4.7163583076326177e-05 - rollout_corr/rollout_is_max:1.9837769269943237 - rollout_corr/rollout_is_min:0.12898790836334229 - rollout_corr/rollout_is_std:0.0356261320412159 - rollout_corr/rollout_is_eff_sample_size:0.9987328632041119 - rollout_corr/rollout_is_seq_mean:1.0002025365829468 - rollout_corr/rollout_is_seq_std:0.0019561529625207186 - rollout_corr/rollout_is_seq_max:1.0071183443069458 - rollout_corr/rollout_is_seq_min:0.9942708015441895 - rollout_corr/rollout_is_seq_max_deviation:0.007118344306945801 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7598580992780626) - rollout_corr/training_log_ppl:np.float64(0.5522571774781682) - rollout_corr/kl:np.float64(0.0006528583797624776) - rollout_corr/k3_kl:np.float64(0.0007534040726682179) - rollout_corr/rollout_ppl:np.float64(1.7586921667680144) - rollout_corr/rollout_log_ppl:np.float64(0.5516043193638325) - rollout_corr/log_ppl_diff:np.float64(0.000652858114335686) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017435748013667762) - rollout_corr/log_ppl_diff_max:np.float64(0.010204493999481201) - rollout_corr/log_ppl_diff_min:np.float64(-0.006472587585449219) - rollout_corr/ppl_ratio:np.float64(1.000655469717458) - rollout_corr/chi2_token:np.float64(0.001773336436599493) - rollout_corr/chi2_seq:np.float64(1.4482529927045107) - actor/pg_loss:np.float64(0.007577315904200077) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005699304592781118) - actor/ppo_kl:np.float64(0.00017036601963926756) - actor/pg_clipfrac_lower:np.float64(1.5439723938470706e-05) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.69921875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.69921875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.33203125) - self_distillation/token_reweight_w_mean:np.float64(40804.10983084305) - self_distillation/token_reweight_w_std:np.float64(550034.2616531486) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.002954869531095) - self_distillation/token_reweight_w_clip_frac:np.float64(0.17178265267284587) - self_distillation/empty_target_batch:np.float64(0.30078125) - actor/grad_norm:np.float64(0.42640380561351776) - perf/mfu/actor:np.float64(0.053623795084905154) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(196.01875686645508) - actor/lr:np.float64(6e-07) - training/global_step:7 - training/epoch:0 - critic/score/mean:0.33203125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.33203125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00599684938788414 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.00599684938788414 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:414.1171875 - respons
|
||||
step:8 - global_seqlen/min:19293 - global_seqlen/max:28026 - global_seqlen/minmax_diff:8733 - global_seqlen/balanced_min:24653 - global_seqlen/balanced_max:24794 - global_seqlen/mean:24675.25 - actor/entropy:0.5080747604370117 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.408370703458786 - training/rollout_probs_diff_mean:0.0044975439086556435 - training/rollout_probs_diff_std:0.009530728682875633 - training/rollout_actor_probs_pearson_corr:0.9994667768478394 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.8828125 - self_distillation/correct_sample_fraction:0.46875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8828125 - rollout_corr/rollout_is_mean:1.0000174045562744 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:8.162332051142585e-06 - rollout_corr/rollout_is_max:1.8054882287979126 - rollout_corr/rollout_is_min:0.462515652179718 - rollout_corr/rollout_is_std:0.0343310683965683 - rollout_corr/rollout_is_eff_sample_size:0.9988226463404877 - rollout_corr/rollout_is_seq_mean:1.000030279159546 - rollout_corr/rollout_is_seq_std:0.0018812705529853702 - rollout_corr/rollout_is_seq_max:1.0089062452316284 - rollout_corr/rollout_is_seq_min:0.9933127760887146 - rollout_corr/rollout_is_seq_max_deviation:0.008906245231628418 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.747896637301892) - rollout_corr/training_log_ppl:np.float64(0.541930019389838) - rollout_corr/kl:np.float64(0.0006643355780084903) - rollout_corr/k3_kl:np.float64(0.0006548753977995148) - rollout_corr/rollout_ppl:np.float64(1.7467446983791888) - rollout_corr/rollout_log_ppl:np.float64(0.5412656838307157) - rollout_corr/log_ppl_diff:np.float64(0.0006643355591222644) - rollout_corr/log_ppl_abs_diff:np.float64(0.00149785284884274) - rollout_corr/log_ppl_diff_max:np.float64(0.007577002048492432) - rollout_corr/log_ppl_diff_min:np.float64(-0.0072296857833862305) - rollout_corr/ppl_ratio:np.float64(1.000666505889967) - rollout_corr/chi2_token:np.float64(0.0012856200337409973) - rollout_corr/chi2_seq:np.float64(0.7123665248509496) - actor/pg_loss:np.float64(0.007868309505283833) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004503007885432453) - actor/ppo_kl:np.float64(5.489206566622329e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8828125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8828125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.46875) - self_distillation/token_reweight_w_mean:np.float64(9891.629087405046) - self_distillation/token_reweight_w_std:np.float64(229237.51225678227) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0036344188265502) - self_distillation/token_reweight_w_clip_frac:np.float64(0.18639719724887982) - self_distillation/empty_target_batch:np.float64(0.1171875) - actor/grad_norm:np.float64(0.3860222175717354) - perf/mfu/actor:np.float64(0.05765871612685469) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(196.10034942626953) - actor/lr:np.float64(7e-07) - training/global_step:8 - training/epoch:0 - critic/score/mean:0.46875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.46875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.008611260913312435 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.008611260913312435 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:478.5703125 - response_length/max:1561.0 - response_leng
|
||||
step:9 - global_seqlen/min:19805 - global_seqlen/max:26584 - global_seqlen/minmax_diff:6779 - global_seqlen/balanced_min:23155 - global_seqlen/balanced_max:23161 - global_seqlen/mean:23157.5 - actor/entropy:0.5331449508666992 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5838292837142944 - training/rollout_probs_diff_mean:0.004796977620571852 - training/rollout_probs_diff_std:0.010206756182014942 - training/rollout_actor_probs_pearson_corr:0.9993701577186584 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.66796875 - self_distillation/correct_sample_fraction:0.359375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.66796875 - rollout_corr/rollout_is_mean:0.9999645352363586 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.796686956367921e-05 - rollout_corr/rollout_is_max:1.6195783615112305 - rollout_corr/rollout_is_min:0.3546355068683624 - rollout_corr/rollout_is_std:0.03580470383167267 - rollout_corr/rollout_is_eff_sample_size:0.9987194268518426 - rollout_corr/rollout_is_seq_mean:0.9999064207077026 - rollout_corr/rollout_is_seq_std:0.001988921547308564 - rollout_corr/rollout_is_seq_max:1.004913330078125 - rollout_corr/rollout_is_seq_min:0.9934803247451782 - rollout_corr/rollout_is_seq_max_deviation:0.006519675254821777 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7555229673162103) - rollout_corr/training_log_ppl:np.float64(0.5492386808618903) - rollout_corr/kl:np.float64(0.000775432549749766) - rollout_corr/k3_kl:np.float64(0.0008021395320838565) - rollout_corr/rollout_ppl:np.float64(1.7541374694555998) - rollout_corr/rollout_log_ppl:np.float64(0.5484632462612353) - rollout_corr/log_ppl_diff:np.float64(0.0007754346006549895) - rollout_corr/log_ppl_abs_diff:np.float64(0.001795541902538389) - rollout_corr/log_ppl_diff_max:np.float64(0.010334789752960205) - rollout_corr/log_ppl_diff_min:np.float64(-0.0038338303565979004) - rollout_corr/ppl_ratio:np.float64(1.000778182875365) - rollout_corr/chi2_token:np.float64(0.0016523946542292833) - rollout_corr/chi2_seq:np.float64(0.9519539077300578) - actor/pg_loss:np.float64(0.007556321215815842) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006206532375472307) - actor/ppo_kl:np.float64(-5.760855917635865e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.66796875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.66796875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.359375) - self_distillation/token_reweight_w_mean:np.float64(22852.849850326777) - self_distillation/token_reweight_w_std:np.float64(403812.0357487643) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0031312247738242) - self_distillation/token_reweight_w_clip_frac:np.float64(0.15560277574695647) - self_distillation/empty_target_batch:np.float64(0.33203125) - actor/grad_norm:np.float64(0.34761351346969604) - perf/mfu/actor:np.float64(0.05432608742997578) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(196.07957077026367) - actor/lr:np.float64(8e-07) - training/global_step:9 - training/epoch:0 - critic/score/mean:0.359375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.359375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.008080599829554558 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.008080599829554558 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:434.828125 - response_length/max:1105.0 - re
|
||||
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 10}
|
||||
validation generation end
|
||||
[prompt] system
|
||||
|
||||
Given a question and four options, please select the right answer. Respond in the following format:
|
||||
<reasoning>
|
||||
...
|
||||
</reasoning>
|
||||
<answer>
|
||||
...
|
||||
</answer>
|
||||
|
||||
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
|
||||
user
|
||||
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||||
|
||||
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||||
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||||
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||||
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||||
Please reason step by step.
|
||||
assistant
|
||||
<think>
|
||||
|
||||
</think>
|
||||
|
||||
|
||||
[response] <reasoning>
|
||||
To determine the correct reactant for the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1", we need to examine the structure of the product and match it with the structure of the reactant provided in the options.
|
||||
|
||||
The product has the following key features:
|
||||
- A nitrogen-containing ring (likely a pyrrole or imidazole-like structure).
|
||||
- A substituent "Cc2ccc(Cl)cc2C(F)(F)F", which is a benzene ring substituted with a chlorine atom and a trifluoromethyl group.
|
||||
- The overall structure suggests that the reactant must have a similar core structure, with the necessary functional groups for the formation of the final product.
|
||||
|
||||
Let's analyze the options:
|
||||
|
||||
**Option A**:
|
||||
"O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1"
|
||||
This structure contains a nitrogen-containing ring with the correct substituent "Cc2ccc(Cl)cc2C(F)(F)F". It also has a carbonyl group and a nitrogen atom, which could be part of a condensation or coupling reaction. This matches the structure of the product closely.
|
||||
|
||||
**Option B**:
|
||||
"O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1"
|
||||
This option introduces a chlorine atom directly on the nitrogen, which is not present in the product. Also, the structure seems to have a different arrangement of substituents, making it unlikely to be the correct reactant.
|
||||
|
||||
**Option C**:
|
||||
"O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1"
|
||||
This structure has a hydroxyl group attached to the benzene ring, which is not present in the product. The presence of this hydroxyl group suggests a different synthetic pathway and thus is not the correct reactant.
|
||||
|
||||
**Option D**:
|
||||
"O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F"
|
||||
This option has a different arrangement of the substituents and introduces a double bond between the carbon atoms, which is not part of the product structure. It also lacks the necessary nitrogen-containing ring.
|
||||
|
||||
**Conclusion**:
|
||||
Only **Option A** contains the correct substituent "Cc2ccc(Cl)cc2C(F)(F)F" and the necessary nitrogen-containing ring structure. It is the only option that matches the structure of the product directly.
|
||||
</reasoning>
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
[ground_truth] A
|
||||
[score] 1.0
|
||||
[acc] 1.0
|
||||
[pred] A
|
||||
[incorrect_format] 1
|
||||
[feedback]
|
||||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_10
|
||||
step:10 - global_seqlen/min:18995 - global_seqlen/max:25222 - global_seqlen/minmax_diff:6227 - global_seqlen/balanced_min:22810 - global_seqlen/balanced_max:22819 - global_seqlen/mean:22815.375 - actor/entropy:0.5086405277252197 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.48631101846694946 - training/rollout_probs_diff_mean:0.004814056679606438 - training/rollout_probs_diff_std:0.010112768970429897 - training/rollout_actor_probs_pearson_corr:0.9993922114372253 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.6796875 - self_distillation/correct_sample_fraction:0.37109375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.6796875 - rollout_corr/rollout_is_mean:1.0000672340393066 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.7611369912629016e-05 - rollout_corr/rollout_is_max:1.758644938468933 - rollout_corr/rollout_is_min:0.47177526354789734 - rollout_corr/rollout_is_std:0.0358346551656723 - rollout_corr/rollout_is_eff_sample_size:0.9987177621962007 - rollout_corr/rollout_is_seq_mean:1.0001494884490967 - rollout_corr/rollout_is_seq_std:0.001915071508847177 - rollout_corr/rollout_is_seq_max:1.0051805973052979 - rollout_corr/rollout_is_seq_min:0.9944745898246765 - rollout_corr/rollout_is_seq_max_deviation:0.005525410175323486 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7283513699658215) - rollout_corr/training_log_ppl:np.float64(0.5342370558064431) - rollout_corr/kl:np.float64(0.00048592705383754264) - rollout_corr/k3_kl:np.float64(0.0007897751748942028) - rollout_corr/rollout_ppl:np.float64(1.7275344631634653) - rollout_corr/rollout_log_ppl:np.float64(0.5337511280667968) - rollout_corr/log_ppl_diff:np.float64(0.00048592773964628577) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017402138910256326) - rollout_corr/log_ppl_diff_max:np.float64(0.008036613464355469) - rollout_corr/log_ppl_diff_min:np.float64(-0.007182866334915161) - rollout_corr/ppl_ratio:np.float64(1.0004885760135949) - rollout_corr/chi2_token:np.float64(0.002185102319344878) - rollout_corr/chi2_seq:np.float64(1.4425214361399412) - actor/pg_loss:np.float64(0.008299664361402392) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006863867054107686) - actor/ppo_kl:np.float64(-5.119195938618759e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.6796875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.6796875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.37109375) - self_distillation/token_reweight_w_mean:np.float64(42394.31746227876) - self_distillation/token_reweight_w_std:np.float64(702394.3469068523) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0012782558333129) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1625702439341694) - self_distillation/empty_target_batch:np.float64(0.3203125) - actor/grad_norm:np.float64(0.3786138668656349) - perf/mfu/actor:np.float64(0.054599702355521036) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(196.04931259155273) - actor/lr:np.float64(9e-07) - val-aux/sciknoweval/reward/mean@16:np.float64(0.4217261904761905) - val-aux/sciknoweval/reward/std@16:np.float64(0.28826355015906874) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.5430761904761905) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.26128001800402223) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.3013761904761905) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.2277461781455412) - val-aux/sciknoweval/reward/maj@2/mean:np
|
||||
step:11 - global_seqlen/min:18238 - global_seqlen/max:25942 - global_seqlen/minmax_diff:7704 - global_seqlen/balanced_min:22325 - global_seqlen/balanced_max:22730 - global_seqlen/mean:22380.25 - actor/entropy:0.5434318780899048 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.38079291582107544 - training/rollout_probs_diff_mean:0.005151212681084871 - training/rollout_probs_diff_std:0.010387294925749302 - training/rollout_actor_probs_pearson_corr:0.9993837475776672 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.86328125 - self_distillation/correct_sample_fraction:0.48046875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.86328125 - rollout_corr/rollout_is_mean:1.0001193284988403 - rollout_corr/rollout_is_ratio_fraction_high:9.758572559803724e-06 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:4.194515228271484 - rollout_corr/rollout_is_min:0.5521866679191589 - rollout_corr/rollout_is_std:0.03714950755238533 - rollout_corr/rollout_is_eff_sample_size:0.9986220538290654 - rollout_corr/rollout_is_seq_mean:1.0001423358917236 - rollout_corr/rollout_is_seq_std:0.0020691656973212957 - rollout_corr/rollout_is_seq_max:1.0071853399276733 - rollout_corr/rollout_is_seq_min:0.993625283241272 - rollout_corr/rollout_is_seq_max_deviation:0.00718533992767334 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7692446084693074) - rollout_corr/training_log_ppl:np.float64(0.5590797059703618) - rollout_corr/kl:np.float64(0.0006710044406990079) - rollout_corr/k3_kl:np.float64(0.0007906234051233696) - rollout_corr/rollout_ppl:np.float64(1.7680208389647305) - rollout_corr/rollout_log_ppl:np.float64(0.5584087027818896) - rollout_corr/log_ppl_diff:np.float64(0.0006710031884722412) - rollout_corr/log_ppl_abs_diff:np.float64(0.001697185740340501) - rollout_corr/log_ppl_diff_max:np.float64(0.006523817777633667) - rollout_corr/log_ppl_diff_min:np.float64(-0.005903065204620361) - rollout_corr/ppl_ratio:np.float64(1.0006734766066074) - rollout_corr/chi2_token:np.float64(0.0018631014972925186) - rollout_corr/chi2_seq:np.float64(1.0500562572851777) - actor/pg_loss:np.float64(0.010553930653259158) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004439412439296575) - actor/ppo_kl:np.float64(6.877072929745154e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.86328125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.86328125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.48046875) - self_distillation/token_reweight_w_mean:np.float64(85283.25226895092) - self_distillation/token_reweight_w_std:np.float64(1114820.7389926258) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0022825964260846) - self_distillation/token_reweight_w_clip_frac:np.float64(0.21137651632307097) - self_distillation/empty_target_batch:np.float64(0.13671875) - actor/grad_norm:np.float64(0.4761762246489525) - perf/mfu/actor:np.float64(0.05262875043012334) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.41241455078125) - actor/lr:np.float64(1e-06) - training/global_step:11 - training/epoch:0 - critic/score/mean:0.48046875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.48046875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.003749731695279479 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.003749731695279479 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:400.2890625 - response_length/max:1
|
||||
step:12 - global_seqlen/min:19921 - global_seqlen/max:27357 - global_seqlen/minmax_diff:7436 - global_seqlen/balanced_min:23213 - global_seqlen/balanced_max:23297 - global_seqlen/mean:23233.875 - actor/entropy:0.5037309527397156 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.16815555095672607 - training/rollout_probs_diff_mean:0.004780740477144718 - training/rollout_probs_diff_std:0.009941385127604008 - training/rollout_actor_probs_pearson_corr:0.9994201064109802 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.85546875 - self_distillation/correct_sample_fraction:0.46484375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.85546875 - rollout_corr/rollout_is_mean:0.999710738658905 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.6984993635560386e-05 - rollout_corr/rollout_is_max:1.6297073364257812 - rollout_corr/rollout_is_min:0.3698853552341461 - rollout_corr/rollout_is_std:0.035432368516922 - rollout_corr/rollout_is_eff_sample_size:0.9987455864550007 - rollout_corr/rollout_is_seq_mean:0.9997212886810303 - rollout_corr/rollout_is_seq_std:0.0021238892804831266 - rollout_corr/rollout_is_seq_max:1.0065120458602905 - rollout_corr/rollout_is_seq_min:0.9934496879577637 - rollout_corr/rollout_is_seq_max_deviation:0.006550312042236328 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7103436021134257) - rollout_corr/training_log_ppl:np.float64(0.5244237306469586) - rollout_corr/kl:np.float64(0.0010640324547948055) - rollout_corr/k3_kl:np.float64(0.0007240620033144296) - rollout_corr/rollout_ppl:np.float64(1.7084603938274086) - rollout_corr/rollout_log_ppl:np.float64(0.5233596969628707) - rollout_corr/log_ppl_diff:np.float64(0.0010640336840879172) - rollout_corr/log_ppl_abs_diff:np.float64(0.001916373468702659) - rollout_corr/log_ppl_diff_max:np.float64(0.007551968097686768) - rollout_corr/log_ppl_diff_min:np.float64(-0.006095826625823975) - rollout_corr/ppl_ratio:np.float64(1.001067083561793) - rollout_corr/chi2_token:np.float64(0.0007682361174374819) - rollout_corr/chi2_seq:np.float64(0.6293782293796539) - actor/pg_loss:np.float64(0.007974700187332928) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00048827441059984267) - actor/ppo_kl:np.float64(0.00010889819034787251) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.85546875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.85546875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.46484375) - self_distillation/token_reweight_w_mean:np.float64(5987.44374671299) - self_distillation/token_reweight_w_std:np.float64(110518.26587392716) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0042279572226107) - self_distillation/token_reweight_w_clip_frac:np.float64(0.17869807308306918) - self_distillation/empty_target_batch:np.float64(0.14453125) - actor/grad_norm:np.float64(0.39314158260822296) - perf/mfu/actor:np.float64(0.05490032892199064) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.42404556274414) - actor/lr:np.float64(1e-06) - training/global_step:12 - training/epoch:0 - critic/score/mean:0.46484375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.46484375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0070593878626823425 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0070593878626823425 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:459.96484375 - response_leng
|
||||
step:13 - global_seqlen/min:18814 - global_seqlen/max:23875 - global_seqlen/minmax_diff:5061 - global_seqlen/balanced_min:21556 - global_seqlen/balanced_max:21562 - global_seqlen/mean:21560.5 - actor/entropy:0.540065348148346 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2304113507270813 - training/rollout_probs_diff_mean:0.005129465367645025 - training/rollout_probs_diff_std:0.010314878076314926 - training/rollout_actor_probs_pearson_corr:0.9993942975997925 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.796875 - self_distillation/correct_sample_fraction:0.375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.796875 - rollout_corr/rollout_is_mean:0.9999523758888245 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:9.84678445092868e-06 - rollout_corr/rollout_is_max:1.643043875694275 - rollout_corr/rollout_is_min:0.3293769061565399 - rollout_corr/rollout_is_std:0.03673160448670387 - rollout_corr/rollout_is_eff_sample_size:0.9986523694012616 - rollout_corr/rollout_is_seq_mean:0.9999930262565613 - rollout_corr/rollout_is_seq_std:0.0019651646725833416 - rollout_corr/rollout_is_seq_max:1.0069950819015503 - rollout_corr/rollout_is_seq_min:0.9950668215751648 - rollout_corr/rollout_is_seq_max_deviation:0.006995081901550293 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.75340425549075) - rollout_corr/training_log_ppl:np.float64(0.5529010602040216) - rollout_corr/kl:np.float64(0.0008567688251606143) - rollout_corr/k3_kl:np.float64(0.0008690143870353495) - rollout_corr/rollout_ppl:np.float64(1.7518776911310852) - rollout_corr/rollout_log_ppl:np.float64(0.5520442884298973) - rollout_corr/log_ppl_diff:np.float64(0.0008567717741243541) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018868366605602205) - rollout_corr/log_ppl_diff_max:np.float64(0.00990372896194458) - rollout_corr/log_ppl_diff_min:np.float64(-0.005962073802947998) - rollout_corr/ppl_ratio:np.float64(1.0008597834967077) - rollout_corr/chi2_token:np.float64(0.0017287565860897303) - rollout_corr/chi2_seq:np.float64(0.7434897748753428) - actor/pg_loss:np.float64(0.009388480219058692) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.001301017089645029) - actor/ppo_kl:np.float64(0.0001421315267471357) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.796875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.796875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.375) - self_distillation/token_reweight_w_mean:np.float64(1917.8817092673853) - self_distillation/token_reweight_w_std:np.float64(43886.33348266815) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0036602555774152) - self_distillation/token_reweight_w_clip_frac:np.float64(0.2082537584064994) - self_distillation/empty_target_batch:np.float64(0.203125) - actor/grad_norm:np.float64(0.43746187537908554) - perf/mfu/actor:np.float64(0.051655857866616195) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.21465253829956) - actor/lr:np.float64(1e-06) - training/global_step:13 - training/epoch:0 - critic/score/mean:0.375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.003404525574296713 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.003404525574296713 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:396.703125 - response_length/max:972.0 - response_length/min:141.0
|
||||
step:14 - global_seqlen/min:17920 - global_seqlen/max:21796 - global_seqlen/minmax_diff:3876 - global_seqlen/balanced_min:20032 - global_seqlen/balanced_max:20152 - global_seqlen/mean:20054.25 - actor/entropy:0.5020557641983032 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18802747130393982 - training/rollout_probs_diff_mean:0.005197192542254925 - training/rollout_probs_diff_std:0.01069364883005619 - training/rollout_actor_probs_pearson_corr:0.9992953538894653 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.76953125 - self_distillation/correct_sample_fraction:0.40625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.76953125 - rollout_corr/rollout_is_mean:1.000118374824524 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6423780918121338 - rollout_corr/rollout_is_min:0.6023645997047424 - rollout_corr/rollout_is_std:0.03736548125743866 - rollout_corr/rollout_is_eff_sample_size:0.9986062429080049 - rollout_corr/rollout_is_seq_mean:1.0001424551010132 - rollout_corr/rollout_is_seq_std:0.002069722395390272 - rollout_corr/rollout_is_seq_max:1.0075104236602783 - rollout_corr/rollout_is_seq_min:0.9939577579498291 - rollout_corr/rollout_is_seq_max_deviation:0.00751042366027832 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7107822382822633) - rollout_corr/training_log_ppl:np.float64(0.5240909514250234) - rollout_corr/kl:np.float64(0.0007750868314850834) - rollout_corr/k3_kl:np.float64(0.0011061159723340097) - rollout_corr/rollout_ppl:np.float64(1.7093983073718846) - rollout_corr/rollout_log_ppl:np.float64(0.5233158656046726) - rollout_corr/log_ppl_diff:np.float64(0.000775085820350796) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019063884974457324) - rollout_corr/log_ppl_diff_max:np.float64(0.009017229080200195) - rollout_corr/log_ppl_diff_min:np.float64(-0.007934510707855225) - rollout_corr/ppl_ratio:np.float64(1.0007783074397594) - rollout_corr/chi2_token:np.float64(0.0027928389608860016) - rollout_corr/chi2_seq:np.float64(0.6327834515832365) - actor/pg_loss:np.float64(0.009141277987509966) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0023054060643517005) - actor/ppo_kl:np.float64(0.00017033189165172047) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.76953125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.76953125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.40625) - self_distillation/token_reweight_w_mean:np.float64(51790.3643477282) - self_distillation/token_reweight_w_std:np.float64(710581.8369147222) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0032089522574097) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1833768331562169) - self_distillation/empty_target_batch:np.float64(0.23046875) - actor/grad_norm:np.float64(0.5500108152627945) - perf/mfu/actor:np.float64(0.04745480566246875) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.28086853027344) - actor/lr:np.float64(1e-06) - training/global_step:14 - training/epoch:0 - critic/score/mean:0.40625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.40625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:1.6309666534652933e-05 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:1.6309666534652933e-05 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:359.2578125 - response_length/max:1234.0 - response_length/min
|
||||
step:15 - global_seqlen/min:17292 - global_seqlen/max:23462 - global_seqlen/minmax_diff:6170 - global_seqlen/balanced_min:20153 - global_seqlen/balanced_max:20223 - global_seqlen/mean:20173.0 - actor/entropy:0.4989277422428131 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23447632789611816 - training/rollout_probs_diff_mean:0.005335042718797922 - training/rollout_probs_diff_std:0.011061806231737137 - training/rollout_actor_probs_pearson_corr:0.99927818775177 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.76953125 - self_distillation/correct_sample_fraction:0.46875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.76953125 - rollout_corr/rollout_is_mean:0.9997453093528748 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5897420644760132 - rollout_corr/rollout_is_min:0.5000474452972412 - rollout_corr/rollout_is_std:0.03860742598772049 - rollout_corr/rollout_is_eff_sample_size:0.9985110313307147 - rollout_corr/rollout_is_seq_mean:0.9997358918190002 - rollout_corr/rollout_is_seq_std:0.0024659933988004923 - rollout_corr/rollout_is_seq_max:1.0084636211395264 - rollout_corr/rollout_is_seq_min:0.9930576682090759 - rollout_corr/rollout_is_seq_max_deviation:0.008463621139526367 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7213433790020645) - rollout_corr/training_log_ppl:np.float64(0.530678162467666) - rollout_corr/kl:np.float64(0.0010814671836847367) - rollout_corr/k3_kl:np.float64(0.000888961919315534) - rollout_corr/rollout_ppl:np.float64(1.7193831540644169) - rollout_corr/rollout_log_ppl:np.float64(0.5295966898556799) - rollout_corr/log_ppl_diff:np.float64(0.0010814726119861007) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019996468909084797) - rollout_corr/log_ppl_diff_max:np.float64(0.012681245803833008) - rollout_corr/log_ppl_diff_min:np.float64(-0.0078064799308776855) - rollout_corr/ppl_ratio:np.float64(1.0010852627456188) - rollout_corr/chi2_token:np.float64(0.0014137444086372852) - rollout_corr/chi2_seq:np.float64(0.5360429051797837) - actor/pg_loss:np.float64(0.007955617504194379) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006377134768627002) - actor/ppo_kl:np.float64(1.769121835337728e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.76953125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.76953125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.46875) - self_distillation/token_reweight_w_mean:np.float64(8754.255326425191) - self_distillation/token_reweight_w_std:np.float64(117017.50059508975) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.001879719318822) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1550164521613624) - self_distillation/empty_target_batch:np.float64(0.23046875) - actor/grad_norm:np.float64(0.4198375418782234) - perf/mfu/actor:np.float64(0.04879610399473919) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.12841796875) - actor/lr:np.float64(1e-06) - training/global_step:15 - training/epoch:0 - critic/score/mean:0.46875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.46875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00847491342574358 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00847491342574358 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:343.5 - response_length/max:1036.0 - response_length/min:108.0 - respo
|
||||
step:16 - global_seqlen/min:17365 - global_seqlen/max:24430 - global_seqlen/minmax_diff:7065 - global_seqlen/balanced_min:20809 - global_seqlen/balanced_max:20815 - global_seqlen/mean:20814.0 - actor/entropy:0.4525703191757202 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.44003361463546753 - training/rollout_probs_diff_mean:0.0051152328960597515 - training/rollout_probs_diff_std:0.011277093552052975 - training/rollout_actor_probs_pearson_corr:0.9991934299468994 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.82421875 - self_distillation/correct_sample_fraction:0.47265625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.82421875 - rollout_corr/rollout_is_mean:1.000049352645874 - rollout_corr/rollout_is_ratio_fraction_high:1.0565463526290841e-05 - rollout_corr/rollout_is_ratio_fraction_low:6.339278479572386e-05 - rollout_corr/rollout_is_max:2.1190927028656006 - rollout_corr/rollout_is_min:0.16761082410812378 - rollout_corr/rollout_is_std:0.03827483579516411 - rollout_corr/rollout_is_eff_sample_size:0.9985372988785958 - rollout_corr/rollout_is_seq_mean:1.0000265836715698 - rollout_corr/rollout_is_seq_std:0.0022894139401614666 - rollout_corr/rollout_is_seq_max:1.0063250064849854 - rollout_corr/rollout_is_seq_min:0.992612361907959 - rollout_corr/rollout_is_seq_max_deviation:0.007387638092041016 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6002246476709843) - rollout_corr/training_log_ppl:np.float64(0.46000494839972816) - rollout_corr/kl:np.float64(0.0006955834787873982) - rollout_corr/k3_kl:np.float64(0.0009704830327450509) - rollout_corr/rollout_ppl:np.float64(1.5991095416247845) - rollout_corr/rollout_log_ppl:np.float64(0.45930936612421647) - rollout_corr/log_ppl_diff:np.float64(0.0006955822755116969) - rollout_corr/log_ppl_abs_diff:np.float64(0.0021134040143806487) - rollout_corr/log_ppl_diff_max:np.float64(0.007766246795654297) - rollout_corr/log_ppl_diff_min:np.float64(-0.005490928888320923) - rollout_corr/ppl_ratio:np.float64(1.0006991014815867) - rollout_corr/chi2_token:np.float64(0.0025813004467636347) - rollout_corr/chi2_seq:np.float64(3.1832106430083513) - actor/pg_loss:np.float64(0.009350149193778634) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0011533425140441977) - actor/ppo_kl:np.float64(-8.012323441874969e-05) - actor/pg_clipfrac_lower:np.float64(1.187310044770129e-05) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.82421875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.82421875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.47265625) - self_distillation/token_reweight_w_mean:np.float64(35795.58769591758) - self_distillation/token_reweight_w_std:np.float64(482687.72877264104) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0030966859776527) - self_distillation/token_reweight_w_clip_frac:np.float64(0.18662144461995922) - self_distillation/empty_target_batch:np.float64(0.17578125) - actor/grad_norm:np.float64(0.49581966549158096) - perf/mfu/actor:np.float64(0.05050208271241423) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.1108741760254) - actor/lr:np.float64(1e-06) - training/global_step:16 - training/epoch:0 - critic/score/mean:0.47265625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.47265625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.01058791484683752 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.01058791484683752 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_leng
|
||||
step:17 - global_seqlen/min:22601 - global_seqlen/max:25800 - global_seqlen/minmax_diff:3199 - global_seqlen/balanced_min:23884 - global_seqlen/balanced_max:24138 - global_seqlen/mean:23919.75 - actor/entropy:0.5084906220436096 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23104846477508545 - training/rollout_probs_diff_mean:0.004854131024330854 - training/rollout_probs_diff_std:0.010198368690907955 - training/rollout_actor_probs_pearson_corr:0.9993425011634827 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.6796875 - self_distillation/correct_sample_fraction:0.41015625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.6796875 - rollout_corr/rollout_is_mean:1.0001150369644165 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.8287475288379937e-05 - rollout_corr/rollout_is_max:1.7099158763885498 - rollout_corr/rollout_is_min:0.4197293221950531 - rollout_corr/rollout_is_std:0.036216843873262405 - rollout_corr/rollout_is_eff_sample_size:0.9986904150765131 - rollout_corr/rollout_is_seq_mean:1.0001122951507568 - rollout_corr/rollout_is_seq_std:0.0019326574401929975 - rollout_corr/rollout_is_seq_max:1.0060887336730957 - rollout_corr/rollout_is_seq_min:0.9933558106422424 - rollout_corr/rollout_is_seq_max_deviation:0.006644189357757568 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7412785273045301) - rollout_corr/training_log_ppl:np.float64(0.5394562017754652) - rollout_corr/kl:np.float64(0.00046429189457519726) - rollout_corr/k3_kl:np.float64(0.0007774811874128318) - rollout_corr/rollout_ppl:np.float64(1.7404227014631033) - rollout_corr/rollout_log_ppl:np.float64(0.538991907320451) - rollout_corr/log_ppl_diff:np.float64(0.0004642944550141692) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017179182032123208) - rollout_corr/log_ppl_diff_max:np.float64(0.007203280925750732) - rollout_corr/log_ppl_diff_min:np.float64(-0.006280422210693359) - rollout_corr/ppl_ratio:np.float64(1.0004668079782277) - rollout_corr/chi2_token:np.float64(0.002199894981458783) - rollout_corr/chi2_seq:np.float64(2.1964850607328117) - actor/pg_loss:np.float64(0.006313061923719943) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00043986083028357825) - actor/ppo_kl:np.float64(-0.00013203885243839864) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.6796875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.6796875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.41015625) - self_distillation/token_reweight_w_mean:np.float64(1803.160887064878) - self_distillation/token_reweight_w_std:np.float64(34584.773948686314) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0014728792011738) - self_distillation/token_reweight_w_clip_frac:np.float64(0.12900599383283406) - self_distillation/empty_target_batch:np.float64(0.3203125) - actor/grad_norm:np.float64(0.38481806963682175) - perf/mfu/actor:np.float64(0.05734825677685727) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.12204360961914) - actor/lr:np.float64(1e-06) - training/global_step:17 - training/epoch:0 - critic/score/mean:0.41015625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.41015625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.005190751980990171 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.005190751980990171 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:414.2734375 - response_lengt
|
||||
step:18 - global_seqlen/min:17607 - global_seqlen/max:24855 - global_seqlen/minmax_diff:7248 - global_seqlen/balanced_min:20730 - global_seqlen/balanced_max:20735 - global_seqlen/mean:20733.125 - actor/entropy:0.4945189952850342 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.30840808153152466 - training/rollout_probs_diff_mean:0.005118089262396097 - training/rollout_probs_diff_std:0.010652771219611168 - training/rollout_actor_probs_pearson_corr:0.9993004202842712 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80078125 - self_distillation/correct_sample_fraction:0.49609375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.80078125 - rollout_corr/rollout_is_mean:0.9998429417610168 - rollout_corr/rollout_is_ratio_fraction_high:2.0853969544987194e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.032266139984131 - rollout_corr/rollout_is_min:0.5130106210708618 - rollout_corr/rollout_is_std:0.03713827580213547 - rollout_corr/rollout_is_eff_sample_size:0.9986221727100675 - rollout_corr/rollout_is_seq_mean:0.9997930526733398 - rollout_corr/rollout_is_seq_std:0.0022385723423212767 - rollout_corr/rollout_is_seq_max:1.007550835609436 - rollout_corr/rollout_is_seq_min:0.9926072359085083 - rollout_corr/rollout_is_seq_max_deviation:0.007550835609436035 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6632476961240172) - rollout_corr/training_log_ppl:np.float64(0.49761353904614225) - rollout_corr/kl:np.float64(0.0008254678116763614) - rollout_corr/k3_kl:np.float64(0.0008354976650934987) - rollout_corr/rollout_ppl:np.float64(1.6618767492473125) - rollout_corr/rollout_log_ppl:np.float64(0.49678807152668014) - rollout_corr/log_ppl_diff:np.float64(0.0008254675194621086) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019702696008607745) - rollout_corr/log_ppl_diff_max:np.float64(0.008582055568695068) - rollout_corr/log_ppl_diff_min:np.float64(-0.006475538015365601) - rollout_corr/ppl_ratio:np.float64(1.0008287394884974) - rollout_corr/chi2_token:np.float64(0.0016966825351119041) - rollout_corr/chi2_seq:np.float64(0.7993498002178967) - actor/pg_loss:np.float64(0.006876209634356201) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0009630748777453846) - actor/ppo_kl:np.float64(-0.00010461421089758005) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.80078125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.80078125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.49609375) - self_distillation/token_reweight_w_mean:np.float64(6797.09068774269) - self_distillation/token_reweight_w_std:np.float64(117938.16381233273) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0013849884271622) - self_distillation/token_reweight_w_clip_frac:np.float64(0.17171310502453707) - self_distillation/empty_target_batch:np.float64(0.19921875) - actor/grad_norm:np.float64(0.4298411011695862) - perf/mfu/actor:np.float64(0.05005069497056926) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.3214340209961) - actor/lr:np.float64(1e-06) - training/global_step:18 - training/epoch:0 - critic/score/mean:0.49609375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.49609375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.005354256834834814 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.005354256834834814 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:374.62890625 - response_le
|
||||
step:19 - global_seqlen/min:16539 - global_seqlen/max:26420 - global_seqlen/minmax_diff:9881 - global_seqlen/balanced_min:21542 - global_seqlen/balanced_max:21548 - global_seqlen/mean:21545.625 - actor/entropy:0.48952940106391907 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.21504220366477966 - training/rollout_probs_diff_mean:0.0051950011402368546 - training/rollout_probs_diff_std:0.010906816460192204 - training/rollout_actor_probs_pearson_corr:0.9992679953575134 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.7421875 - self_distillation/correct_sample_fraction:0.5 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7421875 - rollout_corr/rollout_is_mean:1.0002551078796387 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.993241310119629 - rollout_corr/rollout_is_min:0.508678138256073 - rollout_corr/rollout_is_std:0.03832463547587395 - rollout_corr/rollout_is_eff_sample_size:0.9985340896470543 - rollout_corr/rollout_is_seq_mean:1.000322699546814 - rollout_corr/rollout_is_seq_std:0.0022897543385624886 - rollout_corr/rollout_is_seq_max:1.007610559463501 - rollout_corr/rollout_is_seq_min:0.9937204718589783 - rollout_corr/rollout_is_seq_max_deviation:0.0076105594635009766 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6905396389774978) - rollout_corr/training_log_ppl:np.float64(0.5122466792818159) - rollout_corr/kl:np.float64(0.0005300049446929078) - rollout_corr/k3_kl:np.float64(0.0008489201253496503) - rollout_corr/rollout_ppl:np.float64(1.6896098447032273) - rollout_corr/rollout_log_ppl:np.float64(0.511716675653588) - rollout_corr/log_ppl_diff:np.float64(0.0005300036282278597) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018085126648657024) - rollout_corr/log_ppl_diff_max:np.float64(0.011858522891998291) - rollout_corr/log_ppl_diff_min:np.float64(-0.007662832736968994) - rollout_corr/ppl_ratio:np.float64(1.0005329437553883) - rollout_corr/chi2_token:np.float64(0.00233305711299181) - rollout_corr/chi2_seq:np.float64(0.9715974144637585) - actor/pg_loss:np.float64(0.006131203263066709) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005582695239354507) - actor/ppo_kl:np.float64(7.291653415109067e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.7421875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.7421875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.5) - self_distillation/token_reweight_w_mean:np.float64(19979.935181184206) - self_distillation/token_reweight_w_std:np.float64(310836.0626574742) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0012645116075873) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1186036104336381) - self_distillation/empty_target_batch:np.float64(0.2578125) - actor/grad_norm:np.float64(0.3470497876405716) - perf/mfu/actor:np.float64(0.051614476906136436) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.35358047485352) - actor/lr:np.float64(1e-06) - training/global_step:19 - training/epoch:0 - critic/score/mean:0.5 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.004195125307887793 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.004195125307887793 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:371.17578125 - response_length/max:894.0 - response_length/min:152.0 - response_length
|
||||
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 20}
|
||||
validation generation end
|
||||
[prompt] system
|
||||
|
||||
Given a question and four options, please select the right answer. Respond in the following format:
|
||||
<reasoning>
|
||||
...
|
||||
</reasoning>
|
||||
<answer>
|
||||
...
|
||||
</answer>
|
||||
|
||||
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
|
||||
user
|
||||
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||||
|
||||
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||||
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||||
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||||
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||||
Please reason step by step.
|
||||
assistant
|
||||
<think>
|
||||
|
||||
</think>
|
||||
|
||||
|
||||
[response] <reasoning>
|
||||
To determine the correct reactant for the synthesis of the product **"Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1"**, we must identify the structure of the product and match it with the most appropriate reactant that would lead to its formation through a plausible synthetic pathway.
|
||||
|
||||
The product has the following key structural features:
|
||||
- A **nitrogen-containing ring** (likely a pyrrole or imidazole-like structure).
|
||||
- A **chlorinated aromatic ring** (e.g., **Cl-phenyl**).
|
||||
- A **trifluoromethyl group** (C(F)(F)F).
|
||||
- A **nitrogen atom** bonded to a carbon chain that includes the **Cl-phenyl** and **trifluoromethyl** groups.
|
||||
|
||||
Now, let’s examine the options:
|
||||
|
||||
**Option A**:
|
||||
`O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1`
|
||||
This structure contains a **nitrogen ring** with the same **Cl-phenyl** and **trifluoromethyl** substituents as the product. The **carbonyl groups** and **nitrogen ring** suggest a possible **amide** or **nitrile** structure that could be a **precursor** to the product. It is the **most structurally similar** to the product and would likely be the **correct reactant** in a synthesis involving ring formation or substitution.
|
||||
|
||||
**Option B**:
|
||||
`O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1`
|
||||
This structure contains **two chlorines** and a **nitrogen ring**, but it lacks the **trifluoromethyl group** and the **correct arrangement of substituents** as in the product. It is **not a good match**.
|
||||
|
||||
**Option C**:
|
||||
`O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1`
|
||||
This structure includes a **trifluoromethyl group** and **Cl-phenyl**, but the **nitrogen ring** is **not directly bonded to the Cl-phenyl** or **trifluoromethyl**, and the **structure is more complex**. It might be a **byproduct** or **intermediate**, but not the **direct reactant**.
|
||||
|
||||
**Option D**:
|
||||
`O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F`
|
||||
This structure contains a **trifluoromethyl group**, but it **lacks the nitrogen ring** and the **Cl-phenyl** substituents. It is **structurally different** from the product.
|
||||
|
||||
**Conclusion**:
|
||||
Only **Option A** contains the **correct substituents** (Cl-phenyl, trifluoromethyl, and a nitrogen ring), and the **structure is most consistent** with the product. It is the **only plausible reactant** that would lead to the synthesis of the given product.
|
||||
|
||||
</reasoning>
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
[ground_truth] A
|
||||
[score] 1.0
|
||||
[acc] 1.0
|
||||
[pred] A
|
||||
[incorrect_format] 1
|
||||
[feedback]
|
||||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_20
|
||||
step:20 - global_seqlen/min:18853 - global_seqlen/max:23232 - global_seqlen/minmax_diff:4379 - global_seqlen/balanced_min:20703 - global_seqlen/balanced_max:20714 - global_seqlen/mean:20709.5 - actor/entropy:0.4852030277252197 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3635062277317047 - training/rollout_probs_diff_mean:0.0052037229761481285 - training/rollout_probs_diff_std:0.010991442948579788 - training/rollout_actor_probs_pearson_corr:0.9992546439170837 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - self_distillation/correct_sample_fraction:0.640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83203125 - rollout_corr/rollout_is_mean:1.0001426935195923 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.1487816411536187e-05 - rollout_corr/rollout_is_max:1.6168099641799927 - rollout_corr/rollout_is_min:0.4058627188205719 - rollout_corr/rollout_is_std:0.03768158331513405 - rollout_corr/rollout_is_eff_sample_size:0.9985827057725277 - rollout_corr/rollout_is_seq_mean:1.0002155303955078 - rollout_corr/rollout_is_seq_std:0.002177149523049593 - rollout_corr/rollout_is_seq_max:1.0087448358535767 - rollout_corr/rollout_is_seq_min:0.9949045181274414 - rollout_corr/rollout_is_seq_max_deviation:0.00874483585357666 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.65907796099782) - rollout_corr/training_log_ppl:np.float64(0.49622454700875096) - rollout_corr/kl:np.float64(0.0007081711968908166) - rollout_corr/k3_kl:np.float64(0.000836590931498904) - rollout_corr/rollout_ppl:np.float64(1.6578862462192774) - rollout_corr/rollout_log_ppl:np.float64(0.49551637563854456) - rollout_corr/log_ppl_diff:np.float64(0.0007081713702064008) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018356915388721973) - rollout_corr/log_ppl_diff_max:np.float64(0.007012069225311279) - rollout_corr/log_ppl_diff_min:np.float64(-0.006520330905914307) - rollout_corr/ppl_ratio:np.float64(1.0007108724676073) - rollout_corr/chi2_token:np.float64(0.001943168230354786) - rollout_corr/chi2_seq:np.float64(0.7969548560213298) - actor/pg_loss:np.float64(0.006117190117947757) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006000254938953731) - actor/ppo_kl:np.float64(0.0001683622528645401) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.83203125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.83203125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.640625) - self_distillation/token_reweight_w_mean:np.float64(23049.971880125348) - self_distillation/token_reweight_w_std:np.float64(317665.6890155331) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9982667530421168) - self_distillation/token_reweight_w_clip_frac:np.float64(0.15381890590651892) - self_distillation/empty_target_batch:np.float64(0.16796875) - actor/grad_norm:np.float64(0.3626445010304451) - perf/mfu/actor:np.float64(0.05021794625693002) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.33641052246094) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.5196428571428572) - val-aux/sciknoweval/reward/std@16:np.float64(0.2570318204515784) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.6278380952380953) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.21727863221866373) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.41209523809523807) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.2203367172556475) - val-aux/sciknoweval/reward/maj@2/mean:np.fl
|
||||
step:21 - global_seqlen/min:17376 - global_seqlen/max:26208 - global_seqlen/minmax_diff:8832 - global_seqlen/balanced_min:22075 - global_seqlen/balanced_max:22088 - global_seqlen/mean:22085.25 - actor/entropy:0.4656538665294647 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4102373421192169 - training/rollout_probs_diff_mean:0.004885641857981682 - training/rollout_probs_diff_std:0.010877853259444237 - training/rollout_actor_probs_pearson_corr:0.9992596507072449 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.84765625 - self_distillation/correct_sample_fraction:0.43359375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.84765625 - rollout_corr/rollout_is_mean:0.9998842477798462 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.879465500882361e-05 - rollout_corr/rollout_is_max:1.9200012683868408 - rollout_corr/rollout_is_min:0.25445759296417236 - rollout_corr/rollout_is_std:0.03666176274418831 - rollout_corr/rollout_is_eff_sample_size:0.9986574816202841 - rollout_corr/rollout_is_seq_mean:0.9998617172241211 - rollout_corr/rollout_is_seq_std:0.0022321075666695833 - rollout_corr/rollout_is_seq_max:1.0074141025543213 - rollout_corr/rollout_is_seq_min:0.9920339584350586 - rollout_corr/rollout_is_seq_max_deviation:0.007966041564941406 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6367876464501023) - rollout_corr/training_log_ppl:np.float64(0.47835954203037545) - rollout_corr/kl:np.float64(0.0009155909243130367) - rollout_corr/k3_kl:np.float64(0.0008391680780164279) - rollout_corr/rollout_ppl:np.float64(1.6352251395583153) - rollout_corr/rollout_log_ppl:np.float64(0.4774439476314001) - rollout_corr/log_ppl_diff:np.float64(0.0009155943989753723) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018019898561760783) - rollout_corr/log_ppl_diff_max:np.float64(0.008913397789001465) - rollout_corr/log_ppl_diff_min:np.float64(-0.006456807255744934) - rollout_corr/ppl_ratio:np.float64(1.0009184170048684) - rollout_corr/chi2_token:np.float64(0.001522636041045189) - rollout_corr/chi2_seq:np.float64(0.3302574837580323) - actor/pg_loss:np.float64(0.00977602880448103) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0008731294096833153) - actor/ppo_kl:np.float64(2.3401729779659597e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.84765625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.84765625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.43359375) - self_distillation/token_reweight_w_mean:np.float64(19834.05608875933) - self_distillation/token_reweight_w_std:np.float64(292799.855937861) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0034907914232463) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1998413290712051) - self_distillation/empty_target_batch:np.float64(0.15234375) - actor/grad_norm:np.float64(0.506427951157093) - perf/mfu/actor:np.float64(0.05293222627939937) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(195.2050437927246) - actor/lr:np.float64(1e-06) - training/global_step:21 - training/epoch:0 - critic/score/mean:0.43359375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.43359375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.00826286617666483 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00826286617666483 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:406.9765625 - response_length/max:11
|
||||
step:22 - global_seqlen/min:15780 - global_seqlen/max:20714 - global_seqlen/minmax_diff:4934 - global_seqlen/balanced_min:18793 - global_seqlen/balanced_max:18799 - global_seqlen/mean:18797.25 - actor/entropy:0.4407426416873932 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23966830968856812 - training/rollout_probs_diff_mean:0.0051012844778597355 - training/rollout_probs_diff_std:0.011158647015690804 - training/rollout_actor_probs_pearson_corr:0.9991984963417053 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.6796875 - self_distillation/correct_sample_fraction:0.4375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.6796875 - rollout_corr/rollout_is_mean:0.9999550580978394 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.586863398551941 - rollout_corr/rollout_is_min:0.6212320327758789 - rollout_corr/rollout_is_std:0.03741012141108513 - rollout_corr/rollout_is_eff_sample_size:0.9986023199750391 - rollout_corr/rollout_is_seq_mean:0.9999154806137085 - rollout_corr/rollout_is_seq_std:0.0023052047472447157 - rollout_corr/rollout_is_seq_max:1.0075262784957886 - rollout_corr/rollout_is_seq_min:0.9921414256095886 - rollout_corr/rollout_is_seq_max_deviation:0.007858574390411377 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6121524260379374) - rollout_corr/training_log_ppl:np.float64(0.46336599756614305) - rollout_corr/kl:np.float64(0.0012106716192334943) - rollout_corr/k3_kl:np.float64(0.001044533465915265) - rollout_corr/rollout_ppl:np.float64(1.6101201642304659) - rollout_corr/rollout_log_ppl:np.float64(0.4621553277829662) - rollout_corr/log_ppl_diff:np.float64(0.0012106697831768543) - rollout_corr/log_ppl_abs_diff:np.float64(0.0022202231630217284) - rollout_corr/log_ppl_diff_max:np.float64(0.011315524578094482) - rollout_corr/log_ppl_diff_min:np.float64(-0.007631242275238037) - rollout_corr/ppl_ratio:np.float64(1.0012149058748037) - rollout_corr/chi2_token:np.float64(0.001704103546217084) - rollout_corr/chi2_seq:np.float64(0.4757801443338394) - actor/pg_loss:np.float64(0.006979008438065648) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0011709280256582133) - actor/ppo_kl:np.float64(0.00036738665134805615) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.6796875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.6796875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.4375) - self_distillation/token_reweight_w_mean:np.float64(53386.58389568445) - self_distillation/token_reweight_w_std:np.float64(762098.1771182105) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0011509482283145) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1306488252885174) - self_distillation/empty_target_batch:np.float64(0.3203125) - actor/grad_norm:np.float64(0.4595351442694664) - perf/mfu/actor:np.float64(0.0457439323763552) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(195.23744583129883) - actor/lr:np.float64(1e-06) - training/global_step:22 - training/epoch:0 - critic/score/mean:0.4375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.4375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0024807543959468603 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0024807543959468603 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:331.8515625 - response_length/max:891.0 - response_length/min:134.0
|
||||
step:23 - global_seqlen/min:15655 - global_seqlen/max:22837 - global_seqlen/minmax_diff:7182 - global_seqlen/balanced_min:19794 - global_seqlen/balanced_max:19804 - global_seqlen/mean:19802.0 - actor/entropy:0.43892350792884827 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.22363799810409546 - training/rollout_probs_diff_mean:0.005129741504788399 - training/rollout_probs_diff_std:0.011249431408941746 - training/rollout_actor_probs_pearson_corr:0.9991777539253235 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.9296875 - self_distillation/correct_sample_fraction:0.69921875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.9296875 - rollout_corr/rollout_is_mean:0.9999939203262329 - rollout_corr/rollout_is_ratio_fraction_high:1.1452129911049269e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.1670188903808594 - rollout_corr/rollout_is_min:0.5688050389289856 - rollout_corr/rollout_is_std:0.03843645006418228 - rollout_corr/rollout_is_eff_sample_size:0.9985248186495871 - rollout_corr/rollout_is_seq_mean:0.9999755024909973 - rollout_corr/rollout_is_seq_std:0.0022403132170438766 - rollout_corr/rollout_is_seq_max:1.008310079574585 - rollout_corr/rollout_is_seq_min:0.9935483336448669 - rollout_corr/rollout_is_seq_max_deviation:0.008310079574584961 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.585592514835298) - rollout_corr/training_log_ppl:np.float64(0.45206161157693714) - rollout_corr/kl:np.float64(0.0010271025659691624) - rollout_corr/k3_kl:np.float64(0.0009754648472721783) - rollout_corr/rollout_ppl:np.float64(1.583907154854387) - rollout_corr/rollout_log_ppl:np.float64(0.45103450742317364) - rollout_corr/log_ppl_diff:np.float64(0.0010271041537635028) - rollout_corr/log_ppl_abs_diff:np.float64(0.0021158340969122946) - rollout_corr/log_ppl_diff_max:np.float64(0.008697479963302612) - rollout_corr/log_ppl_diff_min:np.float64(-0.007354050874710083) - rollout_corr/ppl_ratio:np.float64(1.0010308369528502) - rollout_corr/chi2_token:np.float64(0.0018195987213402987) - rollout_corr/chi2_seq:np.float64(1.1621388338971883) - actor/pg_loss:np.float64(0.007446382311172783) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.000959160078309651) - actor/ppo_kl:np.float64(0.00021888582077522756) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.9296875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.9296875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.69921875) - self_distillation/token_reweight_w_mean:np.float64(6609.447738413466) - self_distillation/token_reweight_w_std:np.float64(122546.46291852725) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9963779090903699) - self_distillation/token_reweight_w_clip_frac:np.float64(0.17003195008146577) - self_distillation/empty_target_batch:np.float64(0.0703125) - actor/grad_norm:np.float64(0.43443264812231064) - perf/mfu/actor:np.float64(0.04804480134893495) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(195.1064910888672) - actor/lr:np.float64(1e-06) - training/global_step:23 - training/epoch:0 - critic/score/mean:0.69921875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.69921875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0033225491642951965 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0033225491642951965 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:341.09375 - response_length/max
|
||||
step:24 - global_seqlen/min:18708 - global_seqlen/max:22807 - global_seqlen/minmax_diff:4099 - global_seqlen/balanced_min:20352 - global_seqlen/balanced_max:20355 - global_seqlen/mean:20353.75 - actor/entropy:0.48541581630706787 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.16695591807365417 - training/rollout_probs_diff_mean:0.005245619919151068 - training/rollout_probs_diff_std:0.01105139683932066 - training/rollout_actor_probs_pearson_corr:0.9992536306381226 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.82421875 - self_distillation/correct_sample_fraction:0.58984375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.82421875 - rollout_corr/rollout_is_mean:0.9997487664222717 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6379640102386475 - rollout_corr/rollout_is_min:0.55048668384552 - rollout_corr/rollout_is_std:0.038326967507600784 - rollout_corr/rollout_is_eff_sample_size:0.998532425609191 - rollout_corr/rollout_is_seq_mean:0.9996849894523621 - rollout_corr/rollout_is_seq_std:0.0021940250881016254 - rollout_corr/rollout_is_seq_max:1.0061367750167847 - rollout_corr/rollout_is_seq_min:0.9898834228515625 - rollout_corr/rollout_is_seq_max_deviation:0.0101165771484375 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6475044381804764) - rollout_corr/training_log_ppl:np.float64(0.4905120711773634) - rollout_corr/kl:np.float64(0.0011423102059779922) - rollout_corr/k3_kl:np.float64(0.0009413222521743592) - rollout_corr/rollout_ppl:np.float64(1.6456106547266245) - rollout_corr/rollout_log_ppl:np.float64(0.48936975869582966) - rollout_corr/log_ppl_diff:np.float64(0.001142312481533736) - rollout_corr/log_ppl_abs_diff:np.float64(0.002205758646596223) - rollout_corr/log_ppl_diff_max:np.float64(0.011655747890472412) - rollout_corr/log_ppl_diff_min:np.float64(-0.006167978048324585) - rollout_corr/ppl_ratio:np.float64(1.0011465374846011) - rollout_corr/chi2_token:np.float64(0.001514944015070796) - rollout_corr/chi2_seq:np.float64(0.9692259337753057) - actor/pg_loss:np.float64(0.006828168756328523) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0008421741476922762) - actor/ppo_kl:np.float64(5.722692924337025e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.82421875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.82421875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.58984375) - self_distillation/token_reweight_w_mean:np.float64(16169.235121513018) - self_distillation/token_reweight_w_std:np.float64(204586.64157430374) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000086387153715) - self_distillation/token_reweight_w_clip_frac:np.float64(0.14809982472797856) - self_distillation/empty_target_batch:np.float64(0.17578125) - actor/grad_norm:np.float64(0.38066764548420906) - perf/mfu/actor:np.float64(0.04935736118807353) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.85000133514404) - actor/lr:np.float64(1e-06) - training/global_step:24 - training/epoch:0 - critic/score/mean:0.58984375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.58984375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-8.236870780820027e-05 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-8.236870780820027e-05 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:355.6796875 - response_length/max:641.0 - respon
|
||||
step:25 - global_seqlen/min:16381 - global_seqlen/max:25188 - global_seqlen/minmax_diff:8807 - global_seqlen/balanced_min:21088 - global_seqlen/balanced_max:21097 - global_seqlen/mean:21093.125 - actor/entropy:0.4773712158203125 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.26223766803741455 - training/rollout_probs_diff_mean:0.005165540147572756 - training/rollout_probs_diff_std:0.01117149367928505 - training/rollout_actor_probs_pearson_corr:0.9992371797561646 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.86328125 - self_distillation/correct_sample_fraction:0.5546875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.86328125 - rollout_corr/rollout_is_mean:0.9999747276306152 - rollout_corr/rollout_is_ratio_fraction_high:1.0959985047520604e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.094183921813965 - rollout_corr/rollout_is_min:0.5062851309776306 - rollout_corr/rollout_is_std:0.038087502121925354 - rollout_corr/rollout_is_eff_sample_size:0.9985513246511216 - rollout_corr/rollout_is_seq_mean:1.0000536441802979 - rollout_corr/rollout_is_seq_std:0.0022864702623337507 - rollout_corr/rollout_is_seq_max:1.0067065954208374 - rollout_corr/rollout_is_seq_min:0.9937045574188232 - rollout_corr/rollout_is_seq_max_deviation:0.006706595420837402 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.669403042178601) - rollout_corr/training_log_ppl:np.float64(0.4993552331288811) - rollout_corr/kl:np.float64(0.0006479631488460758) - rollout_corr/k3_kl:np.float64(0.0008453664596572708) - rollout_corr/rollout_ppl:np.float64(1.668324178084731) - rollout_corr/rollout_log_ppl:np.float64(0.4987072692310903) - rollout_corr/log_ppl_diff:np.float64(0.0006479638977907598) - rollout_corr/log_ppl_abs_diff:np.float64(0.001830984663683921) - rollout_corr/log_ppl_diff_max:np.float64(0.008455544710159302) - rollout_corr/log_ppl_diff_min:np.float64(-0.007198035717010498) - rollout_corr/ppl_ratio:np.float64(1.0006508629303426) - rollout_corr/chi2_token:np.float64(0.002105830702930689) - rollout_corr/chi2_seq:np.float64(0.6466955123469234) - actor/pg_loss:np.float64(0.007964683580212295) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.000689070585394802) - actor/ppo_kl:np.float64(-6.814886182837654e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.86328125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.86328125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.5546875) - self_distillation/token_reweight_w_mean:np.float64(26990.30881792493) - self_distillation/token_reweight_w_std:np.float64(421488.5579388592) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.002008835086599) - self_distillation/token_reweight_w_clip_frac:np.float64(0.15447656423202716) - self_distillation/empty_target_batch:np.float64(0.13671875) - actor/grad_norm:np.float64(0.4783755913376808) - perf/mfu/actor:np.float64(0.0494590320538676) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.90256881713867) - actor/lr:np.float64(1e-06) - training/global_step:25 - training/epoch:0 - critic/score/mean:0.5546875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5546875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.02120209112763405 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.02120209112763405 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:356.41015625 - response_length/max:956.0
|
||||
step:26 - global_seqlen/min:18171 - global_seqlen/max:23062 - global_seqlen/minmax_diff:4891 - global_seqlen/balanced_min:20385 - global_seqlen/balanced_max:20394 - global_seqlen/mean:20391.25 - actor/entropy:0.5068786144256592 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.17874199151992798 - training/rollout_probs_diff_mean:0.005472649820148945 - training/rollout_probs_diff_std:0.011403916403651237 - training/rollout_actor_probs_pearson_corr:0.9991965889930725 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.74609375 - self_distillation/correct_sample_fraction:0.6328125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.74609375 - rollout_corr/rollout_is_mean:1.0000735521316528 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.292263525305316e-05 - rollout_corr/rollout_is_max:1.8139365911483765 - rollout_corr/rollout_is_min:0.43659546971321106 - rollout_corr/rollout_is_std:0.039478376507759094 - rollout_corr/rollout_is_eff_sample_size:0.9984440018610561 - rollout_corr/rollout_is_seq_mean:1.0000500679016113 - rollout_corr/rollout_is_seq_std:0.0023560679983347654 - rollout_corr/rollout_is_seq_max:1.0076462030410767 - rollout_corr/rollout_is_seq_min:0.9908225536346436 - rollout_corr/rollout_is_seq_max_deviation:0.009177446365356445 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6881171250715852) - rollout_corr/training_log_ppl:np.float64(0.510973722091876) - rollout_corr/kl:np.float64(0.0007455622756609159) - rollout_corr/k3_kl:np.float64(0.0008864333188967066) - rollout_corr/rollout_ppl:np.float64(1.6867859680205584) - rollout_corr/rollout_log_ppl:np.float64(0.5102281617000699) - rollout_corr/log_ppl_diff:np.float64(0.000745560391806066) - rollout_corr/log_ppl_abs_diff:np.float64(0.002050930866971612) - rollout_corr/log_ppl_diff_max:np.float64(0.00910806655883789) - rollout_corr/log_ppl_diff_min:np.float64(-0.006127357482910156) - rollout_corr/ppl_ratio:np.float64(1.0007490692660213) - rollout_corr/chi2_token:np.float64(0.0020302317570894957) - rollout_corr/chi2_seq:np.float64(1.0315312510356307) - actor/pg_loss:np.float64(0.005001060431823134) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00048566531631877297) - actor/ppo_kl:np.float64(-2.1595971896104516e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.74609375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.74609375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6328125) - self_distillation/token_reweight_w_mean:np.float64(13624.06982974615) - self_distillation/token_reweight_w_std:np.float64(206433.80913565995) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9992130598984659) - self_distillation/token_reweight_w_clip_frac:np.float64(0.09108548518270254) - self_distillation/empty_target_batch:np.float64(0.25390625) - actor/grad_norm:np.float64(0.3169581815600395) - perf/mfu/actor:np.float64(0.04962560549786612) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.88824081420898) - actor/lr:np.float64(1e-06) - training/global_step:26 - training/epoch:0 - critic/score/mean:0.6328125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6328125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0027206302620470524 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0027206302620470524 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:340.8203125 - response_length/m
|
||||
step:27 - global_seqlen/min:17480 - global_seqlen/max:23456 - global_seqlen/minmax_diff:5976 - global_seqlen/balanced_min:20067 - global_seqlen/balanced_max:20076 - global_seqlen/mean:20073.875 - actor/entropy:0.4675213694572449 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23105892539024353 - training/rollout_probs_diff_mean:0.005088143516331911 - training/rollout_probs_diff_std:0.010994603857398033 - training/rollout_actor_probs_pearson_corr:0.9992254972457886 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.85546875 - self_distillation/correct_sample_fraction:0.51171875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.85546875 - rollout_corr/rollout_is_mean:0.9997977018356323 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6167669296264648 - rollout_corr/rollout_is_min:0.5378821492195129 - rollout_corr/rollout_is_std:0.03790003061294556 - rollout_corr/rollout_is_eff_sample_size:0.9985651130827546 - rollout_corr/rollout_is_seq_mean:0.9997107982635498 - rollout_corr/rollout_is_seq_std:0.002149628708139062 - rollout_corr/rollout_is_seq_max:1.007951021194458 - rollout_corr/rollout_is_seq_min:0.9923518896102905 - rollout_corr/rollout_is_seq_max_deviation:0.007951021194458008 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6248419084586203) - rollout_corr/training_log_ppl:np.float64(0.47598731669131666) - rollout_corr/kl:np.float64(0.0010493483772009427) - rollout_corr/k3_kl:np.float64(0.0008248598063573809) - rollout_corr/rollout_ppl:np.float64(1.623152432963252) - rollout_corr/rollout_log_ppl:np.float64(0.47493796655908227) - rollout_corr/log_ppl_diff:np.float64(0.0010493501322343946) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018501168815419078) - rollout_corr/log_ppl_diff_max:np.float64(0.007712453603744507) - rollout_corr/log_ppl_diff_min:np.float64(-0.004121840000152588) - rollout_corr/ppl_ratio:np.float64(1.0010521123185754) - rollout_corr/chi2_token:np.float64(0.0011714110150933266) - rollout_corr/chi2_seq:np.float64(0.32961429841816425) - actor/pg_loss:np.float64(0.008322581415995955) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0008014981349333539) - actor/ppo_kl:np.float64(4.5028834745330926e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.85546875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.85546875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.51171875) - self_distillation/token_reweight_w_mean:np.float64(30577.34729726822) - self_distillation/token_reweight_w_std:np.float64(521484.4462188269) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0020933467894793) - self_distillation/token_reweight_w_clip_frac:np.float64(0.16031731336261146) - self_distillation/empty_target_batch:np.float64(0.14453125) - actor/grad_norm:np.float64(0.42759493738412857) - perf/mfu/actor:np.float64(0.048367516590069666) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.80517578125) - actor/lr:np.float64(1e-06) - training/global_step:27 - training/epoch:0 - critic/score/mean:0.51171875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.51171875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0014940870460122824 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0014940870460122824 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:369.62109375 - response_length/max:943.0 - resp
|
||||
step:28 - global_seqlen/min:18201 - global_seqlen/max:23560 - global_seqlen/minmax_diff:5359 - global_seqlen/balanced_min:20410 - global_seqlen/balanced_max:20497 - global_seqlen/mean:20425.125 - actor/entropy:0.4737909138202667 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2294958233833313 - training/rollout_probs_diff_mean:0.0052576144225895405 - training/rollout_probs_diff_std:0.011254720389842987 - training/rollout_actor_probs_pearson_corr:0.999194860458374 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83984375 - self_distillation/correct_sample_fraction:0.59765625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83984375 - rollout_corr/rollout_is_mean:1.0001050233840942 - rollout_corr/rollout_is_ratio_fraction_high:1.0676346391846891e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:3.8465442657470703 - rollout_corr/rollout_is_min:0.6126700639724731 - rollout_corr/rollout_is_std:0.03862826153635979 - rollout_corr/rollout_is_eff_sample_size:0.9985103182038877 - rollout_corr/rollout_is_seq_mean:1.0001041889190674 - rollout_corr/rollout_is_seq_std:0.002421688986942172 - rollout_corr/rollout_is_seq_max:1.00888991355896 - rollout_corr/rollout_is_seq_min:0.9913090467453003 - rollout_corr/rollout_is_seq_max_deviation:0.008889913558959961 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6370238508097827) - rollout_corr/training_log_ppl:np.float64(0.4820473712461535) - rollout_corr/kl:np.float64(0.0006596640439546775) - rollout_corr/k3_kl:np.float64(0.000909588395757055) - rollout_corr/rollout_ppl:np.float64(1.6358301592990756) - rollout_corr/rollout_log_ppl:np.float64(0.48138770982041024) - rollout_corr/log_ppl_diff:np.float64(0.000659661425743252) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018744217813946307) - rollout_corr/log_ppl_diff_max:np.float64(0.007977485656738281) - rollout_corr/log_ppl_diff_min:np.float64(-0.0074624717235565186) - rollout_corr/ppl_ratio:np.float64(1.0006625556852669) - rollout_corr/chi2_token:np.float64(0.0023598361294716597) - rollout_corr/chi2_seq:np.float64(0.9705478034447879) - actor/pg_loss:np.float64(0.006433643866330385) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006121892383816885) - actor/ppo_kl:np.float64(-2.0443672603676077e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.83984375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.83984375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.59765625) - self_distillation/token_reweight_w_mean:np.float64(22090.04282637965) - self_distillation/token_reweight_w_std:np.float64(363460.48439942) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0007038170006126) - self_distillation/token_reweight_w_clip_frac:np.float64(0.12272839152137749) - self_distillation/empty_target_batch:np.float64(0.16015625) - actor/grad_norm:np.float64(0.4047693833708763) - perf/mfu/actor:np.float64(0.0494901405269842) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.81372833251953) - actor/lr:np.float64(1e-06) - training/global_step:28 - training/epoch:0 - critic/score/mean:0.59765625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.59765625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.004289222415536642 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.004289222415536642 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:365.87890625 - response_length/
|
||||
step:29 - global_seqlen/min:17728 - global_seqlen/max:24689 - global_seqlen/minmax_diff:6961 - global_seqlen/balanced_min:21875 - global_seqlen/balanced_max:21882 - global_seqlen/mean:21878.875 - actor/entropy:0.45051109790802 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.25539880990982056 - training/rollout_probs_diff_mean:0.004729867912828922 - training/rollout_probs_diff_std:0.010498874820768833 - training/rollout_actor_probs_pearson_corr:0.9993234872817993 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.734375 - self_distillation/correct_sample_fraction:0.48046875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.734375 - rollout_corr/rollout_is_mean:0.9999346733093262 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.9196990251657553e-05 - rollout_corr/rollout_is_max:1.6618187427520752 - rollout_corr/rollout_is_min:0.48319488763809204 - rollout_corr/rollout_is_std:0.03575139120221138 - rollout_corr/rollout_is_eff_sample_size:0.998723350704955 - rollout_corr/rollout_is_seq_mean:0.9999228119850159 - rollout_corr/rollout_is_seq_std:0.00180347659625113 - rollout_corr/rollout_is_seq_max:1.0053354501724243 - rollout_corr/rollout_is_seq_min:0.9929618835449219 - rollout_corr/rollout_is_seq_max_deviation:0.007038116455078125 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5977162146009505) - rollout_corr/training_log_ppl:np.float64(0.4593473710701801) - rollout_corr/kl:np.float64(0.0006978295002575408) - rollout_corr/k3_kl:np.float64(0.0007481996697720206) - rollout_corr/rollout_ppl:np.float64(1.5965832527726889) - rollout_corr/rollout_log_ppl:np.float64(0.45864954427815974) - rollout_corr/log_ppl_diff:np.float64(0.0006978267920203507) - rollout_corr/log_ppl_abs_diff:np.float64(0.001622283656615764) - rollout_corr/log_ppl_diff_max:np.float64(0.006626904010772705) - rollout_corr/log_ppl_diff_min:np.float64(-0.00486445426940918) - rollout_corr/ppl_ratio:np.float64(1.0007000237237662) - rollout_corr/chi2_token:np.float64(0.001576458802446723) - rollout_corr/chi2_seq:np.float64(0.6520844814367592) - actor/pg_loss:np.float64(0.0060147366020828485) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005360728382584057) - actor/ppo_kl:np.float64(-6.641209190716424e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.734375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.734375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.48046875) - self_distillation/token_reweight_w_mean:np.float64(26070.862267469056) - self_distillation/token_reweight_w_std:np.float64(478879.7805042105) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0025979080237448) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1094071113911923) - self_distillation/empty_target_batch:np.float64(0.265625) - actor/grad_norm:np.float64(0.3129398711025715) - perf/mfu/actor:np.float64(0.052564897191348076) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.9274024963379) - actor/lr:np.float64(1e-06) - training/global_step:29 - training/epoch:0 - critic/score/mean:0.48046875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.48046875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.01068192534148693 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.01068192534148693 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:406.96484375 - response_length/max:856.0 - resp
|
||||
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 30}
|
||||
validation generation end
|
||||
[prompt] system
|
||||
|
||||
Given a question and four options, please select the right answer. Respond in the following format:
|
||||
<reasoning>
|
||||
...
|
||||
</reasoning>
|
||||
<answer>
|
||||
...
|
||||
</answer>
|
||||
|
||||
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
|
||||
user
|
||||
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||||
|
||||
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||||
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||||
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||||
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||||
Please reason step by step.
|
||||
assistant
|
||||
<think>
|
||||
|
||||
</think>
|
||||
|
||||
|
||||
[response] <reasoning>
|
||||
To determine the correct reactant for the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1", we need to analyze the structure of the product and identify the key functional groups and substituents that must be present in the reactant.
|
||||
|
||||
The product has the following key features:
|
||||
- A nitrogen-containing ring (likely a pyrrole or similar heterocycle).
|
||||
- A benzene ring with a chlorine substituent.
|
||||
- A trifluoromethyl group (C(F)(F)F) attached to the benzene ring.
|
||||
- A substituent group (likely a nitrogen-containing chain) that connects to the ring.
|
||||
|
||||
Now, let's examine each option:
|
||||
|
||||
**Option A:**
|
||||
O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||||
This option contains the trifluoromethyl group (C(F)(F)F) and the chlorinated benzene ring. It also contains a nitrogen-containing chain and a ring structure similar to the product. This option closely matches the structure of the product and is a strong candidate.
|
||||
|
||||
**Option B:**
|
||||
O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||||
This option has a chlorine substituent on the ring, but it lacks the trifluoromethyl group and the correct nitrogen-containing chain. It also has a different ring structure than the product. Therefore, it is not the correct reactant.
|
||||
|
||||
**Option C:**
|
||||
O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||||
This option contains the trifluoromethyl group and the chlorinated benzene ring. However, the structure is more complex and includes an oxygen-containing side chain that is not present in the product. It introduces an unnecessary oxygen group, making it incompatible with the product structure.
|
||||
|
||||
**Option D:**
|
||||
O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||||
This option contains the trifluoromethyl group and a benzene ring. However, it lacks the correct nitrogen-containing chain and the chlorinated benzene ring. The structure is not aligned with the product, so it is not the correct reactant.
|
||||
|
||||
**Conclusion:**
|
||||
Only **Option A** contains all the necessary functional groups and substituents present in the product, making it the only correct reactant for the synthesis of the given product.
|
||||
</reasoning>
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
[ground_truth] A
|
||||
[score] 1.0
|
||||
[acc] 1.0
|
||||
[pred] A
|
||||
[incorrect_format] 1
|
||||
[feedback]
|
||||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_30
|
||||
step:30 - global_seqlen/min:18031 - global_seqlen/max:24561 - global_seqlen/minmax_diff:6530 - global_seqlen/balanced_min:20880 - global_seqlen/balanced_max:20888 - global_seqlen/mean:20884.625 - actor/entropy:0.4658919870853424 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3171748220920563 - training/rollout_probs_diff_mean:0.005262547638267279 - training/rollout_probs_diff_std:0.01141402218490839 - training/rollout_actor_probs_pearson_corr:0.9991620779037476 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.66796875 - self_distillation/correct_sample_fraction:0.44921875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.66796875 - rollout_corr/rollout_is_mean:0.9999656677246094 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:3.412619844311848e-05 - rollout_corr/rollout_is_max:1.9128998517990112 - rollout_corr/rollout_is_min:0.2724512219429016 - rollout_corr/rollout_is_std:0.03873920068144798 - rollout_corr/rollout_is_eff_sample_size:0.9985014042044956 - rollout_corr/rollout_is_seq_mean:1.0000505447387695 - rollout_corr/rollout_is_seq_std:0.002504864940419793 - rollout_corr/rollout_is_seq_max:1.0067648887634277 - rollout_corr/rollout_is_seq_min:0.9914875626564026 - rollout_corr/rollout_is_seq_max_deviation:0.008512437343597412 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6606428353115916) - rollout_corr/training_log_ppl:np.float64(0.49546970275696367) - rollout_corr/kl:np.float64(0.0007864780202773147) - rollout_corr/k3_kl:np.float64(0.000911510776063551) - rollout_corr/rollout_ppl:np.float64(1.659348614513874) - rollout_corr/rollout_log_ppl:np.float64(0.4946832267160062) - rollout_corr/log_ppl_diff:np.float64(0.0007864760409574956) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019896051089745015) - rollout_corr/log_ppl_diff_max:np.float64(0.009873658418655396) - rollout_corr/log_ppl_diff_min:np.float64(-0.007493495941162109) - rollout_corr/ppl_ratio:np.float64(1.0007900123018771) - rollout_corr/chi2_token:np.float64(0.002100849524140358) - rollout_corr/chi2_seq:np.float64(0.6980828407686204) - actor/pg_loss:np.float64(0.0044937689090147614) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004526598731899867) - actor/ppo_kl:np.float64(2.1002829232230624e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.66796875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.66796875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.44921875) - self_distillation/token_reweight_w_mean:np.float64(16658.25212058681) - self_distillation/token_reweight_w_std:np.float64(262596.2304097906) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.002752567641437) - self_distillation/token_reweight_w_clip_frac:np.float64(0.09863085992401466) - self_distillation/empty_target_batch:np.float64(0.33203125) - actor/grad_norm:np.float64(0.3747030645608902) - perf/mfu/actor:np.float64(0.050935440136856504) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(194.77979278564453) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.606845238095238) - val-aux/sciknoweval/reward/std@16:np.float64(0.2138942013523591) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.6934285714285715) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.16904326809342193) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.521195238095238) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.1926339687939083) - val-aux/sciknoweval/reward/maj@2/mean:np.f
|
||||
step:31 - global_seqlen/min:19249 - global_seqlen/max:25029 - global_seqlen/minmax_diff:5780 - global_seqlen/balanced_min:21607 - global_seqlen/balanced_max:21619 - global_seqlen/mean:21615.375 - actor/entropy:0.4569137990474701 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.303112655878067 - training/rollout_probs_diff_mean:0.004967319313436747 - training/rollout_probs_diff_std:0.010901066474616528 - training/rollout_actor_probs_pearson_corr:0.9992502331733704 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - self_distillation/correct_sample_fraction:0.6015625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83203125 - rollout_corr/rollout_is_mean:1.000017762184143 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.816590428352356 - rollout_corr/rollout_is_min:0.5354147553443909 - rollout_corr/rollout_is_std:0.03712382912635803 - rollout_corr/rollout_is_eff_sample_size:0.9986238370470686 - rollout_corr/rollout_is_seq_mean:1.0000417232513428 - rollout_corr/rollout_is_seq_std:0.002106655156239867 - rollout_corr/rollout_is_seq_max:1.0081474781036377 - rollout_corr/rollout_is_seq_min:0.9948439002037048 - rollout_corr/rollout_is_seq_max_deviation:0.008147478103637695 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6235803924500942) - rollout_corr/training_log_ppl:np.float64(0.4749105938244611) - rollout_corr/kl:np.float64(0.0008353461136003659) - rollout_corr/k3_kl:np.float64(0.00080985213742224) - rollout_corr/rollout_ppl:np.float64(1.6222249972634017) - rollout_corr/rollout_log_ppl:np.float64(0.4740752446523402) - rollout_corr/log_ppl_diff:np.float64(0.0008353491721209139) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017292899719905108) - rollout_corr/log_ppl_diff_max:np.float64(0.006912916898727417) - rollout_corr/log_ppl_diff_min:np.float64(-0.0048788487911224365) - rollout_corr/ppl_ratio:np.float64(1.0008378000929952) - rollout_corr/chi2_token:np.float64(0.0016067922115325928) - rollout_corr/chi2_seq:np.float64(0.6984529327601194) - actor/pg_loss:np.float64(0.005961017799563706) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00028992991656195954) - actor/ppo_kl:np.float64(0.0001396356336407223) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.83203125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.83203125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6015625) - self_distillation/token_reweight_w_mean:np.float64(9105.852561026346) - self_distillation/token_reweight_w_std:np.float64(160757.80174179428) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9988879463635385) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1225835490913596) - self_distillation/empty_target_batch:np.float64(0.16796875) - actor/grad_norm:np.float64(0.33824220299720764) - perf/mfu/actor:np.float64(0.05203885094538403) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(196.26508331298828) - actor/lr:np.float64(1e-06) - training/global_step:31 - training/epoch:0 - critic/score/mean:0.6015625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6015625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.001895678462460637 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.001895678462460637 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:395.63671875 - response_length/max:1088.0 - response_le
|
||||
step:32 - global_seqlen/min:17172 - global_seqlen/max:25152 - global_seqlen/minmax_diff:7980 - global_seqlen/balanced_min:20878 - global_seqlen/balanced_max:20886 - global_seqlen/mean:20883.0 - actor/entropy:0.4830181300640106 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3013932406902313 - training/rollout_probs_diff_mean:0.00495893694460392 - training/rollout_probs_diff_std:0.010702594183385372 - training/rollout_actor_probs_pearson_corr:0.9992987513542175 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.55078125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8125 - rollout_corr/rollout_is_mean:1.0001959800720215 - rollout_corr/rollout_is_ratio_fraction_high:1.0022450624092016e-05 - rollout_corr/rollout_is_ratio_fraction_low:3.0067350962781347e-05 - rollout_corr/rollout_is_max:2.244924545288086 - rollout_corr/rollout_is_min:0.39147886633872986 - rollout_corr/rollout_is_std:0.037390995770692825 - rollout_corr/rollout_is_eff_sample_size:0.998604459752836 - rollout_corr/rollout_is_seq_mean:1.0002524852752686 - rollout_corr/rollout_is_seq_std:0.0021919296123087406 - rollout_corr/rollout_is_seq_max:1.0080853700637817 - rollout_corr/rollout_is_seq_min:0.9943554401397705 - rollout_corr/rollout_is_seq_max_deviation:0.008085370063781738 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6529409568756819) - rollout_corr/training_log_ppl:np.float64(0.4933903770870529) - rollout_corr/kl:np.float64(0.0006465979992356097) - rollout_corr/k3_kl:np.float64(0.0008415120931317688) - rollout_corr/rollout_ppl:np.float64(1.651906920131296) - rollout_corr/rollout_log_ppl:np.float64(0.49274377603433095) - rollout_corr/log_ppl_diff:np.float64(0.0006466010527219623) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017735535220708698) - rollout_corr/log_ppl_diff_max:np.float64(0.007363229990005493) - rollout_corr/log_ppl_diff_min:np.float64(-0.010225236415863037) - rollout_corr/ppl_ratio:np.float64(1.000649468274787) - rollout_corr/chi2_token:np.float64(0.0021671864669770002) - rollout_corr/chi2_seq:np.float64(1.2125774417072535) - actor/pg_loss:np.float64(0.007539922837167978) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005248926790955011) - actor/ppo_kl:np.float64(0.00015378757747441796) - actor/pg_clipfrac_lower:np.float64(1.2849506674683653e-05) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.55078125) - self_distillation/token_reweight_w_mean:np.float64(28962.190404822817) - self_distillation/token_reweight_w_std:np.float64(594865.4367179495) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9994088341481984) - self_distillation/token_reweight_w_clip_frac:np.float64(0.15152001177193597) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.4039623737335205) - perf/mfu/actor:np.float64(0.050354520500870484) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(195.79799795150757) - actor/lr:np.float64(1e-06) - training/global_step:32 - training/epoch:0 - critic/score/mean:0.55078125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.55078125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.005901970434933901 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.005901970434933901 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:389.75 - respon
|
||||
step:33 - global_seqlen/min:16273 - global_seqlen/max:24765 - global_seqlen/minmax_diff:8492 - global_seqlen/balanced_min:20358 - global_seqlen/balanced_max:20368 - global_seqlen/mean:20365.75 - actor/entropy:0.5308820605278015 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2607361674308777 - training/rollout_probs_diff_mean:0.005311037413775921 - training/rollout_probs_diff_std:0.010893904604017735 - training/rollout_actor_probs_pearson_corr:0.9993165731430054 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80078125 - self_distillation/correct_sample_fraction:0.60546875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.80078125 - rollout_corr/rollout_is_mean:1.0000280141830444 - rollout_corr/rollout_is_ratio_fraction_high:2.2798258214606903e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.1399129107303452e-05 - rollout_corr/rollout_is_max:2.4295754432678223 - rollout_corr/rollout_is_min:0.4068605899810791 - rollout_corr/rollout_is_std:0.038521647453308105 - rollout_corr/rollout_is_eff_sample_size:0.9985184003675729 - rollout_corr/rollout_is_seq_mean:1.0000193119049072 - rollout_corr/rollout_is_seq_std:0.0023478458169847727 - rollout_corr/rollout_is_seq_max:1.0072718858718872 - rollout_corr/rollout_is_seq_min:0.9926888346672058 - rollout_corr/rollout_is_seq_max_deviation:0.0073111653327941895 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7451868052594364) - rollout_corr/training_log_ppl:np.float64(0.5472010709927417) - rollout_corr/kl:np.float64(0.000750623744014689) - rollout_corr/k3_kl:np.float64(0.0009170059764755933) - rollout_corr/rollout_ppl:np.float64(1.7438560076989233) - rollout_corr/rollout_log_ppl:np.float64(0.5464504445553757) - rollout_corr/log_ppl_diff:np.float64(0.0007506264373660088) - rollout_corr/log_ppl_abs_diff:np.float64(0.0020804237574338913) - rollout_corr/log_ppl_diff_max:np.float64(0.011185050010681152) - rollout_corr/log_ppl_diff_min:np.float64(-0.008495688438415527) - rollout_corr/ppl_ratio:np.float64(1.0007542204111814) - rollout_corr/chi2_token:np.float64(0.0021686023101210594) - rollout_corr/chi2_seq:np.float64(1.0599001871887594) - actor/pg_loss:np.float64(0.0046818595146760345) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.000626342507530353) - actor/ppo_kl:np.float64(-7.711236621688045e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.80078125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.80078125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.60546875) - self_distillation/token_reweight_w_mean:np.float64(406.9383929811884) - self_distillation/token_reweight_w_std:np.float64(6888.8986393570085) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0006252727471292) - self_distillation/token_reweight_w_clip_frac:np.float64(0.10766920002060942) - self_distillation/empty_target_batch:np.float64(0.19921875) - actor/grad_norm:np.float64(0.35059650987386703) - perf/mfu/actor:np.float64(0.04906215016751242) - perf/max_memory_allocated_gb:np.float64(130.98335647583008) - perf/max_memory_reserved_gb:np.float64(140.86328125) - perf/cpu_memory_used_gb:np.float64(195.48283004760742) - actor/lr:np.float64(1e-06) - training/global_step:33 - training/epoch:0 - critic/score/mean:0.60546875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.60546875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0038529057055711746 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0038529057055711746 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:342.
|
||||
step:34 - global_seqlen/min:19407 - global_seqlen/max:28954 - global_seqlen/minmax_diff:9547 - global_seqlen/balanced_min:20764 - global_seqlen/balanced_max:28104 - global_seqlen/mean:21689.625 - actor/entropy:0.4711659848690033 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.26104798913002014 - training/rollout_probs_diff_mean:0.004729302600026131 - training/rollout_probs_diff_std:0.010431492701172829 - training/rollout_actor_probs_pearson_corr:0.9993188977241516 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.85546875 - self_distillation/correct_sample_fraction:0.59765625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.85546875 - rollout_corr/rollout_is_mean:0.9998348355293274 - rollout_corr/rollout_is_ratio_fraction_high:1.0172008842346258e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.372540235519409 - rollout_corr/rollout_is_min:0.5585512518882751 - rollout_corr/rollout_is_std:0.03565623611211777 - rollout_corr/rollout_is_eff_sample_size:0.9987297716220098 - rollout_corr/rollout_is_seq_mean:0.9998084902763367 - rollout_corr/rollout_is_seq_std:0.0021642311476171017 - rollout_corr/rollout_is_seq_max:1.0065914392471313 - rollout_corr/rollout_is_seq_min:0.9937243461608887 - rollout_corr/rollout_is_seq_max_deviation:0.006591439247131348 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.732406364288181) - rollout_corr/training_log_ppl:np.float64(0.53354305683024) - rollout_corr/kl:np.float64(0.0010742717632254706) - rollout_corr/k3_kl:np.float64(0.0007844316083662761) - rollout_corr/rollout_ppl:np.float64(1.7304212134331465) - rollout_corr/rollout_log_ppl:np.float64(0.5324687818174425) - rollout_corr/log_ppl_diff:np.float64(0.0010742750127974432) - rollout_corr/log_ppl_abs_diff:np.float64(0.002004527148528723) - rollout_corr/log_ppl_diff_max:np.float64(0.010212719440460205) - rollout_corr/log_ppl_diff_min:np.float64(-0.0042626261711120605) - rollout_corr/ppl_ratio:np.float64(1.0010776109993458) - rollout_corr/chi2_token:np.float64(0.000989705789834261) - rollout_corr/chi2_seq:np.float64(0.5540260763373226) - actor/pg_loss:np.float64(0.005981583381071687) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004929028409605962) - actor/ppo_kl:np.float64(0.00015171869483765477) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.85546875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.85546875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.59765625) - self_distillation/token_reweight_w_mean:np.float64(11402.142899290891) - self_distillation/token_reweight_w_std:np.float64(223975.93910462921) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0007735511753708) - self_distillation/token_reweight_w_clip_frac:np.float64(0.13933006773004308) - self_distillation/empty_target_batch:np.float64(0.14453125) - actor/grad_norm:np.float64(0.3575190082192421) - perf/mfu/actor:np.float64(0.04880687442998337) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(195.40957641601562) - actor/lr:np.float64(1e-06) - training/global_step:34 - training/epoch:0 - critic/score/mean:0.59765625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.59765625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.05439735949039459 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.05439735949039459 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:384.01953125 - response_length/m
|
||||
step:35 - global_seqlen/min:19463 - global_seqlen/max:21625 - global_seqlen/minmax_diff:2162 - global_seqlen/balanced_min:20560 - global_seqlen/balanced_max:20574 - global_seqlen/mean:20569.125 - actor/entropy:0.4701376259326935 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.27064740657806396 - training/rollout_probs_diff_mean:0.00488562835380435 - training/rollout_probs_diff_std:0.010647308081388474 - training/rollout_actor_probs_pearson_corr:0.999289870262146 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.75 - self_distillation/correct_sample_fraction:0.5078125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.75 - rollout_corr/rollout_is_mean:0.9999282956123352 - rollout_corr/rollout_is_ratio_fraction_high:1.0713864867284428e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.1437816619873047 - rollout_corr/rollout_is_min:0.5562102794647217 - rollout_corr/rollout_is_std:0.036710504442453384 - rollout_corr/rollout_is_eff_sample_size:0.9986539149503298 - rollout_corr/rollout_is_seq_mean:0.9999589323997498 - rollout_corr/rollout_is_seq_std:0.0022408063523471355 - rollout_corr/rollout_is_seq_max:1.0093075037002563 - rollout_corr/rollout_is_seq_min:0.9939238429069519 - rollout_corr/rollout_is_seq_max_deviation:0.009307503700256348 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.643273327499628) - rollout_corr/training_log_ppl:np.float64(0.48507256753509864) - rollout_corr/kl:np.float64(0.0008028178892658389) - rollout_corr/k3_kl:np.float64(0.0007829796426221947) - rollout_corr/rollout_ppl:np.float64(1.641896395944059) - rollout_corr/rollout_log_ppl:np.float64(0.4842697436106391) - rollout_corr/log_ppl_diff:np.float64(0.000802823924459517) - rollout_corr/log_ppl_abs_diff:np.float64(0.002036678488366306) - rollout_corr/log_ppl_diff_max:np.float64(0.008617639541625977) - rollout_corr/log_ppl_diff_min:np.float64(-0.010091245174407959) - rollout_corr/ppl_ratio:np.float64(1.0008062841370702) - rollout_corr/chi2_token:np.float64(0.0015425898600369692) - rollout_corr/chi2_seq:np.float64(1.5650525961536914) - actor/pg_loss:np.float64(0.0024570953100919724) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004893736586382147) - actor/ppo_kl:np.float64(4.371891989851662e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.75) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.75) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.5078125) - self_distillation/token_reweight_w_mean:np.float64(28164.23297266336) - self_distillation/token_reweight_w_std:np.float64(438345.5305433142) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0024336164351553) - self_distillation/token_reweight_w_clip_frac:np.float64(0.09937885942053981) - self_distillation/empty_target_batch:np.float64(0.25) - actor/grad_norm:np.float64(0.2858409211039543) - perf/mfu/actor:np.float64(0.04990645011261699) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(195.3169288635254) - actor/lr:np.float64(1e-06) - training/global_step:35 - training/epoch:0 - critic/score/mean:0.5078125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5078125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.004351168405264616 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.004351168405264616 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:364.59765625 - response_length/max:846.0 - response_length/min:119
|
||||
step:36 - global_seqlen/min:15946 - global_seqlen/max:23368 - global_seqlen/minmax_diff:7422 - global_seqlen/balanced_min:21144 - global_seqlen/balanced_max:21148 - global_seqlen/mean:21146.25 - actor/entropy:0.4903087913990021 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18462920188903809 - training/rollout_probs_diff_mean:0.005082453601062298 - training/rollout_probs_diff_std:0.010714489035308361 - training/rollout_actor_probs_pearson_corr:0.9993000030517578 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.828125 - self_distillation/correct_sample_fraction:0.5625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.828125 - rollout_corr/rollout_is_mean:1.0000641345977783 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6886519193649292 - rollout_corr/rollout_is_min:0.6301243901252747 - rollout_corr/rollout_is_std:0.037293627858161926 - rollout_corr/rollout_is_eff_sample_size:0.9986113546547832 - rollout_corr/rollout_is_seq_mean:1.000097632408142 - rollout_corr/rollout_is_seq_std:0.002074406947940588 - rollout_corr/rollout_is_seq_max:1.0062254667282104 - rollout_corr/rollout_is_seq_min:0.9938981533050537 - rollout_corr/rollout_is_seq_max_deviation:0.006225466728210449 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6528221243061125) - rollout_corr/training_log_ppl:np.float64(0.4915998738142662) - rollout_corr/kl:np.float64(0.0008901074607035042) - rollout_corr/k3_kl:np.float64(0.000780510345350649) - rollout_corr/rollout_ppl:np.float64(1.6513647134415805) - rollout_corr/rollout_log_ppl:np.float64(0.49070976028451696) - rollout_corr/log_ppl_diff:np.float64(0.0008901135297492146) - rollout_corr/log_ppl_abs_diff:np.float64(0.001717470120638609) - rollout_corr/log_ppl_diff_max:np.float64(0.006238460540771484) - rollout_corr/log_ppl_diff_min:np.float64(-0.005200505256652832) - rollout_corr/ppl_ratio:np.float64(1.0008924398571253) - rollout_corr/chi2_token:np.float64(0.0013305631000548601) - rollout_corr/chi2_seq:np.float64(0.19096036697737873) - actor/pg_loss:np.float64(0.0061755290953442454) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0007742183429400029) - actor/ppo_kl:np.float64(0.0002702638142206837) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.828125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.828125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.5625) - self_distillation/token_reweight_w_mean:np.float64(5297.740719528403) - self_distillation/token_reweight_w_std:np.float64(105322.54200576013) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0011047541629523) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1423137153033167) - self_distillation/empty_target_batch:np.float64(0.171875) - actor/grad_norm:np.float64(0.3517342582345009) - perf/mfu/actor:np.float64(0.051460493101783154) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(195.31007385253906) - actor/lr:np.float64(1e-06) - training/global_step:36 - training/epoch:0 - critic/score/mean:0.5625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0027716271579265594 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0027716271579265594 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:382.8203125 - response_length/max:718.0 - response_length/min:171.0 - r
|
||||
step:37 - global_seqlen/min:18794 - global_seqlen/max:24933 - global_seqlen/minmax_diff:6139 - global_seqlen/balanced_min:22097 - global_seqlen/balanced_max:22101 - global_seqlen/mean:22100.0 - actor/entropy:0.4772471785545349 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.8962392210960388 - training/rollout_probs_diff_mean:0.004988708533346653 - training/rollout_probs_diff_std:0.011156635358929634 - training/rollout_actor_probs_pearson_corr:0.9992755651473999 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.7734375 - self_distillation/correct_sample_fraction:0.54296875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7734375 - rollout_corr/rollout_is_mean:0.9999436140060425 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:3.1014806154416874e-05 - rollout_corr/rollout_is_max:1.5039196014404297 - rollout_corr/rollout_is_min:0.0692535936832428 - rollout_corr/rollout_is_std:0.03678025305271149 - rollout_corr/rollout_is_eff_sample_size:0.9986488027678234 - rollout_corr/rollout_is_seq_mean:0.9999807476997375 - rollout_corr/rollout_is_seq_std:0.002165940823033452 - rollout_corr/rollout_is_seq_max:1.0075201988220215 - rollout_corr/rollout_is_seq_min:0.9936956763267517 - rollout_corr/rollout_is_seq_max_deviation:0.007520198822021484 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6701486743986607) - rollout_corr/training_log_ppl:np.float64(0.5012833332875744) - rollout_corr/kl:np.float64(0.0008904281232275935) - rollout_corr/k3_kl:np.float64(0.0009389091193838794) - rollout_corr/rollout_ppl:np.float64(1.6685718595981598) - rollout_corr/rollout_log_ppl:np.float64(0.5003929063677788) - rollout_corr/log_ppl_diff:np.float64(0.0008904269197955728) - rollout_corr/log_ppl_abs_diff:np.float64(0.001979760592803359) - rollout_corr/log_ppl_diff_max:np.float64(0.02197730541229248) - rollout_corr/log_ppl_diff_min:np.float64(-0.00723642110824585) - rollout_corr/ppl_ratio:np.float64(1.0008947688620538) - rollout_corr/chi2_token:np.float64(0.0017904054839164019) - rollout_corr/chi2_seq:np.float64(0.7441312682349235) - actor/pg_loss:np.float64(0.0051974470261484385) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006028023444741848) - actor/ppo_kl:np.float64(0.00011284644474807148) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.7734375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.7734375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.54296875) - self_distillation/token_reweight_w_mean:np.float64(17486.880942083197) - self_distillation/token_reweight_w_std:np.float64(248698.93032729038) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0008525024168193) - self_distillation/token_reweight_w_clip_frac:np.float64(0.10978465108200908) - self_distillation/empty_target_batch:np.float64(0.2265625) - actor/grad_norm:np.float64(0.33661027252674103) - perf/mfu/actor:np.float64(0.053537470413758295) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(195.20091247558594) - actor/lr:np.float64(1e-06) - training/global_step:37 - training/epoch:0 - critic/score/mean:0.54296875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.54296875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00370756140910089 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00370756140910089 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:377.84375 - response_length/max:858.
|
||||
step:38 - global_seqlen/min:17401 - global_seqlen/max:23183 - global_seqlen/minmax_diff:5782 - global_seqlen/balanced_min:20614 - global_seqlen/balanced_max:20619 - global_seqlen/mean:20617.125 - actor/entropy:0.4591502249240875 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.21379658579826355 - training/rollout_probs_diff_mean:0.005224332213401794 - training/rollout_probs_diff_std:0.011259016580879688 - training/rollout_actor_probs_pearson_corr:0.9992035627365112 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.7734375 - self_distillation/correct_sample_fraction:0.6484375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7734375 - rollout_corr/rollout_is_mean:1.0001171827316284 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.0498313460848294e-05 - rollout_corr/rollout_is_max:1.9823302030563354 - rollout_corr/rollout_is_min:0.3327290415763855 - rollout_corr/rollout_is_std:0.03824835270643234 - rollout_corr/rollout_is_eff_sample_size:0.9985396761004448 - rollout_corr/rollout_is_seq_mean:1.0001174211502075 - rollout_corr/rollout_is_seq_std:0.002267003059387207 - rollout_corr/rollout_is_seq_max:1.007875919342041 - rollout_corr/rollout_is_seq_min:0.9941402077674866 - rollout_corr/rollout_is_seq_max_deviation:0.007875919342041016 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.597476608119905) - rollout_corr/training_log_ppl:np.float64(0.4596842197352089) - rollout_corr/kl:np.float64(0.0007120158846873892) - rollout_corr/k3_kl:np.float64(0.0008242313452342387) - rollout_corr/rollout_ppl:np.float64(1.5963355526328087) - rollout_corr/rollout_log_ppl:np.float64(0.45897220366168767) - rollout_corr/log_ppl_diff:np.float64(0.0007120160735212266) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018952106474898756) - rollout_corr/log_ppl_diff_max:np.float64(0.006997823715209961) - rollout_corr/log_ppl_diff_min:np.float64(-0.006651520729064941) - rollout_corr/ppl_ratio:np.float64(1.0007150201126933) - rollout_corr/chi2_token:np.float64(0.001860827673226595) - rollout_corr/chi2_seq:np.float64(0.6607308345846832) - actor/pg_loss:np.float64(0.003047003992833197) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00027509686378834886) - actor/ppo_kl:np.float64(6.150551095629453e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.7734375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.7734375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6484375) - self_distillation/token_reweight_w_mean:np.float64(24324.67800112092) - self_distillation/token_reweight_w_std:np.float64(319731.90878068085) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0001283828169107) - self_distillation/token_reweight_w_clip_frac:np.float64(0.07227780664106831) - self_distillation/empty_target_batch:np.float64(0.2265625) - actor/grad_norm:np.float64(0.25013007409870625) - perf/mfu/actor:np.float64(0.05036454480977212) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(195.1811408996582) - actor/lr:np.float64(1e-06) - training/global_step:38 - training/epoch:0 - critic/score/mean:0.6484375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6484375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0011094711953774095 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0011094711953774095 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:381.12890625 - response_length/max:7
|
||||
step:39 - global_seqlen/min:16098 - global_seqlen/max:24192 - global_seqlen/minmax_diff:8094 - global_seqlen/balanced_min:21469 - global_seqlen/balanced_max:21484 - global_seqlen/mean:21479.5 - actor/entropy:0.4425225853919983 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.17422929406166077 - training/rollout_probs_diff_mean:0.004969022236764431 - training/rollout_probs_diff_std:0.010954358614981174 - training/rollout_actor_probs_pearson_corr:0.9992186427116394 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80078125 - self_distillation/correct_sample_fraction:0.6015625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.80078125 - rollout_corr/rollout_is_mean:1.0003507137298584 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6815071105957031 - rollout_corr/rollout_is_min:0.5040706396102905 - rollout_corr/rollout_is_std:0.03746902570128441 - rollout_corr/rollout_is_eff_sample_size:0.9985991103255378 - rollout_corr/rollout_is_seq_mean:1.0004005432128906 - rollout_corr/rollout_is_seq_std:0.0023794230073690414 - rollout_corr/rollout_is_seq_max:1.012694001197815 - rollout_corr/rollout_is_seq_min:0.9931012392044067 - rollout_corr/rollout_is_seq_max_deviation:0.012694001197814941 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6068592290394008) - rollout_corr/training_log_ppl:np.float64(0.4637760063924361) - rollout_corr/kl:np.float64(0.0005125066249966892) - rollout_corr/k3_kl:np.float64(0.0008192024238837803) - rollout_corr/rollout_ppl:np.float64(1.6059974883683026) - rollout_corr/rollout_log_ppl:np.float64(0.46326349792070687) - rollout_corr/log_ppl_diff:np.float64(0.0005125084717292339) - rollout_corr/log_ppl_abs_diff:np.float64(0.001737583806971088) - rollout_corr/log_ppl_diff_max:np.float64(0.006271004676818848) - rollout_corr/log_ppl_diff_min:np.float64(-0.01115313172340393) - rollout_corr/ppl_ratio:np.float64(1.0005151890218258) - rollout_corr/chi2_token:np.float64(0.0022532844450324774) - rollout_corr/chi2_seq:np.float64(1.0010677098762244) - actor/pg_loss:np.float64(0.004201453877612948) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00037899658491369337) - actor/ppo_kl:np.float64(0.000158508868715046) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.80078125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.80078125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6015625) - self_distillation/token_reweight_w_mean:np.float64(21191.4891625538) - self_distillation/token_reweight_w_std:np.float64(305819.6095345903) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0013274641241878) - self_distillation/token_reweight_w_clip_frac:np.float64(0.08719093064428307) - self_distillation/empty_target_batch:np.float64(0.19921875) - actor/grad_norm:np.float64(0.2567609027028084) - perf/mfu/actor:np.float64(0.05189741811293595) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(195.25640106201172) - actor/lr:np.float64(1e-06) - training/global_step:39 - training/epoch:0 - critic/score/mean:0.6015625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6015625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.00024856385425664485 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00024856385425664485 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:388.953125 - response_length/max:1077.0 - response_
|
||||
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 40}
|
||||
validation generation end
|
||||
[prompt] system
|
||||
|
||||
Given a question and four options, please select the right answer. Respond in the following format:
|
||||
<reasoning>
|
||||
...
|
||||
</reasoning>
|
||||
<answer>
|
||||
...
|
||||
</answer>
|
||||
|
||||
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
|
||||
user
|
||||
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||||
|
||||
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||||
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||||
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||||
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||||
Please reason step by step.
|
||||
assistant
|
||||
<think>
|
||||
|
||||
</think>
|
||||
|
||||
|
||||
[response] <reasoning>
|
||||
To determine the correct reactant for the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1", we must examine the structure of the product and compare it to the structures of the given options.
|
||||
|
||||
The product contains a nitrogen-containing ring with a substituent group that includes a chlorinated benzene ring and a trifluoromethyl group. Specifically, the product has a structure where a nitrogen atom is part of a ring that is connected to a carbon chain leading to a chlorinated benzene ring with a trifluoromethyl group.
|
||||
|
||||
Now, let's analyze the options:
|
||||
|
||||
- **Option A**: Contains a structure with a nitrogen ring and a chlorinated benzene ring, along with a trifluoromethyl group. It closely resembles the structure of the target product, especially in terms of the nitrogen-containing ring and the substituent groups.
|
||||
|
||||
- **Option B**: Contains a chlorine atom directly attached to the nitrogen ring, which is not present in the target product. This makes it an incorrect match.
|
||||
|
||||
- **Option C**: Features a hydroxyl group and a different substituent pattern, which does not match the target structure.
|
||||
|
||||
- **Option D**: Includes a double bond and a fluorinated carbon chain, but not the specific chlorinated benzene ring and nitrogen ring arrangement of the target product.
|
||||
|
||||
Only **Option A** contains the correct structural elements that match the target product.
|
||||
|
||||
</reasoning>
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
[ground_truth] A
|
||||
[score] 1.0
|
||||
[acc] 1.0
|
||||
[pred] A
|
||||
[incorrect_format] 1
|
||||
[feedback]
|
||||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_40
|
||||
step:40 - global_seqlen/min:18616 - global_seqlen/max:26413 - global_seqlen/minmax_diff:7797 - global_seqlen/balanced_min:22648 - global_seqlen/balanced_max:22652 - global_seqlen/mean:22651.25 - actor/entropy:0.46682530641555786 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4102363586425781 - training/rollout_probs_diff_mean:0.005161741282790899 - training/rollout_probs_diff_std:0.011252960190176964 - training/rollout_actor_probs_pearson_corr:0.9991944432258606 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.6484375 - self_distillation/correct_sample_fraction:0.55859375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.6484375 - rollout_corr/rollout_is_mean:1.0001730918884277 - rollout_corr/rollout_is_ratio_fraction_high:9.88513511401834e-06 - rollout_corr/rollout_is_ratio_fraction_low:1.977027022803668e-05 - rollout_corr/rollout_is_max:2.2787070274353027 - rollout_corr/rollout_is_min:0.2646176815032959 - rollout_corr/rollout_is_std:0.03861745819449425 - rollout_corr/rollout_is_eff_sample_size:0.9985113878945102 - rollout_corr/rollout_is_seq_mean:1.0001778602600098 - rollout_corr/rollout_is_seq_std:0.0021594883874058723 - rollout_corr/rollout_is_seq_max:1.0066732168197632 - rollout_corr/rollout_is_seq_min:0.9935801029205322 - rollout_corr/rollout_is_seq_max_deviation:0.006673216819763184 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6297845151275396) - rollout_corr/training_log_ppl:np.float64(0.4774156858911738) - rollout_corr/kl:np.float64(0.0007412694044468537) - rollout_corr/k3_kl:np.float64(0.0009004596647628205) - rollout_corr/rollout_ppl:np.float64(1.628512756433338) - rollout_corr/rollout_log_ppl:np.float64(0.47667441627709195) - rollout_corr/log_ppl_diff:np.float64(0.0007412696140818298) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018867639009840786) - rollout_corr/log_ppl_diff_max:np.float64(0.00960838794708252) - rollout_corr/log_ppl_diff_min:np.float64(-0.005440711975097656) - rollout_corr/ppl_ratio:np.float64(1.0007440785411745) - rollout_corr/chi2_token:np.float64(0.0020869714207947254) - rollout_corr/chi2_seq:np.float64(0.9351563951931894) - actor/pg_loss:np.float64(0.003354526706971228) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.000359561878212844) - actor/ppo_kl:np.float64(0.00013469158743095733) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.6484375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.6484375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.55859375) - self_distillation/token_reweight_w_mean:np.float64(9349.600620643701) - self_distillation/token_reweight_w_std:np.float64(148328.15592557297) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9980465415865183) - self_distillation/token_reweight_w_clip_frac:np.float64(0.08430611382937059) - self_distillation/empty_target_batch:np.float64(0.3515625) - actor/grad_norm:np.float64(0.28926024958491325) - perf/mfu/actor:np.float64(0.054041057160877155) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(195.16180038452148) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.6455357142857143) - val-aux/sciknoweval/reward/std@16:np.float64(0.19709042049186706) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7234142857142857) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.15652759126874422) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.5656380952380953) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.17688873171951364) - val-aux/sciknoweval/r
|
||||
step:41 - global_seqlen/min:18283 - global_seqlen/max:24700 - global_seqlen/minmax_diff:6417 - global_seqlen/balanced_min:21101 - global_seqlen/balanced_max:21109 - global_seqlen/mean:21105.5 - actor/entropy:0.42351460456848145 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9826107025146484 - training/rollout_probs_diff_mean:0.005102348513901234 - training/rollout_probs_diff_std:0.01174638606607914 - training/rollout_actor_probs_pearson_corr:0.9991093873977661 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8046875 - self_distillation/correct_sample_fraction:0.62890625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8046875 - rollout_corr/rollout_is_mean:0.999843180179596 - rollout_corr/rollout_is_ratio_fraction_high:9.873228009382728e-06 - rollout_corr/rollout_is_ratio_fraction_low:9.873228009382728e-06 - rollout_corr/rollout_is_max:2.636786937713623 - rollout_corr/rollout_is_min:0.013885053806006908 - rollout_corr/rollout_is_std:0.03809376060962677 - rollout_corr/rollout_is_eff_sample_size:0.9985506114667393 - rollout_corr/rollout_is_seq_mean:0.9998304843902588 - rollout_corr/rollout_is_seq_std:0.001992590492591262 - rollout_corr/rollout_is_seq_max:1.0068778991699219 - rollout_corr/rollout_is_seq_min:0.9926929473876953 - rollout_corr/rollout_is_seq_max_deviation:0.0073070526123046875 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5631140493787825) - rollout_corr/training_log_ppl:np.float64(0.436920084990561) - rollout_corr/kl:np.float64(0.0011501897124975846) - rollout_corr/k3_kl:np.float64(0.0009579621572584074) - rollout_corr/rollout_ppl:np.float64(1.5612433003261685) - rollout_corr/rollout_log_ppl:np.float64(0.4357698948588222) - rollout_corr/log_ppl_diff:np.float64(0.001150190131738782) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019631539471447468) - rollout_corr/log_ppl_diff_max:np.float64(0.010622143745422363) - rollout_corr/log_ppl_diff_min:np.float64(-0.00589299201965332) - rollout_corr/ppl_ratio:np.float64(1.001153603894636) - rollout_corr/chi2_token:np.float64(0.0014891347382217646) - rollout_corr/chi2_seq:np.float64(0.3713759642560035) - actor/pg_loss:np.float64(0.007653480512090027) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0008902258696252829) - actor/ppo_kl:np.float64(0.0001822680885723571) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8046875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8046875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.62890625) - self_distillation/token_reweight_w_mean:np.float64(1181.0103185500484) - self_distillation/token_reweight_w_std:np.float64(19672.96590518701) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9963315855711699) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1375392735353671) - self_distillation/empty_target_batch:np.float64(0.1953125) - actor/grad_norm:np.float64(0.34661057591438293) - perf/mfu/actor:np.float64(0.05125395817020568) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.22666931152344) - actor/lr:np.float64(1e-06) - training/global_step:41 - training/epoch:0 - critic/score/mean:0.62890625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.62890625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.004591051023453474 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.004591051023453474 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:395.640625 - response
|
||||
step:42 - global_seqlen/min:15988 - global_seqlen/max:22821 - global_seqlen/minmax_diff:6833 - global_seqlen/balanced_min:19797 - global_seqlen/balanced_max:19801 - global_seqlen/mean:19798.875 - actor/entropy:0.39622214436531067 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6605160236358643 - training/rollout_probs_diff_mean:0.005419079214334488 - training/rollout_probs_diff_std:0.012383614666759968 - training/rollout_actor_probs_pearson_corr:0.998907744884491 - self_distillation/success_group_fraction:0.96875 - self_distillation/success_sample_fraction:0.9609375 - self_distillation/correct_sample_fraction:0.6171875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.9609375 - rollout_corr/rollout_is_mean:0.9999876022338867 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.1576137442025356e-05 - rollout_corr/rollout_is_max:1.7245110273361206 - rollout_corr/rollout_is_min:0.2930092215538025 - rollout_corr/rollout_is_std:0.03913719579577446 - rollout_corr/rollout_is_eff_sample_size:0.9984703847272242 - rollout_corr/rollout_is_seq_mean:0.9999865293502808 - rollout_corr/rollout_is_seq_std:0.002267190022394061 - rollout_corr/rollout_is_seq_max:1.0080516338348389 - rollout_corr/rollout_is_seq_min:0.9936971664428711 - rollout_corr/rollout_is_seq_max_deviation:0.008051633834838867 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5080496412701905) - rollout_corr/training_log_ppl:np.float64(0.4042311883531511) - rollout_corr/kl:np.float64(0.0007036857559299392) - rollout_corr/k3_kl:np.float64(0.0009411578781737262) - rollout_corr/rollout_ppl:np.float64(1.5069640525616705) - rollout_corr/rollout_log_ppl:np.float64(0.40352750167949125) - rollout_corr/log_ppl_diff:np.float64(0.0007036866736598313) - rollout_corr/log_ppl_abs_diff:np.float64(0.001983315742108971) - rollout_corr/log_ppl_diff_max:np.float64(0.0076701343059539795) - rollout_corr/log_ppl_diff_min:np.float64(-0.006415665149688721) - rollout_corr/ppl_ratio:np.float64(1.0007067790720612) - rollout_corr/chi2_token:np.float64(0.002371277194470167) - rollout_corr/chi2_seq:np.float64(0.9433175600133836) - actor/pg_loss:np.float64(0.010519832605496049) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0009025154786286294) - actor/ppo_kl:np.float64(-0.00013643282175568316) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.9609375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.9609375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6171875) - self_distillation/token_reweight_w_mean:np.float64(12954.285722455475) - self_distillation/token_reweight_w_std:np.float64(227959.35449937586) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0025726119056344) - self_distillation/token_reweight_w_clip_frac:np.float64(0.14800434719654731) - self_distillation/empty_target_batch:np.float64(0.0390625) - actor/grad_norm:np.float64(0.45579010248184204) - perf/mfu/actor:np.float64(0.04795526011784892) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.25069046020508) - actor/lr:np.float64(1e-06) - training/global_step:42 - training/epoch:0 - critic/score/mean:0.6171875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6171875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0013215383514761925 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0013215383514761925 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:362.08984375 - response_length
|
||||
step:43 - global_seqlen/min:18065 - global_seqlen/max:23096 - global_seqlen/minmax_diff:5031 - global_seqlen/balanced_min:20756 - global_seqlen/balanced_max:20762 - global_seqlen/mean:20760.875 - actor/entropy:0.42775002121925354 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3424738645553589 - training/rollout_probs_diff_mean:0.005239918828010559 - training/rollout_probs_diff_std:0.01176416501402855 - training/rollout_actor_probs_pearson_corr:0.999097466468811 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.79296875 - self_distillation/correct_sample_fraction:0.61328125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.79296875 - rollout_corr/rollout_is_mean:0.9999403357505798 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.1294968317088205e-05 - rollout_corr/rollout_is_max:1.6902360916137695 - rollout_corr/rollout_is_min:0.39213547110557556 - rollout_corr/rollout_is_std:0.038980014622211456 - rollout_corr/rollout_is_eff_sample_size:0.9984827447477223 - rollout_corr/rollout_is_seq_mean:0.9998909831047058 - rollout_corr/rollout_is_seq_std:0.002513354877009988 - rollout_corr/rollout_is_seq_max:1.0098495483398438 - rollout_corr/rollout_is_seq_min:0.9915899038314819 - rollout_corr/rollout_is_seq_max_deviation:0.00984954833984375 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5651745153591037) - rollout_corr/training_log_ppl:np.float64(0.4374735167075414) - rollout_corr/kl:np.float64(0.0007840867953743391) - rollout_corr/k3_kl:np.float64(0.0008964727625198066) - rollout_corr/rollout_ppl:np.float64(1.5639647683128715) - rollout_corr/rollout_log_ppl:np.float64(0.4366894279664848) - rollout_corr/log_ppl_diff:np.float64(0.0007840887410566211) - rollout_corr/log_ppl_abs_diff:np.float64(0.0020498010562732816) - rollout_corr/log_ppl_diff_max:np.float64(0.009105801582336426) - rollout_corr/log_ppl_diff_min:np.float64(-0.00906282663345337) - rollout_corr/ppl_ratio:np.float64(1.0007877822499722) - rollout_corr/chi2_token:np.float64(0.00206923414953053) - rollout_corr/chi2_seq:np.float64(1.3589389722328633) - actor/pg_loss:np.float64(0.004318759543821216) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004535638540801301) - actor/ppo_kl:np.float64(-0.00012510620860073374) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.79296875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.79296875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.61328125) - self_distillation/token_reweight_w_mean:np.float64(39451.46745206439) - self_distillation/token_reweight_w_std:np.float64(461544.07824653096) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9995274562388659) - self_distillation/token_reweight_w_clip_frac:np.float64(0.09244409290840849) - self_distillation/empty_target_batch:np.float64(0.20703125) - actor/grad_norm:np.float64(0.32581663131713867) - perf/mfu/actor:np.float64(0.05044233796581919) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.0129623413086) - actor/lr:np.float64(1e-06) - training/global_step:43 - training/epoch:0 - critic/score/mean:0.61328125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.61328125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0011210255324840546 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0011210255324840546 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:345.83984375 - response_len
|
||||
step:44 - global_seqlen/min:19213 - global_seqlen/max:22622 - global_seqlen/minmax_diff:3409 - global_seqlen/balanced_min:20879 - global_seqlen/balanced_max:20890 - global_seqlen/mean:20886.75 - actor/entropy:0.43929049372673035 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.22547024488449097 - training/rollout_probs_diff_mean:0.0053957682102918625 - training/rollout_probs_diff_std:0.011915630660951138 - training/rollout_actor_probs_pearson_corr:0.9990841150283813 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80859375 - self_distillation/correct_sample_fraction:0.66796875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.80859375 - rollout_corr/rollout_is_mean:1.000036358833313 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.9659661054611206 - rollout_corr/rollout_is_min:0.5844541192054749 - rollout_corr/rollout_is_std:0.039155468344688416 - rollout_corr/rollout_is_eff_sample_size:0.9984691962798411 - rollout_corr/rollout_is_seq_mean:1.0000959634780884 - rollout_corr/rollout_is_seq_std:0.0024589113891124725 - rollout_corr/rollout_is_seq_max:1.0087720155715942 - rollout_corr/rollout_is_seq_min:0.9921355843544006 - rollout_corr/rollout_is_seq_max_deviation:0.008772015571594238 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6047232481651008) - rollout_corr/training_log_ppl:np.float64(0.46260377258295193) - rollout_corr/kl:np.float64(0.0009372506496678756) - rollout_corr/k3_kl:np.float64(0.000976888309367041) - rollout_corr/rollout_ppl:np.float64(1.6032404340803623) - rollout_corr/rollout_log_ppl:np.float64(0.4616665196372196) - rollout_corr/log_ppl_diff:np.float64(0.0009372529457323253) - rollout_corr/log_ppl_abs_diff:np.float64(0.002075349271763116) - rollout_corr/log_ppl_diff_max:np.float64(0.013183057308197021) - rollout_corr/log_ppl_diff_min:np.float64(-0.006263077259063721) - rollout_corr/ppl_ratio:np.float64(1.000940943369642) - rollout_corr/chi2_token:np.float64(0.0020958164241164923) - rollout_corr/chi2_seq:np.float64(0.34359696647152305) - actor/pg_loss:np.float64(0.005748242023400962) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005829084911965765) - actor/ppo_kl:np.float64(0.0002053698056716513) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.80859375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.80859375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.66796875) - self_distillation/token_reweight_w_mean:np.float64(14803.191471951548) - self_distillation/token_reweight_w_std:np.float64(246306.37441055127) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9985546870157123) - self_distillation/token_reweight_w_clip_frac:np.float64(0.10855769729823805) - self_distillation/empty_target_batch:np.float64(0.19140625) - actor/grad_norm:np.float64(0.36420872062444687) - perf/mfu/actor:np.float64(0.05008791245123715) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(195.9407615661621) - actor/lr:np.float64(1e-06) - training/global_step:44 - training/epoch:0 - critic/score/mean:0.66796875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.66796875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.00279244058765471 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00279244058765471 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:341.1484375 - response_length/max:1060.0 - res
|
||||
step:45 - global_seqlen/min:18835 - global_seqlen/max:26317 - global_seqlen/minmax_diff:7482 - global_seqlen/balanced_min:21440 - global_seqlen/balanced_max:21451 - global_seqlen/mean:21447.875 - actor/entropy:0.4101407527923584 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.28218671679496765 - training/rollout_probs_diff_mean:0.004841715097427368 - training/rollout_probs_diff_std:0.01116796862334013 - training/rollout_actor_probs_pearson_corr:0.9992083311080933 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.73046875 - self_distillation/correct_sample_fraction:0.5390625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.73046875 - rollout_corr/rollout_is_mean:0.9999707937240601 - rollout_corr/rollout_is_ratio_fraction_high:1.0986233974108472e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.0986233974108472e-05 - rollout_corr/rollout_is_max:2.0019893646240234 - rollout_corr/rollout_is_min:0.25255540013313293 - rollout_corr/rollout_is_std:0.036969371140003204 - rollout_corr/rollout_is_eff_sample_size:0.998635012024894 - rollout_corr/rollout_is_seq_mean:1.0000436305999756 - rollout_corr/rollout_is_seq_std:0.002479569287970662 - rollout_corr/rollout_is_seq_max:1.0116991996765137 - rollout_corr/rollout_is_seq_min:0.9924298524856567 - rollout_corr/rollout_is_seq_max_deviation:0.011699199676513672 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5578296538442373) - rollout_corr/training_log_ppl:np.float64(0.432097623241134) - rollout_corr/kl:np.float64(0.0007768278248150295) - rollout_corr/k3_kl:np.float64(0.0008250089271086836) - rollout_corr/rollout_ppl:np.float64(1.5565427541732788) - rollout_corr/rollout_log_ppl:np.float64(0.4313207929953933) - rollout_corr/log_ppl_diff:np.float64(0.0007768302457407117) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019471754785627127) - rollout_corr/log_ppl_diff_max:np.float64(0.011698126792907715) - rollout_corr/log_ppl_diff_min:np.float64(-0.010318875312805176) - rollout_corr/ppl_ratio:np.float64(1.0007803542539477) - rollout_corr/chi2_token:np.float64(0.0017588012851774693) - rollout_corr/chi2_seq:np.float64(0.6141417354810983) - actor/pg_loss:np.float64(0.0026712483959272504) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00039884855823402177) - actor/ppo_kl:np.float64(6.892802709934642e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.73046875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.73046875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.5390625) - self_distillation/token_reweight_w_mean:np.float64(38907.24203802785) - self_distillation/token_reweight_w_std:np.float64(568287.8935087718) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0006591160781682) - self_distillation/token_reweight_w_clip_frac:np.float64(0.08028327141073532) - self_distillation/empty_target_batch:np.float64(0.26953125) - actor/grad_norm:np.float64(0.28873957321047783) - perf/mfu/actor:np.float64(0.05157141247510403) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.14898681640625) - actor/lr:np.float64(1e-06) - training/global_step:45 - training/epoch:0 - critic/score/mean:0.5390625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5390625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.005405227188020945 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.005405227188020945 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:355.55859375
|
||||
step:46 - global_seqlen/min:17315 - global_seqlen/max:21947 - global_seqlen/minmax_diff:4632 - global_seqlen/balanced_min:19767 - global_seqlen/balanced_max:19776 - global_seqlen/mean:19773.25 - actor/entropy:0.4300471842288971 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5463374257087708 - training/rollout_probs_diff_mean:0.0050852419808506966 - training/rollout_probs_diff_std:0.011606025509536266 - training/rollout_actor_probs_pearson_corr:0.9991208910942078 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8203125 - self_distillation/correct_sample_fraction:0.5546875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8203125 - rollout_corr/rollout_is_mean:0.9999197125434875 - rollout_corr/rollout_is_ratio_fraction_high:1.1397570233384613e-05 - rollout_corr/rollout_is_ratio_fraction_low:2.2795140466769226e-05 - rollout_corr/rollout_is_max:2.4014859199523926 - rollout_corr/rollout_is_min:0.26859816908836365 - rollout_corr/rollout_is_std:0.03753577917814255 - rollout_corr/rollout_is_eff_sample_size:0.9985928099624436 - rollout_corr/rollout_is_seq_mean:1.0000156164169312 - rollout_corr/rollout_is_seq_std:0.002315930090844631 - rollout_corr/rollout_is_seq_max:1.0097026824951172 - rollout_corr/rollout_is_seq_min:0.9926007986068726 - rollout_corr/rollout_is_seq_max_deviation:0.009702682495117188 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5882981927134097) - rollout_corr/training_log_ppl:np.float64(0.44962009048322216) - rollout_corr/kl:np.float64(0.0011486389046666545) - rollout_corr/k3_kl:np.float64(0.0009466409361493788) - rollout_corr/rollout_ppl:np.float64(1.5864842752926052) - rollout_corr/rollout_log_ppl:np.float64(0.4484714485006407) - rollout_corr/log_ppl_diff:np.float64(0.0011486419825814664) - rollout_corr/log_ppl_abs_diff:np.float64(0.0022174433688633144) - rollout_corr/log_ppl_diff_max:np.float64(0.015058279037475586) - rollout_corr/log_ppl_diff_min:np.float64(-0.00858527421951294) - rollout_corr/ppl_ratio:np.float64(1.0011532127391547) - rollout_corr/chi2_token:np.float64(0.0015598325990140438) - rollout_corr/chi2_seq:np.float64(0.6058120087254792) - actor/pg_loss:np.float64(0.005358678288757801) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006527249729515461) - actor/ppo_kl:np.float64(0.0004008172310321356) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8203125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8203125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.5546875) - self_distillation/token_reweight_w_mean:np.float64(13853.629367629066) - self_distillation/token_reweight_w_std:np.float64(223926.18253548152) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000649391906336) - self_distillation/token_reweight_w_clip_frac:np.float64(0.12465728094684891) - self_distillation/empty_target_batch:np.float64(0.1796875) - actor/grad_norm:np.float64(0.37411294877529144) - perf/mfu/actor:np.float64(0.04759140321566382) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(195.96075439453125) - actor/lr:np.float64(1e-06) - training/global_step:46 - training/epoch:0 - critic/score/mean:0.5546875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5546875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.004386639688163996 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.004386639688163996 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:342.7265625 - res
|
||||
step:47 - global_seqlen/min:16338 - global_seqlen/max:23481 - global_seqlen/minmax_diff:7143 - global_seqlen/balanced_min:20960 - global_seqlen/balanced_max:20969 - global_seqlen/mean:20966.0 - actor/entropy:0.4562165141105652 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24135512113571167 - training/rollout_probs_diff_mean:0.004941638559103012 - training/rollout_probs_diff_std:0.01093968003988266 - training/rollout_actor_probs_pearson_corr:0.9992673993110657 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.8671875 - self_distillation/correct_sample_fraction:0.64453125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8671875 - rollout_corr/rollout_is_mean:1.0000667572021484 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.9514927864074707 - rollout_corr/rollout_is_min:0.5686209797859192 - rollout_corr/rollout_is_std:0.03698226809501648 - rollout_corr/rollout_is_eff_sample_size:0.998634417604882 - rollout_corr/rollout_is_seq_mean:1.0001049041748047 - rollout_corr/rollout_is_seq_std:0.0021275379694998264 - rollout_corr/rollout_is_seq_max:1.0070573091506958 - rollout_corr/rollout_is_seq_min:0.9932472705841064 - rollout_corr/rollout_is_seq_max_deviation:0.007057309150695801 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6164640178903937) - rollout_corr/training_log_ppl:np.float64(0.46599303945549764) - rollout_corr/kl:np.float64(0.0011402188001117253) - rollout_corr/k3_kl:np.float64(0.0008741923521711215) - rollout_corr/rollout_ppl:np.float64(1.6145951971411705) - rollout_corr/rollout_log_ppl:np.float64(0.46485281962668523) - rollout_corr/log_ppl_diff:np.float64(0.0011402198288124055) - rollout_corr/log_ppl_abs_diff:np.float64(0.001965065923286602) - rollout_corr/log_ppl_diff_max:np.float64(0.008194833993911743) - rollout_corr/log_ppl_diff_min:np.float64(-0.005479037761688232) - rollout_corr/ppl_ratio:np.float64(1.0011436399072409) - rollout_corr/chi2_token:np.float64(0.0012568566016852856) - rollout_corr/chi2_seq:np.float64(0.40265263244509697) - actor/pg_loss:np.float64(0.006021094392053783) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005619370990643802) - actor/ppo_kl:np.float64(0.0005166811963022155) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8671875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8671875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.64453125) - self_distillation/token_reweight_w_mean:np.float64(12667.028565393994) - self_distillation/token_reweight_w_std:np.float64(223391.28332038334) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0007588975131512) - self_distillation/token_reweight_w_clip_frac:np.float64(0.12597810916486196) - self_distillation/empty_target_batch:np.float64(0.1328125) - actor/grad_norm:np.float64(0.35431264340877533) - perf/mfu/actor:np.float64(0.050087304585172576) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(195.9956817626953) - actor/lr:np.float64(1e-06) - training/global_step:47 - training/epoch:0 - critic/score/mean:0.64453125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.64453125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.005755655933171511 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.005755655933171511 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:361.90625 - response_length/max:784.0 - response_leng
|
||||
step:48 - global_seqlen/min:16138 - global_seqlen/max:20539 - global_seqlen/minmax_diff:4401 - global_seqlen/balanced_min:19360 - global_seqlen/balanced_max:19367 - global_seqlen/mean:19365.625 - actor/entropy:0.45748016238212585 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.423759400844574 - training/rollout_probs_diff_mean:0.00517674908041954 - training/rollout_probs_diff_std:0.011389809660613537 - training/rollout_actor_probs_pearson_corr:0.9992020130157471 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80859375 - self_distillation/correct_sample_fraction:0.671875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.80859375 - rollout_corr/rollout_is_mean:1.0000624656677246 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:3.4522043279139325e-05 - rollout_corr/rollout_is_max:1.9825339317321777 - rollout_corr/rollout_is_min:0.29720374941825867 - rollout_corr/rollout_is_std:0.038425594568252563 - rollout_corr/rollout_is_eff_sample_size:0.9985258883712781 - rollout_corr/rollout_is_seq_mean:1.0000547170639038 - rollout_corr/rollout_is_seq_std:0.0022232853807508945 - rollout_corr/rollout_is_seq_max:1.0069233179092407 - rollout_corr/rollout_is_seq_min:0.9934024810791016 - rollout_corr/rollout_is_seq_max_deviation:0.006923317909240723 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6189399659633636) - rollout_corr/training_log_ppl:np.float64(0.47103061771485955) - rollout_corr/kl:np.float64(0.0008533685957203829) - rollout_corr/k3_kl:np.float64(0.0008472814761262271) - rollout_corr/rollout_ppl:np.float64(1.6175275961868465) - rollout_corr/rollout_log_ppl:np.float64(0.47017724718898535) - rollout_corr/log_ppl_diff:np.float64(0.0008533705258741975) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018438339466229081) - rollout_corr/log_ppl_diff_max:np.float64(0.009492874145507812) - rollout_corr/log_ppl_diff_min:np.float64(-0.004758328199386597) - rollout_corr/ppl_ratio:np.float64(1.000856165541336) - rollout_corr/chi2_token:np.float64(0.0016905255615711212) - rollout_corr/chi2_seq:np.float64(0.5319139340426773) - actor/pg_loss:np.float64(0.004102788749150932) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002375331782786816) - actor/ppo_kl:np.float64(0.00012016280814464153) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.80859375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.80859375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.671875) - self_distillation/token_reweight_w_mean:np.float64(1381.5336034777574) - self_distillation/token_reweight_w_std:np.float64(20333.09212728584) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9975523005705327) - self_distillation/token_reweight_w_clip_frac:np.float64(0.11302358307875693) - self_distillation/empty_target_batch:np.float64(0.19140625) - actor/grad_norm:np.float64(0.2960100993514061) - perf/mfu/actor:np.float64(0.04606089601890587) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.1104965209961) - actor/lr:np.float64(1e-06) - training/global_step:48 - training/epoch:0 - critic/score/mean:0.671875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.671875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0013794433325529099 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0013794433325529099 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:339.45703125 - response_length/ma
|
||||
step:49 - global_seqlen/min:19946 - global_seqlen/max:22692 - global_seqlen/minmax_diff:2746 - global_seqlen/balanced_min:21730 - global_seqlen/balanced_max:21733 - global_seqlen/mean:21731.75 - actor/entropy:0.4979252517223358 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.43991324305534363 - training/rollout_probs_diff_mean:0.004984842147678137 - training/rollout_probs_diff_std:0.01066806074231863 - training/rollout_actor_probs_pearson_corr:0.9993179440498352 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.86328125 - self_distillation/correct_sample_fraction:0.640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.86328125 - rollout_corr/rollout_is_mean:0.9998939037322998 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:9.758192391018383e-06 - rollout_corr/rollout_is_max:1.9085274934768677 - rollout_corr/rollout_is_min:0.12017349898815155 - rollout_corr/rollout_is_std:0.03726325184106827 - rollout_corr/rollout_is_eff_sample_size:0.9986130189557251 - rollout_corr/rollout_is_seq_mean:0.9998945593833923 - rollout_corr/rollout_is_seq_std:0.001821837155148387 - rollout_corr/rollout_is_seq_max:1.005088448524475 - rollout_corr/rollout_is_seq_min:0.9940054416656494 - rollout_corr/rollout_is_seq_max_deviation:0.005994558334350586 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6754766842350364) - rollout_corr/training_log_ppl:np.float64(0.5048167807399295) - rollout_corr/kl:np.float64(0.0009508746554258707) - rollout_corr/k3_kl:np.float64(0.0007735003869697721) - rollout_corr/rollout_ppl:np.float64(1.6738910162821412) - rollout_corr/rollout_log_ppl:np.float64(0.503865905397106) - rollout_corr/log_ppl_diff:np.float64(0.000950875342823565) - rollout_corr/log_ppl_abs_diff:np.float64(0.001617251429706812) - rollout_corr/log_ppl_diff_max:np.float64(0.008522868156433105) - rollout_corr/log_ppl_diff_min:np.float64(-0.00493285059928894) - rollout_corr/ppl_ratio:np.float64(1.0009531062096357) - rollout_corr/chi2_token:np.float64(0.0011934449430555105) - rollout_corr/chi2_seq:np.float64(0.1550909043289721) - actor/pg_loss:np.float64(0.005771090276539326) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002795115901790268) - actor/ppo_kl:np.float64(0.00010383528175317291) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.86328125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.86328125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.640625) - self_distillation/token_reweight_w_mean:np.float64(4803.391529331682) - self_distillation/token_reweight_w_std:np.float64(100370.09618178767) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.001389476004988) - self_distillation/token_reweight_w_clip_frac:np.float64(0.10803735570516437) - self_distillation/empty_target_batch:np.float64(0.13671875) - actor/grad_norm:np.float64(0.3376929610967636) - perf/mfu/actor:np.float64(0.05224311765341433) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(195.75490951538086) - actor/lr:np.float64(1e-06) - training/global_step:49 - training/epoch:0 - critic/score/mean:0.640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0023309881798923016 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.0023309881798923016 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:400.3046875 - response_length/max:826.0 - res
|
||||
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 50}
|
||||
validation generation end
|
||||
[prompt] system
|
||||
|
||||
Given a question and four options, please select the right answer. Respond in the following format:
|
||||
<reasoning>
|
||||
...
|
||||
</reasoning>
|
||||
<answer>
|
||||
...
|
||||
</answer>
|
||||
|
||||
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
|
||||
user
|
||||
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||||
|
||||
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||||
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||||
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||||
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||||
Please reason step by step.
|
||||
assistant
|
||||
<think>
|
||||
|
||||
</think>
|
||||
|
||||
|
||||
[response] <reasoning>
|
||||
To determine the correct reactant for the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1", we need to examine the structure of the product and compare it with the structures of the given options.
|
||||
|
||||
The product has a nitrogen-containing ring with a substituent group: "Cc2ccc(Cl)cc2C(F)(F)F". This suggests that the molecule contains a chlorinated aromatic ring attached to a trifluoromethyl group, which is connected to a nitrogen-containing ring.
|
||||
|
||||
Now, let's evaluate the options:
|
||||
|
||||
- **Option A**: Contains a structure with a nitrogen ring and a chlorinated aromatic ring connected to a trifluoromethyl group. The structure is very similar to the product, with the correct substituents and connectivity.
|
||||
- **Option B**: Contains a chlorinated nitrogen ring but lacks the trifluoromethyl group and the correct aromatic ring structure.
|
||||
- **Option C**: Contains a hydroxyl group and a different substituent, not matching the product.
|
||||
- **Option D**: Contains a different ring structure and lacks the trifluoromethyl group and the correct aromatic ring.
|
||||
|
||||
Only **Option A** matches the structure and substituents of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1".
|
||||
|
||||
</reasoning>
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
[ground_truth] A
|
||||
[score] 1.0
|
||||
[acc] 1.0
|
||||
[pred] A
|
||||
[incorrect_format] 1
|
||||
[feedback]
|
||||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_50
|
||||
step:50 - global_seqlen/min:19045 - global_seqlen/max:30156 - global_seqlen/minmax_diff:11111 - global_seqlen/balanced_min:21698 - global_seqlen/balanced_max:21699 - global_seqlen/mean:21698.5 - actor/entropy:0.39573362469673157 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24491804838180542 - training/rollout_probs_diff_mean:0.004589053802192211 - training/rollout_probs_diff_std:0.01072707585990429 - training/rollout_actor_probs_pearson_corr:0.9992213845252991 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - self_distillation/correct_sample_fraction:0.609375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83203125 - rollout_corr/rollout_is_mean:1.0000442266464233 - rollout_corr/rollout_is_ratio_fraction_high:9.806613888940774e-06 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.004254102706909 - rollout_corr/rollout_is_min:0.6065316796302795 - rollout_corr/rollout_is_std:0.03603039309382439 - rollout_corr/rollout_is_eff_sample_size:0.9987034939469002 - rollout_corr/rollout_is_seq_mean:1.000090479850769 - rollout_corr/rollout_is_seq_std:0.0021530245430767536 - rollout_corr/rollout_is_seq_max:1.0064139366149902 - rollout_corr/rollout_is_seq_min:0.991096556186676 - rollout_corr/rollout_is_seq_max_deviation:0.008903443813323975 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.524656287394464) - rollout_corr/training_log_ppl:np.float64(0.4108033778902609) - rollout_corr/kl:np.float64(0.0007370334822254243) - rollout_corr/k3_kl:np.float64(0.0008599899813646061) - rollout_corr/rollout_ppl:np.float64(1.5234609101898968) - rollout_corr/rollout_log_ppl:np.float64(0.41006634262157604) - rollout_corr/log_ppl_diff:np.float64(0.0007370352686848491) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018268449639435858) - rollout_corr/log_ppl_diff_max:np.float64(0.00987979769706726) - rollout_corr/log_ppl_diff_min:np.float64(-0.00624692440032959) - rollout_corr/ppl_ratio:np.float64(1.000740017509088) - rollout_corr/chi2_token:np.float64(0.0019624873530119658) - rollout_corr/chi2_seq:np.float64(0.6737326441798359) - actor/pg_loss:np.float64(0.00585430464707315) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0007581320141980541) - actor/ppo_kl:np.float64(0.00011956099752552518) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.83203125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.83203125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.609375) - self_distillation/token_reweight_w_mean:np.float64(4577.494679181604) - self_distillation/token_reweight_w_std:np.float64(75540.59034832189) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9987221618648618) - self_distillation/token_reweight_w_clip_frac:np.float64(0.11705390218412504) - self_distillation/empty_target_batch:np.float64(0.16796875) - actor/grad_norm:np.float64(0.3779703937470913) - perf/mfu/actor:np.float64(0.0522593230068327) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(195.6965789794922) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.66875) - val-aux/sciknoweval/reward/std@16:np.float64(0.17551919057486814) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7372904761904763) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.13660825245257055) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6003428571428571) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.15818906699634022) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.6687047
|
||||
step:51 - global_seqlen/min:19988 - global_seqlen/max:24236 - global_seqlen/minmax_diff:4248 - global_seqlen/balanced_min:21993 - global_seqlen/balanced_max:22000 - global_seqlen/mean:21997.625 - actor/entropy:0.47643905878067017 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6292527914047241 - training/rollout_probs_diff_mean:0.004974652547389269 - training/rollout_probs_diff_std:0.010949385352432728 - training/rollout_actor_probs_pearson_corr:0.9992679953575134 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - self_distillation/correct_sample_fraction:0.5546875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83203125 - rollout_corr/rollout_is_mean:1.000044584274292 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.0419596947031096e-05 - rollout_corr/rollout_is_max:1.6338132619857788 - rollout_corr/rollout_is_min:0.19046327471733093 - rollout_corr/rollout_is_std:0.03751512989401817 - rollout_corr/rollout_is_eff_sample_size:0.998594711950472 - rollout_corr/rollout_is_seq_mean:0.9999472498893738 - rollout_corr/rollout_is_seq_std:0.002286979230120778 - rollout_corr/rollout_is_seq_max:1.0055614709854126 - rollout_corr/rollout_is_seq_min:0.9931483864784241 - rollout_corr/rollout_is_seq_max_deviation:0.006851613521575928 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6695182812400162) - rollout_corr/training_log_ppl:np.float64(0.49964700039708987) - rollout_corr/kl:np.float64(0.0010526214147859037) - rollout_corr/k3_kl:np.float64(0.0008514240415706809) - rollout_corr/rollout_ppl:np.float64(1.6677322825416923) - rollout_corr/rollout_log_ppl:np.float64(0.49859437631675974) - rollout_corr/log_ppl_diff:np.float64(0.0010526240803301334) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019851572578772902) - rollout_corr/log_ppl_diff_max:np.float64(0.008082211017608643) - rollout_corr/log_ppl_diff_min:np.float64(-0.008058786392211914) - rollout_corr/ppl_ratio:np.float64(1.0010559679940343) - rollout_corr/chi2_token:np.float64(0.0013342886231839657) - rollout_corr/chi2_seq:np.float64(1.1409833203069866) - actor/pg_loss:np.float64(0.005735161481425166) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006951400860089052) - actor/ppo_kl:np.float64(0.00024209431599508946) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.83203125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.83203125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.5546875) - self_distillation/token_reweight_w_mean:np.float64(7385.579081017524) - self_distillation/token_reweight_w_std:np.float64(136267.17127367936) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0020049593877047) - self_distillation/token_reweight_w_clip_frac:np.float64(0.11803711034008302) - self_distillation/empty_target_batch:np.float64(0.16796875) - actor/grad_norm:np.float64(0.3889169320464134) - perf/mfu/actor:np.float64(0.053154140926217126) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(199.41073989868164) - actor/lr:np.float64(1e-06) - training/global_step:51 - training/epoch:0 - critic/score/mean:0.5546875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5546875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0006512248073704541 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0006512248073704541 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:374.89453125 - response_len
|
||||
step:52 - global_seqlen/min:16418 - global_seqlen/max:23760 - global_seqlen/minmax_diff:7342 - global_seqlen/balanced_min:20391 - global_seqlen/balanced_max:20400 - global_seqlen/mean:20398.0 - actor/entropy:0.4327685832977295 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4141288995742798 - training/rollout_probs_diff_mean:0.004852033220231533 - training/rollout_probs_diff_std:0.011078495532274246 - training/rollout_actor_probs_pearson_corr:0.9992164969444275 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.859375 - self_distillation/correct_sample_fraction:0.61328125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.859375 - rollout_corr/rollout_is_mean:0.999881386756897 - rollout_corr/rollout_is_ratio_fraction_high:1.0429704161651898e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.107421636581421 - rollout_corr/rollout_is_min:0.5139061212539673 - rollout_corr/rollout_is_std:0.03718959912657738 - rollout_corr/rollout_is_eff_sample_size:0.9986184874121644 - rollout_corr/rollout_is_seq_mean:0.9998266696929932 - rollout_corr/rollout_is_seq_std:0.0023117384407669306 - rollout_corr/rollout_is_seq_max:1.010014533996582 - rollout_corr/rollout_is_seq_min:0.9893986582756042 - rollout_corr/rollout_is_seq_max_deviation:0.010601341724395752 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5747977024875581) - rollout_corr/training_log_ppl:np.float64(0.44207731005735695) - rollout_corr/kl:np.float64(0.0007742779266513011) - rollout_corr/k3_kl:np.float64(0.0008074032141394127) - rollout_corr/rollout_ppl:np.float64(1.5735519244335592) - rollout_corr/rollout_log_ppl:np.float64(0.44130302700796165) - rollout_corr/log_ppl_diff:np.float64(0.0007742830493953079) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017560763808432966) - rollout_corr/log_ppl_diff_max:np.float64(0.011528313159942627) - rollout_corr/log_ppl_diff_min:np.float64(-0.008214741945266724) - rollout_corr/ppl_ratio:np.float64(1.0007772489916533) - rollout_corr/chi2_token:np.float64(0.0016987936105579138) - rollout_corr/chi2_seq:np.float64(0.6159643346909434) - actor/pg_loss:np.float64(0.004326079739257693) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003785648150369525) - actor/ppo_kl:np.float64(-0.00013583274305517534) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.859375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.859375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.61328125) - self_distillation/token_reweight_w_mean:np.float64(23087.692809279542) - self_distillation/token_reweight_w_std:np.float64(411971.67904203804) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0013474761508405) - self_distillation/token_reweight_w_clip_frac:np.float64(0.10722107632318512) - self_distillation/empty_target_batch:np.float64(0.140625) - actor/grad_norm:np.float64(0.31624583154916763) - perf/mfu/actor:np.float64(0.04994189212775807) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(199.2727928161621) - actor/lr:np.float64(1e-06) - training/global_step:52 - training/epoch:0 - critic/score/mean:0.61328125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.61328125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.00683667091652751 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00683667091652751 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:374.53125 - response_length/max:748.0 -
|
||||
step:53 - global_seqlen/min:19241 - global_seqlen/max:23446 - global_seqlen/minmax_diff:4205 - global_seqlen/balanced_min:21257 - global_seqlen/balanced_max:21274 - global_seqlen/mean:21261.625 - actor/entropy:0.4501931071281433 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2294929027557373 - training/rollout_probs_diff_mean:0.0048794192261993885 - training/rollout_probs_diff_std:0.010888302698731422 - training/rollout_actor_probs_pearson_corr:0.9992465376853943 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.73828125 - self_distillation/correct_sample_fraction:0.62109375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.73828125 - rollout_corr/rollout_is_mean:0.9998956918716431 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.1179934265092015e-05 - rollout_corr/rollout_is_max:1.6338140964508057 - rollout_corr/rollout_is_min:0.37676042318344116 - rollout_corr/rollout_is_std:0.036510247737169266 - rollout_corr/rollout_is_eff_sample_size:0.9986683006749928 - rollout_corr/rollout_is_seq_mean:0.9998424053192139 - rollout_corr/rollout_is_seq_std:0.0020514947827905416 - rollout_corr/rollout_is_seq_max:1.0057796239852905 - rollout_corr/rollout_is_seq_min:0.9921916127204895 - rollout_corr/rollout_is_seq_max_deviation:0.007808387279510498 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6163378930650651) - rollout_corr/training_log_ppl:np.float64(0.4686294782150071) - rollout_corr/kl:np.float64(0.000762243592463463) - rollout_corr/k3_kl:np.float64(0.000810159646334796) - rollout_corr/rollout_ppl:np.float64(1.6150716617703438) - rollout_corr/rollout_log_ppl:np.float64(0.46786723379045725) - rollout_corr/log_ppl_diff:np.float64(0.0007622444245498627) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019291590724606067) - rollout_corr/log_ppl_diff_max:np.float64(0.008153051137924194) - rollout_corr/log_ppl_diff_min:np.float64(-0.005449235439300537) - rollout_corr/ppl_ratio:np.float64(1.000765364151448) - rollout_corr/chi2_token:np.float64(0.001715675462037325) - rollout_corr/chi2_seq:np.float64(0.6910448384005576) - actor/pg_loss:np.float64(0.0023863104870542884) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002726298084780865) - actor/ppo_kl:np.float64(-0.00011610987491650349) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.73828125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.73828125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.62109375) - self_distillation/token_reweight_w_mean:np.float64(1507.4308137190528) - self_distillation/token_reweight_w_std:np.float64(33027.9989340055) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9995625410228968) - self_distillation/token_reweight_w_clip_frac:np.float64(0.06903543486259878) - self_distillation/empty_target_batch:np.float64(0.26171875) - actor/grad_norm:np.float64(0.21907289698719978) - perf/mfu/actor:np.float64(0.051939957351732395) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(199.24185180664062) - actor/lr:np.float64(1e-06) - training/global_step:53 - training/epoch:0 - critic/score/mean:0.62109375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.62109375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.003284213598817587 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.003284213598817587 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:368.86328125 - response_l
|
||||
step:54 - global_seqlen/min:16771 - global_seqlen/max:23434 - global_seqlen/minmax_diff:6663 - global_seqlen/balanced_min:20121 - global_seqlen/balanced_max:20134 - global_seqlen/mean:20129.375 - actor/entropy:0.39255428314208984 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2144627422094345 - training/rollout_probs_diff_mean:0.004672019276767969 - training/rollout_probs_diff_std:0.011115854606032372 - training/rollout_actor_probs_pearson_corr:0.9991840720176697 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8046875 - self_distillation/correct_sample_fraction:0.5546875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8046875 - rollout_corr/rollout_is_mean:0.9999427795410156 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.969622254371643 - rollout_corr/rollout_is_min:0.5503029227256775 - rollout_corr/rollout_is_std:0.03688544034957886 - rollout_corr/rollout_is_eff_sample_size:0.998641194034968 - rollout_corr/rollout_is_seq_mean:0.9999832510948181 - rollout_corr/rollout_is_seq_std:0.002377620432525873 - rollout_corr/rollout_is_seq_max:1.0109039545059204 - rollout_corr/rollout_is_seq_min:0.9921522736549377 - rollout_corr/rollout_is_seq_max_deviation:0.01090395450592041 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5304169207811356) - rollout_corr/training_log_ppl:np.float64(0.4121860411542002) - rollout_corr/kl:np.float64(0.0008754907883350427) - rollout_corr/k3_kl:np.float64(0.0008680712330715323) - rollout_corr/rollout_ppl:np.float64(1.5290372776798904) - rollout_corr/rollout_log_ppl:np.float64(0.411310548952315) - rollout_corr/log_ppl_diff:np.float64(0.0008754922018852085) - rollout_corr/log_ppl_abs_diff:np.float64(0.002070396934868768) - rollout_corr/log_ppl_diff_max:np.float64(0.016439110040664673) - rollout_corr/log_ppl_diff_min:np.float64(-0.00867973268032074) - rollout_corr/ppl_ratio:np.float64(1.0008794867899269) - rollout_corr/chi2_token:np.float64(0.0017218261491507292) - rollout_corr/chi2_seq:np.float64(0.5381895105820149) - actor/pg_loss:np.float64(0.006413624971173704) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005543413315081125) - actor/ppo_kl:np.float64(9.862682850325655e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8046875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8046875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.5546875) - self_distillation/token_reweight_w_mean:np.float64(21560.225262402324) - self_distillation/token_reweight_w_std:np.float64(340430.1742267248) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0004913297016174) - self_distillation/token_reweight_w_clip_frac:np.float64(0.10845586232608184) - self_distillation/empty_target_batch:np.float64(0.1953125) - actor/grad_norm:np.float64(0.3543579652905464) - perf/mfu/actor:np.float64(0.04868076222396018) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(199.25546646118164) - actor/lr:np.float64(1e-06) - training/global_step:54 - training/epoch:0 - critic/score/mean:0.5546875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5546875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.011497573927044868 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.011497573927044868 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:358.26171875 - response_length/max:1148.0 - response_length/m
|
||||
step:55 - global_seqlen/min:19138 - global_seqlen/max:21905 - global_seqlen/minmax_diff:2767 - global_seqlen/balanced_min:20379 - global_seqlen/balanced_max:20392 - global_seqlen/mean:20387.375 - actor/entropy:0.46607786417007446 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.44003328680992126 - training/rollout_probs_diff_mean:0.005262596998363733 - training/rollout_probs_diff_std:0.011458019725978374 - training/rollout_actor_probs_pearson_corr:0.9992002844810486 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8046875 - self_distillation/correct_sample_fraction:0.640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8046875 - rollout_corr/rollout_is_mean:1.0000497102737427 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:3.523318446241319e-05 - rollout_corr/rollout_is_max:1.6244064569473267 - rollout_corr/rollout_is_min:0.396142840385437 - rollout_corr/rollout_is_std:0.03780318796634674 - rollout_corr/rollout_is_eff_sample_size:0.99857307726417 - rollout_corr/rollout_is_seq_mean:0.9999815225601196 - rollout_corr/rollout_is_seq_std:0.002393631497398019 - rollout_corr/rollout_is_seq_max:1.0126123428344727 - rollout_corr/rollout_is_seq_min:0.9941840171813965 - rollout_corr/rollout_is_seq_max_deviation:0.012612342834472656 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.630201779305935) - rollout_corr/training_log_ppl:np.float64(0.47879417467629537) - rollout_corr/kl:np.float64(0.0010309481362873463) - rollout_corr/k3_kl:np.float64(0.0008061554526079817) - rollout_corr/rollout_ppl:np.float64(1.6284840498119593) - rollout_corr/rollout_log_ppl:np.float64(0.4777632247423753) - rollout_corr/log_ppl_diff:np.float64(0.0010309499339200556) - rollout_corr/log_ppl_abs_diff:np.float64(0.0020579289994202554) - rollout_corr/log_ppl_diff_max:np.float64(0.007941722869873047) - rollout_corr/log_ppl_diff_min:np.float64(-0.0057179927825927734) - rollout_corr/ppl_ratio:np.float64(1.0010342469904572) - rollout_corr/chi2_token:np.float64(0.00116553227417171) - rollout_corr/chi2_seq:np.float64(0.674807739444077) - actor/pg_loss:np.float64(0.004637886886484921) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00022305224865704076) - actor/ppo_kl:np.float64(0.0002447084712073888) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8046875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8046875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.640625) - self_distillation/token_reweight_w_mean:np.float64(5661.525033625308) - self_distillation/token_reweight_w_std:np.float64(108029.05970195204) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0004851918201894) - self_distillation/token_reweight_w_clip_frac:np.float64(0.08942344659590162) - self_distillation/empty_target_batch:np.float64(0.1953125) - actor/grad_norm:np.float64(0.34092599898576736) - perf/mfu/actor:np.float64(0.04916186108901947) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(199.24573516845703) - actor/lr:np.float64(1e-06) - training/global_step:55 - training/epoch:0 - critic/score/mean:0.640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0027775494381785393 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0027775494381785393 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:332.60546875 - response_length/max:929.0 -
|
||||
step:56 - global_seqlen/min:16165 - global_seqlen/max:23611 - global_seqlen/minmax_diff:7446 - global_seqlen/balanced_min:19603 - global_seqlen/balanced_max:19611 - global_seqlen/mean:19608.875 - actor/entropy:0.42714759707450867 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2066904902458191 - training/rollout_probs_diff_mean:0.004840976092964411 - training/rollout_probs_diff_std:0.011059464886784554 - training/rollout_actor_probs_pearson_corr:0.9992184638977051 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.87109375 - self_distillation/correct_sample_fraction:0.6484375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.87109375 - rollout_corr/rollout_is_mean:1.0001850128173828 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.7041244506835938 - rollout_corr/rollout_is_min:0.5909921526908875 - rollout_corr/rollout_is_std:0.03720882534980774 - rollout_corr/rollout_is_eff_sample_size:0.9986178930118242 - rollout_corr/rollout_is_seq_mean:1.0002026557922363 - rollout_corr/rollout_is_seq_std:0.0023563196882605553 - rollout_corr/rollout_is_seq_max:1.0108153820037842 - rollout_corr/rollout_is_seq_min:0.9901830554008484 - rollout_corr/rollout_is_seq_max_deviation:0.01081538200378418 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5864489898085594) - rollout_corr/training_log_ppl:np.float64(0.4489033415738959) - rollout_corr/kl:np.float64(0.0005826077157036202) - rollout_corr/k3_kl:np.float64(0.0008374885150033151) - rollout_corr/rollout_ppl:np.float64(1.5854855361394584) - rollout_corr/rollout_log_ppl:np.float64(0.4483207310840953) - rollout_corr/log_ppl_diff:np.float64(0.0005826104898005724) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019630271708592772) - rollout_corr/log_ppl_diff_max:np.float64(0.011552989482879639) - rollout_corr/log_ppl_diff_min:np.float64(-0.00930735468864441) - rollout_corr/ppl_ratio:np.float64(1.0005861730314791) - rollout_corr/chi2_token:np.float64(0.0022083548828959465) - rollout_corr/chi2_seq:np.float64(1.1513643299695104) - actor/pg_loss:np.float64(0.0053109838627278805) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004679971348195977) - actor/ppo_kl:np.float64(4.2931282393965375e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.87109375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.87109375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6484375) - self_distillation/token_reweight_w_mean:np.float64(133.22875495348126) - self_distillation/token_reweight_w_std:np.float64(2217.535897645983) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0020651405211538) - self_distillation/token_reweight_w_clip_frac:np.float64(0.08932297839783132) - self_distillation/empty_target_batch:np.float64(0.12890625) - actor/grad_norm:np.float64(0.3138376846909523) - perf/mfu/actor:np.float64(0.04696252700818672) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(199.17254257202148) - actor/lr:np.float64(1e-06) - training/global_step:56 - training/epoch:0 - critic/score/mean:0.6484375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6484375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.004433032590895891 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.004433032590895891 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:347.62109375 - response_length/max:909.0 - response_
|
||||
step:57 - global_seqlen/min:17187 - global_seqlen/max:22734 - global_seqlen/minmax_diff:5547 - global_seqlen/balanced_min:19426 - global_seqlen/balanced_max:19436 - global_seqlen/mean:19433.875 - actor/entropy:0.419066846370697 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.45647868514060974 - training/rollout_probs_diff_mean:0.005257493816316128 - training/rollout_probs_diff_std:0.012042389251291752 - training/rollout_actor_probs_pearson_corr:0.9990561008453369 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - self_distillation/correct_sample_fraction:0.66796875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83203125 - rollout_corr/rollout_is_mean:0.9998766779899597 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:4.874779187957756e-05 - rollout_corr/rollout_is_max:1.9517626762390137 - rollout_corr/rollout_is_min:0.412737101316452 - rollout_corr/rollout_is_std:0.039478376507759094 - rollout_corr/rollout_is_eff_sample_size:0.9984436453453855 - rollout_corr/rollout_is_seq_mean:0.9998682141304016 - rollout_corr/rollout_is_seq_std:0.002758133690804243 - rollout_corr/rollout_is_seq_max:1.010321855545044 - rollout_corr/rollout_is_seq_min:0.9932553768157959 - rollout_corr/rollout_is_seq_max_deviation:0.010321855545043945 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5485623660497367) - rollout_corr/training_log_ppl:np.float64(0.4282529861666262) - rollout_corr/kl:np.float64(0.0008480568455153659) - rollout_corr/k3_kl:np.float64(0.0008880625406391118) - rollout_corr/rollout_ppl:np.float64(1.5472203940153122) - rollout_corr/rollout_log_ppl:np.float64(0.42740493212477304) - rollout_corr/log_ppl_diff:np.float64(0.0008480540418531746) - rollout_corr/log_ppl_abs_diff:np.float64(0.002164474077289924) - rollout_corr/log_ppl_diff_max:np.float64(0.007681757211685181) - rollout_corr/log_ppl_diff_min:np.float64(-0.008211076259613037) - rollout_corr/ppl_ratio:np.float64(1.0008519866969436) - rollout_corr/chi2_token:np.float64(0.0018828820902854204) - rollout_corr/chi2_seq:np.float64(1.0425616323482245) - actor/pg_loss:np.float64(0.002906394307501614) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00029778121393064794) - actor/ppo_kl:np.float64(-0.00014163734993388744) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.83203125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.83203125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.66796875) - self_distillation/token_reweight_w_mean:np.float64(3090.494025749387) - self_distillation/token_reweight_w_std:np.float64(77747.68742029642) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9994963144417852) - self_distillation/token_reweight_w_clip_frac:np.float64(0.06993347342358902) - self_distillation/empty_target_batch:np.float64(0.16796875) - actor/grad_norm:np.float64(0.22349200583994389) - perf/mfu/actor:np.float64(0.04662792309159809) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(199.20368194580078) - actor/lr:np.float64(1e-06) - training/global_step:57 - training/epoch:0 - critic/score/mean:0.66796875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.66796875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0013542745728045702 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0013542745728045702 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:320.52734375 - response_
|
||||
step:58 - global_seqlen/min:17591 - global_seqlen/max:22208 - global_seqlen/minmax_diff:4617 - global_seqlen/balanced_min:19936 - global_seqlen/balanced_max:19980 - global_seqlen/mean:19944.0 - actor/entropy:0.39805784821510315 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5443886518478394 - training/rollout_probs_diff_mean:0.0047958018258214 - training/rollout_probs_diff_std:0.011325978673994541 - training/rollout_actor_probs_pearson_corr:0.9991068243980408 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.8671875 - self_distillation/correct_sample_fraction:0.703125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8671875 - rollout_corr/rollout_is_mean:0.9999314546585083 - rollout_corr/rollout_is_ratio_fraction_high:1.1276500117674004e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.1276500117674004e-05 - rollout_corr/rollout_is_max:2.245145320892334 - rollout_corr/rollout_is_min:0.3792571425437927 - rollout_corr/rollout_is_std:0.037710048258304596 - rollout_corr/rollout_is_eff_sample_size:0.9985797339908784 - rollout_corr/rollout_is_seq_mean:0.9998851418495178 - rollout_corr/rollout_is_seq_std:0.002318696118891239 - rollout_corr/rollout_is_seq_max:1.007555365562439 - rollout_corr/rollout_is_seq_min:0.9894842505455017 - rollout_corr/rollout_is_seq_max_deviation:0.010515749454498291 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5350726060569286) - rollout_corr/training_log_ppl:np.float64(0.4149510355782695) - rollout_corr/kl:np.float64(0.0008157491595808253) - rollout_corr/k3_kl:np.float64(0.0008106242991630097) - rollout_corr/rollout_ppl:np.float64(1.533764130435884) - rollout_corr/rollout_log_ppl:np.float64(0.41413528152043) - rollout_corr/log_ppl_diff:np.float64(0.0008157540578395128) - rollout_corr/log_ppl_abs_diff:np.float64(0.00186522479634732) - rollout_corr/log_ppl_diff_max:np.float64(0.012087315320968628) - rollout_corr/log_ppl_diff_min:np.float64(-0.005733311176300049) - rollout_corr/ppl_ratio:np.float64(1.0008189915679395) - rollout_corr/chi2_token:np.float64(0.0016144579276442528) - rollout_corr/chi2_seq:np.float64(0.5739311659708619) - actor/pg_loss:np.float64(0.0032648365013301373) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00039790487971913535) - actor/ppo_kl:np.float64(-6.927822899172043e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8671875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8671875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.703125) - self_distillation/token_reweight_w_mean:np.float64(20.40596836898476) - self_distillation/token_reweight_w_std:np.float64(398.28047909680754) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0010924299713224) - self_distillation/token_reweight_w_clip_frac:np.float64(0.060014044371200725) - self_distillation/empty_target_batch:np.float64(0.1328125) - actor/grad_norm:np.float64(0.22483210638165474) - perf/mfu/actor:np.float64(0.048109296920814684) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(199.0834503173828) - actor/lr:np.float64(1e-06) - training/global_step:58 - training/epoch:0 - critic/score/mean:0.703125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.703125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00522806728258729 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00522806728258729 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:346.40625 - response_length/max
|
||||
step:59 - global_seqlen/min:16516 - global_seqlen/max:23158 - global_seqlen/minmax_diff:6642 - global_seqlen/balanced_min:19285 - global_seqlen/balanced_max:19288 - global_seqlen/mean:19286.5 - actor/entropy:0.39392659068107605 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23105931282043457 - training/rollout_probs_diff_mean:0.004885942675173283 - training/rollout_probs_diff_std:0.01138182170689106 - training/rollout_actor_probs_pearson_corr:0.9991214275360107 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65234375 - self_distillation/correct_sample_fraction:0.5390625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.65234375 - rollout_corr/rollout_is_mean:0.9998358488082886 - rollout_corr/rollout_is_ratio_fraction_high:1.1690437531797215e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.1690437531797215e-05 - rollout_corr/rollout_is_max:3.892078161239624 - rollout_corr/rollout_is_min:0.4614630937576294 - rollout_corr/rollout_is_std:0.03779766708612442 - rollout_corr/rollout_is_eff_sample_size:0.9985728395255719 - rollout_corr/rollout_is_seq_mean:0.999951183795929 - rollout_corr/rollout_is_seq_std:0.002800993388518691 - rollout_corr/rollout_is_seq_max:1.0143204927444458 - rollout_corr/rollout_is_seq_min:0.9928278923034668 - rollout_corr/rollout_is_seq_max_deviation:0.0143204927444458 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5359033830463886) - rollout_corr/training_log_ppl:np.float64(0.4180586190195754) - rollout_corr/kl:np.float64(0.0011108192077600165) - rollout_corr/k3_kl:np.float64(0.0008730394356462057) - rollout_corr/rollout_ppl:np.float64(1.5341403023339808) - rollout_corr/rollout_log_ppl:np.float64(0.41694779935642146) - rollout_corr/log_ppl_diff:np.float64(0.0011108196631539613) - rollout_corr/log_ppl_abs_diff:np.float64(0.002291953394887969) - rollout_corr/log_ppl_diff_max:np.float64(0.008277744054794312) - rollout_corr/log_ppl_diff_min:np.float64(-0.007644861936569214) - rollout_corr/ppl_ratio:np.float64(1.0011151337530464) - rollout_corr/chi2_token:np.float64(0.0013092008884996176) - rollout_corr/chi2_seq:np.float64(0.4192327775526792) - actor/pg_loss:np.float64(0.0025457896990701556) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003396774759494292) - actor/ppo_kl:np.float64(0.0002558892238049282) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.65234375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.65234375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.5390625) - self_distillation/token_reweight_w_mean:np.float64(9165.766619424336) - self_distillation/token_reweight_w_std:np.float64(181822.19756039116) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9998559767846018) - self_distillation/token_reweight_w_clip_frac:np.float64(0.057435350317973644) - self_distillation/empty_target_batch:np.float64(0.34765625) - actor/grad_norm:np.float64(0.21796750277280807) - perf/mfu/actor:np.float64(0.0472734036832577) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(198.940327167511) - actor/lr:np.float64(1e-06) - training/global_step:59 - training/epoch:0 - critic/score/mean:0.5390625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5390625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0026581131387501955 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0026581131387501955 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:334.140625 - re
|
||||
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 60}
|
||||
validation generation end
|
||||
[prompt] system
|
||||
|
||||
Given a question and four options, please select the right answer. Respond in the following format:
|
||||
<reasoning>
|
||||
...
|
||||
</reasoning>
|
||||
<answer>
|
||||
...
|
||||
</answer>
|
||||
|
||||
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
|
||||
user
|
||||
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||||
|
||||
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||||
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||||
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||||
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||||
Please reason step by step.
|
||||
assistant
|
||||
<think>
|
||||
|
||||
</think>
|
||||
|
||||
|
||||
[response] <reasoning>
|
||||
To determine the correct reactant for the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1", we need to analyze the structure of the product and compare it with the structures of the given options.
|
||||
|
||||
The product contains a nitrogen-containing ring with a substituent group: a chlorine-substituted benzene ring connected to a trifluoromethyl group (C(F)(F)F). This suggests that the correct reactant must contain a similar structure, specifically a nitrogen-containing ring with the same substituents.
|
||||
|
||||
Option A: "O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" contains a nitrogen-containing ring with the exact substituents: a chlorine-substituted benzene ring and a trifluoromethyl group. This matches the structure of the product exactly.
|
||||
|
||||
Option B: Contains a nitrogen but not the correct substituents.
|
||||
|
||||
Option C: Contains a hydroxyl group and a different substituent, not matching the product.
|
||||
|
||||
Option D: Contains a different structure with a carbon chain and a fluorinated group, not matching the product.
|
||||
|
||||
Therefore, the only correct reactant is Option A.
|
||||
</reasoning>
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
[ground_truth] A
|
||||
[score] 1.0
|
||||
[acc] 1.0
|
||||
[pred] A
|
||||
[incorrect_format] 1
|
||||
[feedback]
|
||||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_60
|
||||
step:60 - global_seqlen/min:15969 - global_seqlen/max:20753 - global_seqlen/minmax_diff:4784 - global_seqlen/balanced_min:19004 - global_seqlen/balanced_max:19008 - global_seqlen/mean:19006.625 - actor/entropy:0.40371620655059814 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.8045270442962646 - training/rollout_probs_diff_mean:0.005058089271187782 - training/rollout_probs_diff_std:0.01212889701128006 - training/rollout_actor_probs_pearson_corr:0.9990420341491699 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.859375 - self_distillation/correct_sample_fraction:0.59375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.859375 - rollout_corr/rollout_is_mean:0.9997034668922424 - rollout_corr/rollout_is_ratio_fraction_high:1.1908589840459172e-05 - rollout_corr/rollout_is_ratio_fraction_low:4.763435936183669e-05 - rollout_corr/rollout_is_max:2.487139940261841 - rollout_corr/rollout_is_min:0.0640438050031662 - rollout_corr/rollout_is_std:0.03801152855157852 - rollout_corr/rollout_is_eff_sample_size:0.9985564358356739 - rollout_corr/rollout_is_seq_mean:0.9996296167373657 - rollout_corr/rollout_is_seq_std:0.0025690472684800625 - rollout_corr/rollout_is_seq_max:1.0076467990875244 - rollout_corr/rollout_is_seq_min:0.9901750683784485 - rollout_corr/rollout_is_seq_max_deviation:0.009824931621551514 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.534271144773811) - rollout_corr/training_log_ppl:np.float64(0.41600528117851354) - rollout_corr/kl:np.float64(0.001141798264093552) - rollout_corr/k3_kl:np.float64(0.000918725037195145) - rollout_corr/rollout_ppl:np.float64(1.5324818124063313) - rollout_corr/rollout_log_ppl:np.float64(0.4148634799348656) - rollout_corr/log_ppl_diff:np.float64(0.001141801243647933) - rollout_corr/log_ppl_abs_diff:np.float64(0.0023173787049017847) - rollout_corr/log_ppl_diff_max:np.float64(0.009830474853515625) - rollout_corr/log_ppl_diff_min:np.float64(-0.006625443696975708) - rollout_corr/ppl_ratio:np.float64(1.0011462338734418) - rollout_corr/chi2_token:np.float64(0.0013863001950085163) - rollout_corr/chi2_seq:np.float64(0.6778646670281887) - actor/pg_loss:np.float64(0.005434742895886302) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0011875206650984182) - actor/ppo_kl:np.float64(-4.705816970229648e-05) - actor/pg_clipfrac_lower:np.float64(2.3182492441264912e-05) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.859375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.859375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.59375) - self_distillation/token_reweight_w_mean:np.float64(18888.49168556533) - self_distillation/token_reweight_w_std:np.float64(311593.95773308317) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0030431225895882) - self_distillation/token_reweight_w_clip_frac:np.float64(0.10766072731348686) - self_distillation/empty_target_batch:np.float64(0.140625) - actor/grad_norm:np.float64(0.35879671573638916) - perf/mfu/actor:np.float64(0.045344773814958786) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(199.29048919677734) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.6875) - val-aux/sciknoweval/reward/std@16:np.float64(0.16029546205916068) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7511857142857143) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.13365319867771744) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6229333333333333) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.13860952641542026) - val-aux/sciknoweval/reward/
|
||||
step:61 - global_seqlen/min:15598 - global_seqlen/max:23498 - global_seqlen/minmax_diff:7900 - global_seqlen/balanced_min:19661 - global_seqlen/balanced_max:19666 - global_seqlen/mean:19664.0 - actor/entropy:0.375679075717926 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4102344214916229 - training/rollout_probs_diff_mean:0.00489032082259655 - training/rollout_probs_diff_std:0.011647011153399944 - training/rollout_actor_probs_pearson_corr:0.9990482330322266 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83984375 - self_distillation/correct_sample_fraction:0.66015625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83984375 - rollout_corr/rollout_is_mean:1.00008225440979 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.1082543096563313e-05 - rollout_corr/rollout_is_max:1.639464259147644 - rollout_corr/rollout_is_min:0.3959808051586151 - rollout_corr/rollout_is_std:0.03841473534703255 - rollout_corr/rollout_is_eff_sample_size:0.9985268392369275 - rollout_corr/rollout_is_seq_mean:1.000138759613037 - rollout_corr/rollout_is_seq_std:0.002528289332985878 - rollout_corr/rollout_is_seq_max:1.007035493850708 - rollout_corr/rollout_is_seq_min:0.993572473526001 - rollout_corr/rollout_is_seq_max_deviation:0.007035493850708008 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4903184147551656) - rollout_corr/training_log_ppl:np.float64(0.3903155774460174) - rollout_corr/kl:np.float64(0.0006912411112880434) - rollout_corr/k3_kl:np.float64(0.0009254292584728319) - rollout_corr/rollout_ppl:np.float64(1.4893131870776415) - rollout_corr/rollout_log_ppl:np.float64(0.38962433370761573) - rollout_corr/log_ppl_diff:np.float64(0.0006912437384016812) - rollout_corr/log_ppl_abs_diff:np.float64(0.0020783296204172075) - rollout_corr/log_ppl_diff_max:np.float64(0.008693814277648926) - rollout_corr/log_ppl_diff_min:np.float64(-0.006119638681411743) - rollout_corr/ppl_ratio:np.float64(1.0006948560476303) - rollout_corr/chi2_token:np.float64(0.002362349070608616) - rollout_corr/chi2_seq:np.float64(0.9650198624003679) - actor/pg_loss:np.float64(0.005624711047858) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006064838689781027) - actor/ppo_kl:np.float64(1.730111737785478e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.83984375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.83984375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.66015625) - self_distillation/token_reweight_w_mean:np.float64(5193.730784527026) - self_distillation/token_reweight_w_std:np.float64(104429.90018021135) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9993646144866943) - self_distillation/token_reweight_w_clip_frac:np.float64(0.08162092015845701) - self_distillation/empty_target_batch:np.float64(0.16015625) - actor/grad_norm:np.float64(0.354709729552269) - perf/mfu/actor:np.float64(0.04738440538771457) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(200.02718353271484) - actor/lr:np.float64(1e-06) - training/global_step:61 - training/epoch:1 - critic/score/mean:0.66015625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.66015625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.004777961410582066 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.004777961410582066 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:352.46875 - response_length/max:838.0 - r
|
||||
step:62 - global_seqlen/min:17175 - global_seqlen/max:22516 - global_seqlen/minmax_diff:5341 - global_seqlen/balanced_min:19524 - global_seqlen/balanced_max:19535 - global_seqlen/mean:19531.875 - actor/entropy:0.4294959306716919 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.20105724036693573 - training/rollout_probs_diff_mean:0.005539817735552788 - training/rollout_probs_diff_std:0.012213723734021187 - training/rollout_actor_probs_pearson_corr:0.9990189075469971 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80078125 - self_distillation/correct_sample_fraction:0.62890625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.80078125 - rollout_corr/rollout_is_mean:1.0000041723251343 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.9759001731872559 - rollout_corr/rollout_is_min:0.5684848427772522 - rollout_corr/rollout_is_std:0.04096618667244911 - rollout_corr/rollout_is_eff_sample_size:0.9983247021636537 - rollout_corr/rollout_is_seq_mean:0.9999959468841553 - rollout_corr/rollout_is_seq_std:0.0028718190733343363 - rollout_corr/rollout_is_seq_max:1.008870244026184 - rollout_corr/rollout_is_seq_min:0.987141489982605 - rollout_corr/rollout_is_seq_max_deviation:0.01285851001739502 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.584152389317751) - rollout_corr/training_log_ppl:np.float64(0.4482707906863652) - rollout_corr/kl:np.float64(0.001287838817653153) - rollout_corr/k3_kl:np.float64(0.0013142074097913792) - rollout_corr/rollout_ppl:np.float64(1.5821097679436207) - rollout_corr/rollout_log_ppl:np.float64(0.4469829489826225) - rollout_corr/log_ppl_diff:np.float64(0.0012878417037427425) - rollout_corr/log_ppl_abs_diff:np.float64(0.0026649965438991785) - rollout_corr/log_ppl_diff_max:np.float64(0.014283061027526855) - rollout_corr/log_ppl_diff_min:np.float64(-0.010016441345214844) - rollout_corr/ppl_ratio:np.float64(1.0012943353503942) - rollout_corr/chi2_token:np.float64(0.002664147410541773) - rollout_corr/chi2_seq:np.float64(0.6478703722823411) - actor/pg_loss:np.float64(0.005159451276995242) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.001237628416674852) - actor/ppo_kl:np.float64(0.0003923675134416271) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.80078125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.80078125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.62890625) - self_distillation/token_reweight_w_mean:np.float64(9608.619164069882) - self_distillation/token_reweight_w_std:np.float64(147490.825859931) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000067490618676) - self_distillation/token_reweight_w_clip_frac:np.float64(0.10762355217593722) - self_distillation/empty_target_batch:np.float64(0.19921875) - actor/grad_norm:np.float64(0.41205330193042755) - perf/mfu/actor:np.float64(0.04771627973118822) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(199.9746551513672) - actor/lr:np.float64(1e-06) - training/global_step:62 - training/epoch:1 - critic/score/mean:0.62890625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.62890625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.012106545269489288 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.012106545269489288 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:311.12109375 - response_length/max:1015.0 - response_le
|
||||
step:63 - global_seqlen/min:15986 - global_seqlen/max:20551 - global_seqlen/minmax_diff:4565 - global_seqlen/balanced_min:18758 - global_seqlen/balanced_max:18761 - global_seqlen/mean:18759.625 - actor/entropy:0.35515502095222473 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24133962392807007 - training/rollout_probs_diff_mean:0.004971614107489586 - training/rollout_probs_diff_std:0.012004957534372807 - training/rollout_actor_probs_pearson_corr:0.9989368915557861 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.53125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8125 - rollout_corr/rollout_is_mean:1.0001026391983032 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.3019985999562778e-05 - rollout_corr/rollout_is_max:1.8590831756591797 - rollout_corr/rollout_is_min:0.4984112083911896 - rollout_corr/rollout_is_std:0.03878994286060333 - rollout_corr/rollout_is_eff_sample_size:0.998497719797901 - rollout_corr/rollout_is_seq_mean:1.000045895576477 - rollout_corr/rollout_is_seq_std:0.002819458954036236 - rollout_corr/rollout_is_seq_max:1.0079857110977173 - rollout_corr/rollout_is_seq_min:0.9885396957397461 - rollout_corr/rollout_is_seq_max_deviation:0.011460304260253906 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4885337855666876) - rollout_corr/training_log_ppl:np.float64(0.38630677238688804) - rollout_corr/kl:np.float64(0.0008763675811707117) - rollout_corr/k3_kl:np.float64(0.0011061657141908654) - rollout_corr/rollout_ppl:np.float64(1.4872424700297415) - rollout_corr/rollout_log_ppl:np.float64(0.3854304051492363) - rollout_corr/log_ppl_diff:np.float64(0.0008763672376517206) - rollout_corr/log_ppl_abs_diff:np.float64(0.0025414098927285522) - rollout_corr/log_ppl_diff_max:np.float64(0.016067981719970703) - rollout_corr/log_ppl_diff_min:np.float64(-0.010530859231948853) - rollout_corr/ppl_ratio:np.float64(1.0008826348930597) - rollout_corr/chi2_token:np.float64(0.002648768713697791) - rollout_corr/chi2_seq:np.float64(1.621614676201716) - actor/pg_loss:np.float64(0.004201351199299097) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006837096566414402) - actor/ppo_kl:np.float64(9.11460165120559e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.53125) - self_distillation/token_reweight_w_mean:np.float64(328.2442679554224) - self_distillation/token_reweight_w_std:np.float64(5757.5124940559035) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0031547287944704) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0877059873600956) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.3450646325945854) - perf/mfu/actor:np.float64(0.04625351205637495) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(199.83617782592773) - actor/lr:np.float64(1e-06) - training/global_step:63 - training/epoch:1 - critic/score/mean:0.53125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.53125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.001894407905638218 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.001894407905638218 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:300.01953125 - response_length/max:723.0 - response_length/min
|
||||
step:64 - global_seqlen/min:15100 - global_seqlen/max:20599 - global_seqlen/minmax_diff:5499 - global_seqlen/balanced_min:18082 - global_seqlen/balanced_max:18090 - global_seqlen/mean:18088.125 - actor/entropy:0.3758540451526642 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24836724996566772 - training/rollout_probs_diff_mean:0.005264882929623127 - training/rollout_probs_diff_std:0.012344393879175186 - training/rollout_actor_probs_pearson_corr:0.9989309310913086 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.8984375 - self_distillation/correct_sample_fraction:0.765625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8984375 - rollout_corr/rollout_is_mean:1.0000416040420532 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.8669313905993477e-05 - rollout_corr/rollout_is_max:1.78631591796875 - rollout_corr/rollout_is_min:0.3966720402240753 - rollout_corr/rollout_is_std:0.04027950391173363 - rollout_corr/rollout_is_eff_sample_size:0.9983804272587137 - rollout_corr/rollout_is_seq_mean:1.0000438690185547 - rollout_corr/rollout_is_seq_std:0.0027673090808093548 - rollout_corr/rollout_is_seq_max:1.0073356628417969 - rollout_corr/rollout_is_seq_min:0.9913079738616943 - rollout_corr/rollout_is_seq_max_deviation:0.008692026138305664 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.488935824483633) - rollout_corr/training_log_ppl:np.float64(0.3868299189489335) - rollout_corr/kl:np.float64(0.0007672079092806428) - rollout_corr/k3_kl:np.float64(0.0009736951585637144) - rollout_corr/rollout_ppl:np.float64(1.4878136315383017) - rollout_corr/rollout_log_ppl:np.float64(0.38606271165190265) - rollout_corr/log_ppl_diff:np.float64(0.0007672072970308363) - rollout_corr/log_ppl_abs_diff:np.float64(0.0023564034490846097) - rollout_corr/log_ppl_diff_max:np.float64(0.011275023221969604) - rollout_corr/log_ppl_diff_min:np.float64(-0.008923828601837158) - rollout_corr/ppl_ratio:np.float64(1.000771946972236) - rollout_corr/chi2_token:np.float64(0.002396456664428115) - rollout_corr/chi2_seq:np.float64(0.5857332549057901) - actor/pg_loss:np.float64(0.0037766702007502317) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00045624542644873145) - actor/ppo_kl:np.float64(-8.097473682511236e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8984375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8984375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.765625) - self_distillation/token_reweight_w_mean:np.float64(34646.944526499836) - self_distillation/token_reweight_w_std:np.float64(444681.0361109813) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9998654320370406) - self_distillation/token_reweight_w_clip_frac:np.float64(0.06305540914763696) - self_distillation/empty_target_batch:np.float64(0.1015625) - actor/grad_norm:np.float64(0.2998654544353485) - perf/mfu/actor:np.float64(0.04428300784041782) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(199.82562017440796) - actor/lr:np.float64(1e-06) - training/global_step:64 - training/epoch:1 - critic/score/mean:0.765625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.765625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.009668726474046707 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.009668726474046707 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:272.50390625 - response_length/max:914.
|
||||
step:65 - global_seqlen/min:15051 - global_seqlen/max:21236 - global_seqlen/minmax_diff:6185 - global_seqlen/balanced_min:18004 - global_seqlen/balanced_max:18008 - global_seqlen/mean:18006.625 - actor/entropy:0.3955051898956299 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1947164535522461 - training/rollout_probs_diff_mean:0.0055443341843783855 - training/rollout_probs_diff_std:0.012429744005203247 - training/rollout_actor_probs_pearson_corr:0.9989457726478577 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.86328125 - self_distillation/correct_sample_fraction:0.71484375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.86328125 - rollout_corr/rollout_is_mean:0.9999876618385315 - rollout_corr/rollout_is_ratio_fraction_high:1.3995213521411642e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.031123399734497 - rollout_corr/rollout_is_min:0.504069447517395 - rollout_corr/rollout_is_std:0.041325442492961884 - rollout_corr/rollout_is_eff_sample_size:0.9982950004968502 - rollout_corr/rollout_is_seq_mean:0.9999707341194153 - rollout_corr/rollout_is_seq_std:0.002948605688288808 - rollout_corr/rollout_is_seq_max:1.009615182876587 - rollout_corr/rollout_is_seq_min:0.9913778305053711 - rollout_corr/rollout_is_seq_max_deviation:0.009615182876586914 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.525312386918813) - rollout_corr/training_log_ppl:np.float64(0.4125790528487414) - rollout_corr/kl:np.float64(0.00100548441882653) - rollout_corr/k3_kl:np.float64(0.001056205478960237) - rollout_corr/rollout_ppl:np.float64(1.5237079304642975) - rollout_corr/rollout_log_ppl:np.float64(0.4115735680970829) - rollout_corr/log_ppl_diff:np.float64(0.0010054847516585141) - rollout_corr/log_ppl_abs_diff:np.float64(0.0024253172159660608) - rollout_corr/log_ppl_diff_max:np.float64(0.012961804866790771) - rollout_corr/log_ppl_diff_min:np.float64(-0.007540702819824219) - rollout_corr/ppl_ratio:np.float64(1.0010107702109963) - rollout_corr/chi2_token:np.float64(0.0022052584681659937) - rollout_corr/chi2_seq:np.float64(0.9272577015217394) - actor/pg_loss:np.float64(0.003355647437274456) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.000539568421572767) - actor/ppo_kl:np.float64(7.455754089491506e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.86328125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.86328125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.71484375) - self_distillation/token_reweight_w_mean:np.float64(178.37819157890044) - self_distillation/token_reweight_w_std:np.float64(3052.0747698339983) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9989271517843008) - self_distillation/token_reweight_w_clip_frac:np.float64(0.07668937157723121) - self_distillation/empty_target_batch:np.float64(0.13671875) - actor/grad_norm:np.float64(0.31669436022639275) - perf/mfu/actor:np.float64(0.04449943785630652) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.69052124023438) - actor/lr:np.float64(1e-06) - training/global_step:65 - training/epoch:1 - critic/score/mean:0.71484375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.71484375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.013841266743838787 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.013841266743838787 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:279.11328125 - response_length/ma
|
||||
step:66 - global_seqlen/min:16554 - global_seqlen/max:19900 - global_seqlen/minmax_diff:3346 - global_seqlen/balanced_min:18653 - global_seqlen/balanced_max:18704 - global_seqlen/mean:18662.625 - actor/entropy:0.3692823350429535 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24191009998321533 - training/rollout_probs_diff_mean:0.005031523294746876 - training/rollout_probs_diff_std:0.011805592104792595 - training/rollout_actor_probs_pearson_corr:0.998999834060669 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.7109375 - self_distillation/correct_sample_fraction:0.6171875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7109375 - rollout_corr/rollout_is_mean:0.9999994039535522 - rollout_corr/rollout_is_ratio_fraction_high:1.3333866263565142e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.2511892318725586 - rollout_corr/rollout_is_min:0.5517903566360474 - rollout_corr/rollout_is_std:0.03924669697880745 - rollout_corr/rollout_is_eff_sample_size:0.998462065654954 - rollout_corr/rollout_is_seq_mean:1.0000253915786743 - rollout_corr/rollout_is_seq_std:0.002698667347431183 - rollout_corr/rollout_is_seq_max:1.0073356628417969 - rollout_corr/rollout_is_seq_min:0.992901086807251 - rollout_corr/rollout_is_seq_max_deviation:0.007335662841796875 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5085271298885345) - rollout_corr/training_log_ppl:np.float64(0.4001465483015636) - rollout_corr/kl:np.float64(0.0009092299301867168) - rollout_corr/k3_kl:np.float64(0.001055825624348472) - rollout_corr/rollout_ppl:np.float64(1.5071750050410628) - rollout_corr/rollout_log_ppl:np.float64(0.3992373197543202) - rollout_corr/log_ppl_diff:np.float64(0.0009092285472434014) - rollout_corr/log_ppl_abs_diff:np.float64(0.002248107542982325) - rollout_corr/log_ppl_diff_max:np.float64(0.008349627256393433) - rollout_corr/log_ppl_diff_min:np.float64(-0.008058488368988037) - rollout_corr/ppl_ratio:np.float64(1.0009135268628597) - rollout_corr/chi2_token:np.float64(0.0025282949209213257) - rollout_corr/chi2_seq:np.float64(0.6103038967121392) - actor/pg_loss:np.float64(0.002379831741563976) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00033230165354325436) - actor/ppo_kl:np.float64(8.482311253033004e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.7109375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.7109375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6171875) - self_distillation/token_reweight_w_mean:np.float64(14.794726703781635) - self_distillation/token_reweight_w_std:np.float64(315.56329661473865) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9989614612422884) - self_distillation/token_reweight_w_clip_frac:np.float64(0.04611051009851508) - self_distillation/empty_target_batch:np.float64(0.2890625) - actor/grad_norm:np.float64(0.24986116960644722) - perf/mfu/actor:np.float64(0.044863735441974) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.69231033325195) - actor/lr:np.float64(1e-06) - training/global_step:66 - training/epoch:1 - critic/score/mean:0.6171875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6171875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0023150925990194082 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0023150925990194082 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:292.95703125 - response_length/max:10
|
||||
step:67 - global_seqlen/min:14936 - global_seqlen/max:21018 - global_seqlen/minmax_diff:6082 - global_seqlen/balanced_min:18548 - global_seqlen/balanced_max:18548 - global_seqlen/mean:18548.0 - actor/entropy:0.4010680615901947 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18490827083587646 - training/rollout_probs_diff_mean:0.005276753567159176 - training/rollout_probs_diff_std:0.011970635503530502 - training/rollout_actor_probs_pearson_corr:0.9990053772926331 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.75 - self_distillation/correct_sample_fraction:0.6796875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.75 - rollout_corr/rollout_is_mean:1.0001569986343384 - rollout_corr/rollout_is_ratio_fraction_high:1.357367818854982e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.357367818854982e-05 - rollout_corr/rollout_is_max:2.239859104156494 - rollout_corr/rollout_is_min:0.4357939660549164 - rollout_corr/rollout_is_std:0.03935437649488449 - rollout_corr/rollout_is_eff_sample_size:0.9984541032441931 - rollout_corr/rollout_is_seq_mean:1.0001627206802368 - rollout_corr/rollout_is_seq_std:0.002580940490588546 - rollout_corr/rollout_is_seq_max:1.0100871324539185 - rollout_corr/rollout_is_seq_min:0.9937511682510376 - rollout_corr/rollout_is_seq_max_deviation:0.010087132453918457 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5274818786419928) - rollout_corr/training_log_ppl:np.float64(0.41380137618398294) - rollout_corr/kl:np.float64(0.0010146739747582956) - rollout_corr/k3_kl:np.float64(0.0009835226737777703) - rollout_corr/rollout_ppl:np.float64(1.5259358463808894) - rollout_corr/rollout_log_ppl:np.float64(0.41278669840539806) - rollout_corr/log_ppl_diff:np.float64(0.0010146777785848826) - rollout_corr/log_ppl_abs_diff:np.float64(0.0021784571872558445) - rollout_corr/log_ppl_diff_max:np.float64(0.014455139636993408) - rollout_corr/log_ppl_diff_min:np.float64(-0.0060721635818481445) - rollout_corr/ppl_ratio:np.float64(1.0010192152112722) - rollout_corr/chi2_token:np.float64(0.0019758078269660473) - rollout_corr/chi2_seq:np.float64(0.4387206360697746) - actor/pg_loss:np.float64(0.0029250025982037187) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00039720970880807727) - actor/ppo_kl:np.float64(0.0003566578742226767) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.75) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.75) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6796875) - self_distillation/token_reweight_w_mean:np.float64(547.347474707989) - self_distillation/token_reweight_w_std:np.float64(8326.106724163285) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9982913499698043) - self_distillation/token_reweight_w_clip_frac:np.float64(0.06155556399608031) - self_distillation/empty_target_batch:np.float64(0.25) - actor/grad_norm:np.float64(0.2857639491558075) - perf/mfu/actor:np.float64(0.04548089224569089) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.682861328125) - actor/lr:np.float64(1e-06) - training/global_step:67 - training/epoch:1 - critic/score/mean:0.6796875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6796875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0016373249236494303 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0016373249236494303 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:287.78125 - response_length/max:673.0 - respon
|
||||
step:68 - global_seqlen/min:15394 - global_seqlen/max:21848 - global_seqlen/minmax_diff:6454 - global_seqlen/balanced_min:19111 - global_seqlen/balanced_max:19113 - global_seqlen/mean:19112.0 - actor/entropy:0.454579621553421 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2294895052909851 - training/rollout_probs_diff_mean:0.005558677017688751 - training/rollout_probs_diff_std:0.011987541802227497 - training/rollout_actor_probs_pearson_corr:0.999093234539032 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.6484375 - self_distillation/correct_sample_fraction:0.46875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.6484375 - rollout_corr/rollout_is_mean:0.999990701675415 - rollout_corr/rollout_is_ratio_fraction_high:2.770083119685296e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.329737901687622 - rollout_corr/rollout_is_min:0.5801910758018494 - rollout_corr/rollout_is_std:0.04091522842645645 - rollout_corr/rollout_is_eff_sample_size:0.9983287417268669 - rollout_corr/rollout_is_seq_mean:0.9999210834503174 - rollout_corr/rollout_is_seq_std:0.0029238073620945215 - rollout_corr/rollout_is_seq_max:1.010511040687561 - rollout_corr/rollout_is_seq_min:0.9896543025970459 - rollout_corr/rollout_is_seq_max_deviation:0.010511040687561035 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6329940822906792) - rollout_corr/training_log_ppl:np.float64(0.4769620407023467) - rollout_corr/kl:np.float64(0.0010583482470849503) - rollout_corr/k3_kl:np.float64(0.0009939358607766735) - rollout_corr/rollout_ppl:np.float64(1.631263299845159) - rollout_corr/rollout_log_ppl:np.float64(0.47590369096724316) - rollout_corr/log_ppl_diff:np.float64(0.0010583497351035476) - rollout_corr/log_ppl_abs_diff:np.float64(0.0024261822691187263) - rollout_corr/log_ppl_diff_max:np.float64(0.012225866317749023) - rollout_corr/log_ppl_diff_min:np.float64(-0.008922696113586426) - rollout_corr/ppl_ratio:np.float64(1.0010634786449373) - rollout_corr/chi2_token:np.float64(0.0018958586733788252) - rollout_corr/chi2_seq:np.float64(0.7360206483863294) - actor/pg_loss:np.float64(0.0039513889933004975) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004996858078811783) - actor/ppo_kl:np.float64(9.023317768352968e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.6484375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.6484375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.46875) - self_distillation/token_reweight_w_mean:np.float64(16211.263574679615) - self_distillation/token_reweight_w_std:np.float64(247760.0435439156) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0013059759512544) - self_distillation/token_reweight_w_clip_frac:np.float64(0.09038829631754197) - self_distillation/empty_target_batch:np.float64(0.3515625) - actor/grad_norm:np.float64(0.35276055335998535) - perf/mfu/actor:np.float64(0.04664085001944063) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.4664192199707) - actor/lr:np.float64(1e-06) - training/global_step:68 - training/epoch:1 - critic/score/mean:0.46875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.46875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0022922437638044357 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0022922437638044357 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:282.03125 - response_length/max:777.0 - response_l
|
||||
step:69 - global_seqlen/min:17581 - global_seqlen/max:21958 - global_seqlen/minmax_diff:4377 - global_seqlen/balanced_min:20257 - global_seqlen/balanced_max:20260 - global_seqlen/mean:20258.25 - actor/entropy:0.46467649936676025 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24135446548461914 - training/rollout_probs_diff_mean:0.00523096090182662 - training/rollout_probs_diff_std:0.011185083538293839 - training/rollout_actor_probs_pearson_corr:0.999234676361084 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - self_distillation/correct_sample_fraction:0.6796875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8359375 - rollout_corr/rollout_is_mean:1.0000218152999878 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6324684619903564 - rollout_corr/rollout_is_min:0.5706769227981567 - rollout_corr/rollout_is_std:0.03825926035642624 - rollout_corr/rollout_is_eff_sample_size:0.998538606349212 - rollout_corr/rollout_is_seq_mean:1.0000791549682617 - rollout_corr/rollout_is_seq_std:0.002675000810995698 - rollout_corr/rollout_is_seq_max:1.0090395212173462 - rollout_corr/rollout_is_seq_min:0.989310622215271 - rollout_corr/rollout_is_seq_max_deviation:0.010689377784729004 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.646835464052856) - rollout_corr/training_log_ppl:np.float64(0.4856712990440428) - rollout_corr/kl:np.float64(0.0008948077715515979) - rollout_corr/k3_kl:np.float64(0.0010960273730233894) - rollout_corr/rollout_ppl:np.float64(1.6454020515084267) - rollout_corr/rollout_log_ppl:np.float64(0.48477648955304176) - rollout_corr/log_ppl_diff:np.float64(0.0008948094910010695) - rollout_corr/log_ppl_abs_diff:np.float64(0.0023694130359217525) - rollout_corr/log_ppl_diff_max:np.float64(0.012593626976013184) - rollout_corr/log_ppl_diff_min:np.float64(-0.010994911193847656) - rollout_corr/ppl_ratio:np.float64(1.0008999034762383) - rollout_corr/chi2_token:np.float64(0.0027323926333338022) - rollout_corr/chi2_seq:np.float64(1.6403908263891935) - actor/pg_loss:np.float64(0.004149097250774503) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006782121126889251) - actor/ppo_kl:np.float64(0.0001744897412141455) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8359375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8359375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6796875) - self_distillation/token_reweight_w_mean:np.float64(112.03770920354873) - self_distillation/token_reweight_w_std:np.float64(1537.8677468675305) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9990846202708781) - self_distillation/token_reweight_w_clip_frac:np.float64(0.10416805295972154) - self_distillation/empty_target_batch:np.float64(0.1640625) - actor/grad_norm:np.float64(0.3852573037147522) - perf/mfu/actor:np.float64(0.04884632323523585) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.50916624069214) - actor/lr:np.float64(1e-06) - training/global_step:69 - training/epoch:1 - critic/score/mean:0.6796875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6796875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.001025630161166191 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.001025630161166191 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:323.2578125 - response_length/max:740.0 - response_length/m
|
||||
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 70}
|
||||
validation generation end
|
||||
[prompt] system
|
||||
|
||||
Given a question and four options, please select the right answer. Respond in the following format:
|
||||
<reasoning>
|
||||
...
|
||||
</reasoning>
|
||||
<answer>
|
||||
...
|
||||
</answer>
|
||||
|
||||
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
|
||||
user
|
||||
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||||
|
||||
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||||
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||||
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||||
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||||
Please reason step by step.
|
||||
assistant
|
||||
<think>
|
||||
|
||||
</think>
|
||||
|
||||
|
||||
[response] <reasoning>
|
||||
To determine the correct reactant for the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1", we must analyze the structure and functional groups present in both the product and the options.
|
||||
|
||||
The product contains:
|
||||
- A nitrogen-containing ring (likely a pyrrole or similar heterocycle).
|
||||
- A chlorinated aromatic ring (–Cl group).
|
||||
- A trifluoromethyl group (–C(F)(F)F).
|
||||
- A nitrogen atom bonded to a carbon chain that includes the chlorinated aromatic ring.
|
||||
|
||||
Now, let's examine the options:
|
||||
|
||||
**Option A**: Contains a chlorinated aromatic ring, a trifluoromethyl group, and a nitrogen-containing ring with a nitrogen atom bonded to a carbon chain that includes the chlorinated aromatic ring. It closely matches the structure of the product and is a strong candidate.
|
||||
|
||||
**Option B**: Contains a chlorinated aromatic ring but lacks the trifluoromethyl group and the correct nitrogen-containing ring structure. It is not a match.
|
||||
|
||||
**Option C**: Contains a chlorinated aromatic ring and a trifluoromethyl group, but the nitrogen-containing ring is not correctly structured. It does not match the product.
|
||||
|
||||
**Option D**: Contains a trifluoromethyl group and a carbon chain with a chlorinated aromatic ring, but the nitrogen-containing ring is not correctly structured. It also does not match the product.
|
||||
|
||||
Thus, **Option A** is the only one that contains all the necessary functional groups and structural features required to synthesize the product.
|
||||
|
||||
</reasoning>
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
[ground_truth] A
|
||||
[score] 1.0
|
||||
[acc] 1.0
|
||||
[pred] A
|
||||
[incorrect_format] 1
|
||||
[feedback]
|
||||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_70
|
||||
step:70 - global_seqlen/min:14508 - global_seqlen/max:21779 - global_seqlen/minmax_diff:7271 - global_seqlen/balanced_min:19295 - global_seqlen/balanced_max:19376 - global_seqlen/mean:19309.25 - actor/entropy:0.48265349864959717 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2931375205516815 - training/rollout_probs_diff_mean:0.005450577940791845 - training/rollout_probs_diff_std:0.011540403589606285 - training/rollout_actor_probs_pearson_corr:0.9992024302482605 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.796875 - self_distillation/correct_sample_fraction:0.62890625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.796875 - rollout_corr/rollout_is_mean:1.0000245571136475 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6136518716812134 - rollout_corr/rollout_is_min:0.5989248752593994 - rollout_corr/rollout_is_std:0.03922086954116821 - rollout_corr/rollout_is_eff_sample_size:0.9984642048317262 - rollout_corr/rollout_is_seq_mean:1.0000163316726685 - rollout_corr/rollout_is_seq_std:0.002710202941671014 - rollout_corr/rollout_is_seq_max:1.0095508098602295 - rollout_corr/rollout_is_seq_min:0.9921322464942932 - rollout_corr/rollout_is_seq_max_deviation:0.009550809860229492 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6673680869862437) - rollout_corr/training_log_ppl:np.float64(0.4960062170866877) - rollout_corr/kl:np.float64(0.0012155248211556113) - rollout_corr/k3_kl:np.float64(0.0012827583032049006) - rollout_corr/rollout_ppl:np.float64(1.6654691863805056) - rollout_corr/rollout_log_ppl:np.float64(0.4947906881570816) - rollout_corr/log_ppl_diff:np.float64(0.00121552892960608) - rollout_corr/log_ppl_abs_diff:np.float64(0.0026022420497611165) - rollout_corr/log_ppl_diff_max:np.float64(0.016035795211791992) - rollout_corr/log_ppl_diff_min:np.float64(-0.00719606876373291) - rollout_corr/ppl_ratio:np.float64(1.0012216775212437) - rollout_corr/chi2_token:np.float64(0.002924366621300578) - rollout_corr/chi2_seq:np.float64(1.3708638157695532) - actor/pg_loss:np.float64(0.0026991143822669983) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006329814200398687) - actor/ppo_kl:np.float64(0.0004067609955740181) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.796875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.796875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.62890625) - self_distillation/token_reweight_w_mean:np.float64(1099.6124461798463) - self_distillation/token_reweight_w_std:np.float64(17668.276791150623) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0005380015354604) - self_distillation/token_reweight_w_clip_frac:np.float64(0.08962540709762834) - self_distillation/empty_target_batch:np.float64(0.203125) - actor/grad_norm:np.float64(0.30823376774787903) - perf/mfu/actor:np.float64(0.04660907978159935) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.58329391479492) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.6967261904761904) - val-aux/sciknoweval/reward/std@16:np.float64(0.15836902316671747) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7580857142857141) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.12446919093953822) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6347571428571429) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.1415950973799327) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.696961904761904
|
||||
step:71 - global_seqlen/min:19243 - global_seqlen/max:22825 - global_seqlen/minmax_diff:3582 - global_seqlen/balanced_min:21239 - global_seqlen/balanced_max:21249 - global_seqlen/mean:21244.625 - actor/entropy:0.48243123292922974 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1782420575618744 - training/rollout_probs_diff_mean:0.004945038352161646 - training/rollout_probs_diff_std:0.010490105487406254 - training/rollout_actor_probs_pearson_corr:0.9993298649787903 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.73828125 - self_distillation/correct_sample_fraction:0.58984375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.73828125 - rollout_corr/rollout_is_mean:1.0001386404037476 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.646081805229187 - rollout_corr/rollout_is_min:0.6092609167098999 - rollout_corr/rollout_is_std:0.036472681909799576 - rollout_corr/rollout_is_eff_sample_size:0.998671986340567 - rollout_corr/rollout_is_seq_mean:1.00009286403656 - rollout_corr/rollout_is_seq_std:0.0023478141520172358 - rollout_corr/rollout_is_seq_max:1.0064475536346436 - rollout_corr/rollout_is_seq_min:0.9918783903121948 - rollout_corr/rollout_is_seq_max_deviation:0.008121609687805176 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6814591218717396) - rollout_corr/training_log_ppl:np.float64(0.5044128161389381) - rollout_corr/kl:np.float64(0.0009273746180333831) - rollout_corr/k3_kl:np.float64(0.0009558818163668548) - rollout_corr/rollout_ppl:np.float64(1.679951566271484) - rollout_corr/rollout_log_ppl:np.float64(0.5034854415280279) - rollout_corr/log_ppl_diff:np.float64(0.0009273746109101921) - rollout_corr/log_ppl_abs_diff:np.float64(0.0021467784827109426) - rollout_corr/log_ppl_diff_max:np.float64(0.009796887636184692) - rollout_corr/log_ppl_diff_min:np.float64(-0.007009267807006836) - rollout_corr/ppl_ratio:np.float64(1.0009312871843576) - rollout_corr/chi2_token:np.float64(0.002058735117316246) - rollout_corr/chi2_seq:np.float64(0.9684193118009716) - actor/pg_loss:np.float64(0.002505662152543664) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00036506839978756034) - actor/ppo_kl:np.float64(0.0002983230811750559) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.73828125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.73828125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.58984375) - self_distillation/token_reweight_w_mean:np.float64(1927.8832940307911) - self_distillation/token_reweight_w_std:np.float64(37339.10144847393) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0009577083401382) - self_distillation/token_reweight_w_clip_frac:np.float64(0.08062945469282568) - self_distillation/empty_target_batch:np.float64(0.26171875) - actor/grad_norm:np.float64(0.5023428723216057) - perf/mfu/actor:np.float64(0.051698067264255104) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(201.09897232055664) - actor/lr:np.float64(1e-06) - training/global_step:71 - training/epoch:1 - critic/score/mean:0.58984375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.58984375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0013479345943778753 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0013479345943778753 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:361.51953125 - response_length/max:967.0 - respons
|
||||
step:72 - global_seqlen/min:16380 - global_seqlen/max:21611 - global_seqlen/minmax_diff:5231 - global_seqlen/balanced_min:19858 - global_seqlen/balanced_max:19861 - global_seqlen/mean:19860.0 - actor/entropy:0.5429559946060181 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.15169164538383484 - training/rollout_probs_diff_mean:0.005359958857297897 - training/rollout_probs_diff_std:0.010805261321365833 - training/rollout_actor_probs_pearson_corr:0.9993027448654175 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83984375 - self_distillation/correct_sample_fraction:0.671875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83984375 - rollout_corr/rollout_is_mean:1.0000475645065308 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.644222617149353 - rollout_corr/rollout_is_min:0.6008924245834351 - rollout_corr/rollout_is_std:0.03811253234744072 - rollout_corr/rollout_is_eff_sample_size:0.9985496605558143 - rollout_corr/rollout_is_seq_mean:0.9999592304229736 - rollout_corr/rollout_is_seq_std:0.0024577102158218622 - rollout_corr/rollout_is_seq_max:1.0064175128936768 - rollout_corr/rollout_is_seq_min:0.9917982220649719 - rollout_corr/rollout_is_seq_max_deviation:0.008201777935028076 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7471187515184283) - rollout_corr/training_log_ppl:np.float64(0.5459236325696111) - rollout_corr/kl:np.float64(0.0010623894131924772) - rollout_corr/k3_kl:np.float64(0.0008694290827975237) - rollout_corr/rollout_ppl:np.float64(1.745131317526102) - rollout_corr/rollout_log_ppl:np.float64(0.5448612435720861) - rollout_corr/log_ppl_diff:np.float64(0.0010623889975249767) - rollout_corr/log_ppl_abs_diff:np.float64(0.0021515846019610763) - rollout_corr/log_ppl_diff_max:np.float64(0.010035932064056396) - rollout_corr/log_ppl_diff_min:np.float64(-0.00523415207862854) - rollout_corr/ppl_ratio:np.float64(1.0010662653949112) - rollout_corr/chi2_token:np.float64(0.001382074085995555) - rollout_corr/chi2_seq:np.float64(0.6654008245095611) - actor/pg_loss:np.float64(0.005195157486014068) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003912408333235362) - actor/ppo_kl:np.float64(0.0002666896097824889) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.83984375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.83984375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.671875) - self_distillation/token_reweight_w_mean:np.float64(7515.255449632183) - self_distillation/token_reweight_w_std:np.float64(120734.30576296372) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9992036235053092) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1153781593311578) - self_distillation/empty_target_batch:np.float64(0.16015625) - actor/grad_norm:np.float64(0.3598603457212448) - perf/mfu/actor:np.float64(0.048463804802359345) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(200.67909240722656) - actor/lr:np.float64(1e-06) - training/global_step:72 - training/epoch:1 - critic/score/mean:0.671875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.671875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0006963299238123 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0006963299238123 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:333.78125 - response_length/max:799.0 - response_length/min:152.
|
||||
step:73 - global_seqlen/min:19647 - global_seqlen/max:22925 - global_seqlen/minmax_diff:3278 - global_seqlen/balanced_min:21192 - global_seqlen/balanced_max:21202 - global_seqlen/mean:21199.0 - actor/entropy:0.5374614596366882 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.301607221364975 - training/rollout_probs_diff_mean:0.004932303912937641 - training/rollout_probs_diff_std:0.010089674964547157 - training/rollout_actor_probs_pearson_corr:0.9993661642074585 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.7734375 - self_distillation/correct_sample_fraction:0.640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7734375 - rollout_corr/rollout_is_mean:0.9999767541885376 - rollout_corr/rollout_is_ratio_fraction_high:1.021909702103585e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.1998536586761475 - rollout_corr/rollout_is_min:0.6301840543746948 - rollout_corr/rollout_is_std:0.03611796349287033 - rollout_corr/rollout_is_eff_sample_size:0.9986969544688866 - rollout_corr/rollout_is_seq_mean:1.0000014305114746 - rollout_corr/rollout_is_seq_std:0.0019163991091772914 - rollout_corr/rollout_is_seq_max:1.0062246322631836 - rollout_corr/rollout_is_seq_min:0.9927696585655212 - rollout_corr/rollout_is_seq_max_deviation:0.00723034143447876 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7496894244104624) - rollout_corr/training_log_ppl:np.float64(0.5468992951209657) - rollout_corr/kl:np.float64(0.0007367991016540287) - rollout_corr/k3_kl:np.float64(0.0007361613289731395) - rollout_corr/rollout_ppl:np.float64(1.7483539679087698) - rollout_corr/rollout_log_ppl:np.float64(0.5461624904419295) - rollout_corr/log_ppl_diff:np.float64(0.0007368046790361404) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016733732772991061) - rollout_corr/log_ppl_diff_max:np.float64(0.009091198444366455) - rollout_corr/log_ppl_diff_min:np.float64(-0.005505651235580444) - rollout_corr/ppl_ratio:np.float64(1.0007393239066005) - rollout_corr/chi2_token:np.float64(0.0014872262254357338) - rollout_corr/chi2_seq:np.float64(0.43628761847503483) - actor/pg_loss:np.float64(0.0029139972757548094) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00018539705752118607) - actor/ppo_kl:np.float64(6.380892117796577e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.7734375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.7734375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.640625) - self_distillation/token_reweight_w_mean:np.float64(1624.953690484399) - self_distillation/token_reweight_w_std:np.float64(32560.52424334915) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9991668974980712) - self_distillation/token_reweight_w_clip_frac:np.float64(0.06964316591620445) - self_distillation/empty_target_batch:np.float64(0.2265625) - actor/grad_norm:np.float64(0.21447288990020752) - perf/mfu/actor:np.float64(0.050516435918862526) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(200.0999493598938) - actor/lr:np.float64(1e-06) - training/global_step:73 - training/epoch:1 - critic/score/mean:0.640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.007218515034765005 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.007218515034765005 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:382.25 - response_length/max:996.0 - respon
|
||||
step:74 - global_seqlen/min:18287 - global_seqlen/max:25091 - global_seqlen/minmax_diff:6804 - global_seqlen/balanced_min:20904 - global_seqlen/balanced_max:20912 - global_seqlen/mean:20910.0 - actor/entropy:0.5127137303352356 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.27330654859542847 - training/rollout_probs_diff_mean:0.004899736028164625 - training/rollout_probs_diff_std:0.010343815200030804 - training/rollout_actor_probs_pearson_corr:0.9993671774864197 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - self_distillation/correct_sample_fraction:0.703125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83203125 - rollout_corr/rollout_is_mean:1.000099539756775 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.0278759873472154e-05 - rollout_corr/rollout_is_max:1.7854251861572266 - rollout_corr/rollout_is_min:0.48449382185935974 - rollout_corr/rollout_is_std:0.0365249402821064 - rollout_corr/rollout_is_eff_sample_size:0.9986679439991228 - rollout_corr/rollout_is_seq_mean:1.000190258026123 - rollout_corr/rollout_is_seq_std:0.0022744936868548393 - rollout_corr/rollout_is_seq_max:1.0073591470718384 - rollout_corr/rollout_is_seq_min:0.9917874336242676 - rollout_corr/rollout_is_seq_max_deviation:0.008212566375732422 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7097132932394743) - rollout_corr/training_log_ppl:np.float64(0.5229829574818723) - rollout_corr/kl:np.float64(0.0006870645115757412) - rollout_corr/k3_kl:np.float64(0.0007633993515128168) - rollout_corr/rollout_ppl:np.float64(1.7085823002271354) - rollout_corr/rollout_log_ppl:np.float64(0.5222958918311633) - rollout_corr/log_ppl_diff:np.float64(0.0006870656507089734) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019122406374663115) - rollout_corr/log_ppl_diff_max:np.float64(0.008660197257995605) - rollout_corr/log_ppl_diff_min:np.float64(-0.005268260836601257) - rollout_corr/ppl_ratio:np.float64(1.000690019223839) - rollout_corr/chi2_token:np.float64(0.0016905386000871658) - rollout_corr/chi2_seq:np.float64(1.4013629674445838) - actor/pg_loss:np.float64(0.0029980618273839355) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00017913388501256122) - actor/ppo_kl:np.float64(0.00016305825718454514) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.83203125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.83203125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.703125) - self_distillation/token_reweight_w_mean:np.float64(15.748328928370029) - self_distillation/token_reweight_w_std:np.float64(289.3728471599752) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0004254090599716) - self_distillation/token_reweight_w_clip_frac:np.float64(0.06948782474501058) - self_distillation/empty_target_batch:np.float64(0.16796875) - actor/grad_norm:np.float64(0.2460213266313076) - perf/mfu/actor:np.float64(0.0504502030203242) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.804443359375) - actor/lr:np.float64(1e-06) - training/global_step:74 - training/epoch:1 - critic/score/mean:0.703125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.703125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.005830626469105482 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.005830626469105482 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:380.03125 - response_length/max:953.0 -
|
||||
step:75 - global_seqlen/min:16481 - global_seqlen/max:26987 - global_seqlen/minmax_diff:10506 - global_seqlen/balanced_min:22439 - global_seqlen/balanced_max:22447 - global_seqlen/mean:22444.625 - actor/entropy:0.5642927289009094 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2400461733341217 - training/rollout_probs_diff_mean:0.0051166643388569355 - training/rollout_probs_diff_std:0.010140416212379932 - training/rollout_actor_probs_pearson_corr:0.9994109869003296 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8046875 - self_distillation/correct_sample_fraction:0.71484375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8046875 - rollout_corr/rollout_is_mean:0.9999061822891235 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.0068769370263908e-05 - rollout_corr/rollout_is_max:1.5713963508605957 - rollout_corr/rollout_is_min:0.4978005588054657 - rollout_corr/rollout_is_std:0.03701609745621681 - rollout_corr/rollout_is_eff_sample_size:0.9986314455154365 - rollout_corr/rollout_is_seq_mean:0.999901294708252 - rollout_corr/rollout_is_seq_std:0.0019736075773835182 - rollout_corr/rollout_is_seq_max:1.0060852766036987 - rollout_corr/rollout_is_seq_min:0.9931412935256958 - rollout_corr/rollout_is_seq_max_deviation:0.006858706474304199 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.791621611919254) - rollout_corr/training_log_ppl:np.float64(0.570267406059429) - rollout_corr/kl:np.float64(0.0008821988329259511) - rollout_corr/k3_kl:np.float64(0.0008163573830870519) - rollout_corr/rollout_ppl:np.float64(1.7901061396114528) - rollout_corr/rollout_log_ppl:np.float64(0.56938520638505) - rollout_corr/log_ppl_diff:np.float64(0.0008821996743790805) - rollout_corr/log_ppl_abs_diff:np.float64(0.00185499427607283) - rollout_corr/log_ppl_diff_max:np.float64(0.00786745548248291) - rollout_corr/log_ppl_diff_min:np.float64(-0.005721062421798706) - rollout_corr/ppl_ratio:np.float64(1.000884982291609) - rollout_corr/chi2_token:np.float64(0.001510836649686098) - rollout_corr/chi2_seq:np.float64(0.9038489798549563) - actor/pg_loss:np.float64(0.002076247939839959) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00023045666375764995) - actor/ppo_kl:np.float64(5.9166708297198056e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8046875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8046875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.71484375) - self_distillation/token_reweight_w_mean:np.float64(6473.88584861136) - self_distillation/token_reweight_w_std:np.float64(110608.01085398125) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9989379325415939) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0619521141925361) - self_distillation/empty_target_batch:np.float64(0.1953125) - actor/grad_norm:np.float64(0.19757572747766972) - perf/mfu/actor:np.float64(0.053976922098505964) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.7879409790039) - actor/lr:np.float64(1e-06) - training/global_step:75 - training/epoch:1 - critic/score/mean:0.71484375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.71484375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0026480865199118853 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0026480865199118853 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:387.95703125 - response_length/max:91
|
||||
step:76 - global_seqlen/min:20385 - global_seqlen/max:23735 - global_seqlen/minmax_diff:3350 - global_seqlen/balanced_min:22458 - global_seqlen/balanced_max:22461 - global_seqlen/mean:22459.875 - actor/entropy:0.5038673281669617 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.17917871475219727 - training/rollout_probs_diff_mean:0.004544694907963276 - training/rollout_probs_diff_std:0.009781488217413425 - training/rollout_actor_probs_pearson_corr:0.9994357228279114 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.7734375 - self_distillation/correct_sample_fraction:0.6171875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7734375 - rollout_corr/rollout_is_mean:0.999882698059082 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.531444787979126 - rollout_corr/rollout_is_min:0.6298530101776123 - rollout_corr/rollout_is_std:0.03470920771360397 - rollout_corr/rollout_is_eff_sample_size:0.9987963637779437 - rollout_corr/rollout_is_seq_mean:0.9998737573623657 - rollout_corr/rollout_is_seq_std:0.001968828495591879 - rollout_corr/rollout_is_seq_max:1.0092756748199463 - rollout_corr/rollout_is_seq_min:0.9928333163261414 - rollout_corr/rollout_is_seq_max_deviation:0.009275674819946289 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7229888113215566) - rollout_corr/training_log_ppl:np.float64(0.5273766612517647) - rollout_corr/kl:np.float64(0.0006758117173214506) - rollout_corr/k3_kl:np.float64(0.0006499772979395857) - rollout_corr/rollout_ppl:np.float64(1.72180867055431) - rollout_corr/rollout_log_ppl:np.float64(0.526700847200118) - rollout_corr/log_ppl_diff:np.float64(0.0006758140516467392) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016295489040203393) - rollout_corr/log_ppl_diff_max:np.float64(0.007842868566513062) - rollout_corr/log_ppl_diff_min:np.float64(-0.005845367908477783) - rollout_corr/ppl_ratio:np.float64(1.0006780109833926) - rollout_corr/chi2_token:np.float64(0.0012538495939224958) - rollout_corr/chi2_seq:np.float64(0.6266895108856261) - actor/pg_loss:np.float64(0.0024852192727848887) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00022367794190358836) - actor/ppo_kl:np.float64(-8.031681899822729e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.7734375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.7734375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6171875) - self_distillation/token_reweight_w_mean:np.float64(11.586508150445297) - self_distillation/token_reweight_w_std:np.float64(205.42800806532614) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9991163620725274) - self_distillation/token_reweight_w_clip_frac:np.float64(0.07528238961822353) - self_distillation/empty_target_batch:np.float64(0.2265625) - actor/grad_norm:np.float64(0.21515963971614838) - perf/mfu/actor:np.float64(0.05377171741192728) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.86333084106445) - actor/lr:np.float64(1e-06) - training/global_step:76 - training/epoch:1 - critic/score/mean:0.6171875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6171875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0009749584714882076 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0009749584714882076 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:416.18359375 - response_length/max:835.0 - response_
|
||||
step:77 - global_seqlen/min:19528 - global_seqlen/max:26165 - global_seqlen/minmax_diff:6637 - global_seqlen/balanced_min:22384 - global_seqlen/balanced_max:22391 - global_seqlen/mean:22389.375 - actor/entropy:0.5504021048545837 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2723144292831421 - training/rollout_probs_diff_mean:0.005151774734258652 - training/rollout_probs_diff_std:0.01037499401718378 - training/rollout_actor_probs_pearson_corr:0.9993544816970825 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.8984375 - self_distillation/correct_sample_fraction:0.7421875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8984375 - rollout_corr/rollout_is_mean:0.9999526739120483 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6774260997772217 - rollout_corr/rollout_is_min:0.6362783908843994 - rollout_corr/rollout_is_std:0.037128645926713943 - rollout_corr/rollout_is_eff_sample_size:0.9986231237591031 - rollout_corr/rollout_is_seq_mean:0.9999860525131226 - rollout_corr/rollout_is_seq_std:0.002109500812366605 - rollout_corr/rollout_is_seq_max:1.0075759887695312 - rollout_corr/rollout_is_seq_min:0.993508517742157 - rollout_corr/rollout_is_seq_max_deviation:0.00757598876953125 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7856992213055491) - rollout_corr/training_log_ppl:np.float64(0.5679072652128525) - rollout_corr/kl:np.float64(0.0009176925997458341) - rollout_corr/k3_kl:np.float64(0.0007757482836723284) - rollout_corr/rollout_ppl:np.float64(1.784037644509226) - rollout_corr/rollout_log_ppl:np.float64(0.5669895681203343) - rollout_corr/log_ppl_diff:np.float64(0.0009176970925182104) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019323080778121948) - rollout_corr/log_ppl_diff_max:np.float64(0.007754087448120117) - rollout_corr/log_ppl_diff_min:np.float64(-0.00577545166015625) - rollout_corr/ppl_ratio:np.float64(1.00092074717395) - rollout_corr/chi2_token:np.float64(0.00127337034791708) - rollout_corr/chi2_seq:np.float64(0.6404874103609473) - actor/pg_loss:np.float64(0.0034807646879926324) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002253228935842344) - actor/ppo_kl:np.float64(0.00018203446227005315) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8984375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8984375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.7421875) - self_distillation/token_reweight_w_mean:np.float64(130.1160973869264) - self_distillation/token_reweight_w_std:np.float64(2325.1732160223182) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0002648527733982) - self_distillation/token_reweight_w_clip_frac:np.float64(0.091672543407185) - self_distillation/empty_target_batch:np.float64(0.1015625) - actor/grad_norm:np.float64(0.2861635349690914) - perf/mfu/actor:np.float64(0.053185240362257795) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.9845733642578) - actor/lr:np.float64(1e-06) - training/global_step:77 - training/epoch:1 - critic/score/mean:0.7421875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7421875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0007428093231283128 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0007428093231283128 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:382.57421875 - response_length/max:872.0 - response_length/mi
|
||||
step:78 - global_seqlen/min:20134 - global_seqlen/max:31492 - global_seqlen/minmax_diff:11358 - global_seqlen/balanced_min:22482 - global_seqlen/balanced_max:29618 - global_seqlen/mean:23385.375 - actor/entropy:0.48590248823165894 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.40464895963668823 - training/rollout_probs_diff_mean:0.004600349348038435 - training/rollout_probs_diff_std:0.009926710277795792 - training/rollout_actor_probs_pearson_corr:0.9993968605995178 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.89453125 - self_distillation/correct_sample_fraction:0.68359375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.89453125 - rollout_corr/rollout_is_mean:0.9998959302902222 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:8.656884347146843e-06 - rollout_corr/rollout_is_max:1.8092979192733765 - rollout_corr/rollout_is_min:0.48850229382514954 - rollout_corr/rollout_is_std:0.03525106608867645 - rollout_corr/rollout_is_eff_sample_size:0.9987587856840187 - rollout_corr/rollout_is_seq_mean:0.9998641014099121 - rollout_corr/rollout_is_seq_std:0.0020814789459109306 - rollout_corr/rollout_is_seq_max:1.0097109079360962 - rollout_corr/rollout_is_seq_min:0.9942805171012878 - rollout_corr/rollout_is_seq_max_deviation:0.009710907936096191 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7207122039981186) - rollout_corr/training_log_ppl:np.float64(0.5288956113836321) - rollout_corr/kl:np.float64(0.0008265634318824766) - rollout_corr/k3_kl:np.float64(0.0007540007430009155) - rollout_corr/rollout_ppl:np.float64(1.7192450258880854) - rollout_corr/rollout_log_ppl:np.float64(0.5280690497056639) - rollout_corr/log_ppl_diff:np.float64(0.0008265616779681295) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017743823991622776) - rollout_corr/log_ppl_diff_max:np.float64(0.009331107139587402) - rollout_corr/log_ppl_diff_min:np.float64(-0.004447638988494873) - rollout_corr/ppl_ratio:np.float64(1.0008291078265756) - rollout_corr/chi2_token:np.float64(0.0013703966978937387) - rollout_corr/chi2_seq:np.float64(0.8413228269200772) - actor/pg_loss:np.float64(0.004617663682438433) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00030698799764650175) - actor/ppo_kl:np.float64(-3.2745020313029727e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.89453125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.89453125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.68359375) - self_distillation/token_reweight_w_mean:np.float64(38.89375674771145) - self_distillation/token_reweight_w_std:np.float64(700.807301894878) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.001390076475218) - self_distillation/token_reweight_w_clip_frac:np.float64(0.09478579316055402) - self_distillation/empty_target_batch:np.float64(0.10546875) - actor/grad_norm:np.float64(0.28953084722161293) - perf/mfu/actor:np.float64(0.0530806510376341) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(197.00171661376953) - actor/lr:np.float64(1e-06) - training/global_step:78 - training/epoch:1 - critic/score/mean:0.68359375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.68359375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.025150412693619728 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.025150412693619728 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:451.23046875 - response
|
||||
step:79 - global_seqlen/min:20379 - global_seqlen/max:23902 - global_seqlen/minmax_diff:3523 - global_seqlen/balanced_min:22483 - global_seqlen/balanced_max:22828 - global_seqlen/mean:22531.875 - actor/entropy:0.47975319623947144 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.999380886554718 - training/rollout_probs_diff_mean:0.004774122033268213 - training/rollout_probs_diff_std:0.010841231793165207 - training/rollout_actor_probs_pearson_corr:0.9993086457252502 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - self_distillation/correct_sample_fraction:0.609375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83203125 - rollout_corr/rollout_is_mean:0.9999167919158936 - rollout_corr/rollout_is_ratio_fraction_high:8.872878424881492e-06 - rollout_corr/rollout_is_ratio_fraction_low:1.7745756849762984e-05 - rollout_corr/rollout_is_max:2.256157159805298 - rollout_corr/rollout_is_min:1.8565188497632334e-07 - rollout_corr/rollout_is_std:0.035090066492557526 - rollout_corr/rollout_is_eff_sample_size:0.9987699636823175 - rollout_corr/rollout_is_seq_mean:0.9999032020568848 - rollout_corr/rollout_is_seq_std:0.002013471210375428 - rollout_corr/rollout_is_seq_max:1.0075592994689941 - rollout_corr/rollout_is_seq_min:0.9934999942779541 - rollout_corr/rollout_is_seq_max_deviation:0.007559299468994141 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6734817675314844) - rollout_corr/training_log_ppl:np.float64(0.5028578019700944) - rollout_corr/kl:np.float64(0.0007867071553491201) - rollout_corr/k3_kl:np.float64(0.0008091878738127889) - rollout_corr/rollout_ppl:np.float64(1.6721355388872325) - rollout_corr/rollout_log_ppl:np.float64(0.5020710933895316) - rollout_corr/log_ppl_diff:np.float64(0.0007867085805628449) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017737957823555917) - rollout_corr/log_ppl_diff_max:np.float64(0.022169530391693115) - rollout_corr/log_ppl_diff_min:np.float64(-0.006566405296325684) - rollout_corr/ppl_ratio:np.float64(1.000790162710473) - rollout_corr/chi2_token:np.float64(0.0015391185879707336) - rollout_corr/chi2_seq:np.float64(1.4496739441528916) - actor/pg_loss:np.float64(0.005306927720084786) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00039829218349041184) - actor/ppo_kl:np.float64(-4.736486153222863e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.83203125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.83203125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.609375) - self_distillation/token_reweight_w_mean:np.float64(76.99012057809159) - self_distillation/token_reweight_w_std:np.float64(1541.8618622039212) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9995152873452753) - self_distillation/token_reweight_w_clip_frac:np.float64(0.11978079983964562) - self_distillation/empty_target_batch:np.float64(0.16796875) - actor/grad_norm:np.float64(0.2994599714875221) - perf/mfu/actor:np.float64(0.052974969811297) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(197.06073379516602) - actor/lr:np.float64(1e-06) - training/global_step:79 - training/epoch:1 - critic/score/mean:0.609375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.609375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.009753511287271976 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.009753511287271976 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:440.24609375 - r
|
||||
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 80}
|
||||
validation generation end
|
||||
[prompt] system
|
||||
|
||||
Given a question and four options, please select the right answer. Respond in the following format:
|
||||
<reasoning>
|
||||
...
|
||||
</reasoning>
|
||||
<answer>
|
||||
...
|
||||
</answer>
|
||||
|
||||
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
|
||||
user
|
||||
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||||
|
||||
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||||
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||||
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||||
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||||
Please reason step by step.
|
||||
assistant
|
||||
<think>
|
||||
|
||||
</think>
|
||||
|
||||
|
||||
[response] <reasoning>
|
||||
To determine the correct reactant for the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1", we must analyze the structure and functional groups of the target molecule and compare it to the structures of the given options.
|
||||
|
||||
The target molecule contains:
|
||||
- A nitrogen-containing ring (likely an azole or similar).
|
||||
- A chlorinated aromatic ring (–Cl).
|
||||
- A trifluoromethyl group (–C(F)(F)F).
|
||||
- A nitrogen-containing side chain that is connected to a carbon chain with a chlorinated aromatic group and a trifluoromethyl group.
|
||||
|
||||
Now, we analyze the options:
|
||||
|
||||
**Option A:**
|
||||
"O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1"
|
||||
- Contains a chlorinated aromatic ring (–Cl).
|
||||
- Contains a trifluoromethyl group (–C(F)(F)F).
|
||||
- Contains a nitrogen-containing ring with a side chain that connects to the chlorinated aromatic ring and the trifluoromethyl group.
|
||||
- Matches the structure of the target molecule closely.
|
||||
|
||||
**Option B:**
|
||||
"O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1"
|
||||
- Contains a chlorinated ring, but the side chain and functional groups do not match the target molecule.
|
||||
- The nitrogen-containing ring is different and does not include the trifluoromethyl group.
|
||||
|
||||
**Option C:**
|
||||
"O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1"
|
||||
- Contains a chlorinated aromatic ring and a trifluoromethyl group.
|
||||
- However, the functional group arrangement and the nitrogen-containing ring do not match the target molecule's structure.
|
||||
|
||||
**Option D:**
|
||||
"O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F"
|
||||
- Contains a trifluoromethyl group and a chlorinated ring.
|
||||
- However, the nitrogen-containing ring and the side chain are not structured correctly to match the target molecule.
|
||||
|
||||
**Conclusion:**
|
||||
Only **Option A** contains all the necessary functional groups and structural elements that match the target molecule "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1".
|
||||
|
||||
</reasoning>
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
[ground_truth] A
|
||||
[score] 1.0
|
||||
[acc] 1.0
|
||||
[pred] A
|
||||
[incorrect_format] 1
|
||||
[feedback]
|
||||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_80
|
||||
step:80 - global_seqlen/min:21379 - global_seqlen/max:27736 - global_seqlen/minmax_diff:6357 - global_seqlen/balanced_min:24041 - global_seqlen/balanced_max:24046 - global_seqlen/mean:24044.0 - actor/entropy:0.5321111679077148 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.22554826736450195 - training/rollout_probs_diff_mean:0.005127668380737305 - training/rollout_probs_diff_std:0.010481650941073895 - training/rollout_actor_probs_pearson_corr:0.999333381652832 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.90234375 - self_distillation/correct_sample_fraction:0.7109375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.90234375 - rollout_corr/rollout_is_mean:0.9999201893806458 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.784020185470581 - rollout_corr/rollout_is_min:0.5773587822914124 - rollout_corr/rollout_is_std:0.03691612929105759 - rollout_corr/rollout_is_eff_sample_size:0.9986386974447631 - rollout_corr/rollout_is_seq_mean:0.9999967813491821 - rollout_corr/rollout_is_seq_std:0.002032036194577813 - rollout_corr/rollout_is_seq_max:1.0090333223342896 - rollout_corr/rollout_is_seq_min:0.99375981092453 - rollout_corr/rollout_is_seq_max_deviation:0.00903332233428955 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7447218438610435) - rollout_corr/training_log_ppl:np.float64(0.5451616903301328) - rollout_corr/kl:np.float64(0.0007389033336671069) - rollout_corr/k3_kl:np.float64(0.0007813431365661927) - rollout_corr/rollout_ppl:np.float64(1.743419658858329) - rollout_corr/rollout_log_ppl:np.float64(0.5444227863335982) - rollout_corr/log_ppl_diff:np.float64(0.0007389039965346456) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016596049536019564) - rollout_corr/log_ppl_diff_max:np.float64(0.005907297134399414) - rollout_corr/log_ppl_diff_min:np.float64(-0.0042874813079833984) - rollout_corr/ppl_ratio:np.float64(1.0007411232218146) - rollout_corr/chi2_token:np.float64(0.001654008636251092) - rollout_corr/chi2_seq:np.float64(0.4709114315919578) - actor/pg_loss:np.float64(0.004708616062998772) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004071622186074819) - actor/ppo_kl:np.float64(2.530917598608795e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.90234375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.90234375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.7109375) - self_distillation/token_reweight_w_mean:np.float64(13418.10702541261) - self_distillation/token_reweight_w_std:np.float64(249965.62142065133) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0001714508980513) - self_distillation/token_reweight_w_clip_frac:np.float64(0.10478754635551013) - self_distillation/empty_target_batch:np.float64(0.09765625) - actor/grad_norm:np.float64(0.277141023427248) - perf/mfu/actor:np.float64(0.05686537883473178) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(197.12682485580444) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7098214285714286) - val-aux/sciknoweval/reward/std@16:np.float64(0.14355997006676555) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7664952380952381) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.10898051269342317) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6531333333333332) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.13319323391175897) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.710485714285
|
||||
step:81 - global_seqlen/min:19063 - global_seqlen/max:27921 - global_seqlen/minmax_diff:8858 - global_seqlen/balanced_min:22607 - global_seqlen/balanced_max:22618 - global_seqlen/mean:22615.0 - actor/entropy:0.4837610721588135 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18462881445884705 - training/rollout_probs_diff_mean:0.004895538557320833 - training/rollout_probs_diff_std:0.010500817559659481 - training/rollout_actor_probs_pearson_corr:0.9993140697479248 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.7734375 - self_distillation/correct_sample_fraction:0.63671875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7734375 - rollout_corr/rollout_is_mean:0.9999274611473083 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:9.790866897674277e-06 - rollout_corr/rollout_is_max:1.9132018089294434 - rollout_corr/rollout_is_min:0.4658448398113251 - rollout_corr/rollout_is_std:0.036682888865470886 - rollout_corr/rollout_is_eff_sample_size:0.998655936060176 - rollout_corr/rollout_is_seq_mean:0.9999848008155823 - rollout_corr/rollout_is_seq_std:0.002200948540121317 - rollout_corr/rollout_is_seq_max:1.0080451965332031 - rollout_corr/rollout_is_seq_min:0.9924934506416321 - rollout_corr/rollout_is_seq_max_deviation:0.008045196533203125 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6707927286624908) - rollout_corr/training_log_ppl:np.float64(0.496672905923333) - rollout_corr/kl:np.float64(0.0007028676274636325) - rollout_corr/k3_kl:np.float64(0.0008774436023486487) - rollout_corr/rollout_ppl:np.float64(1.6695120050571859) - rollout_corr/rollout_log_ppl:np.float64(0.4959700393374078) - rollout_corr/log_ppl_diff:np.float64(0.0007028665859252214) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018403253052383661) - rollout_corr/log_ppl_diff_max:np.float64(0.008645951747894287) - rollout_corr/log_ppl_diff_min:np.float64(-0.005466446280479431) - rollout_corr/ppl_ratio:np.float64(1.0007056852336973) - rollout_corr/chi2_token:np.float64(0.002068758476525545) - rollout_corr/chi2_seq:np.float64(0.5465720517095178) - actor/pg_loss:np.float64(0.0029664566973224282) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00035411453472988796) - actor/ppo_kl:np.float64(-4.833459730235745e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.7734375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.7734375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.63671875) - self_distillation/token_reweight_w_mean:np.float64(46.35069849877618) - self_distillation/token_reweight_w_std:np.float64(809.1466785562807) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0000212090089917) - self_distillation/token_reweight_w_clip_frac:np.float64(0.07078566148993559) - self_distillation/empty_target_batch:np.float64(0.2265625) - actor/grad_norm:np.float64(0.2487758994102478) - perf/mfu/actor:np.float64(0.0539087676600922) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(200.76457595825195) - actor/lr:np.float64(1e-06) - training/global_step:81 - training/epoch:1 - critic/score/mean:0.63671875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.63671875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.008355281315743923 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.008355281315743923 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:398.96875 - response_length/max:113
|
||||
step:82 - global_seqlen/min:20637 - global_seqlen/max:24388 - global_seqlen/minmax_diff:3751 - global_seqlen/balanced_min:22213 - global_seqlen/balanced_max:22222 - global_seqlen/mean:22220.125 - actor/entropy:0.44993990659713745 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24135524034500122 - training/rollout_probs_diff_mean:0.0047300467267632484 - training/rollout_probs_diff_std:0.010585080832242966 - training/rollout_actor_probs_pearson_corr:0.9992892146110535 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.828125 - self_distillation/correct_sample_fraction:0.6015625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.828125 - rollout_corr/rollout_is_mean:0.9999090433120728 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.8240019926452078e-05 - rollout_corr/rollout_is_max:1.7895359992980957 - rollout_corr/rollout_is_min:0.000389088352676481 - rollout_corr/rollout_is_std:0.03580636531114578 - rollout_corr/rollout_is_eff_sample_size:0.9987193079476842 - rollout_corr/rollout_is_seq_mean:0.9998470544815063 - rollout_corr/rollout_is_seq_std:0.0019274608930572867 - rollout_corr/rollout_is_seq_max:1.0058091878890991 - rollout_corr/rollout_is_seq_min:0.9948349595069885 - rollout_corr/rollout_is_seq_max_deviation:0.005809187889099121 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6126991645433009) - rollout_corr/training_log_ppl:np.float64(0.4650081274448894) - rollout_corr/kl:np.float64(0.0009749815301454134) - rollout_corr/k3_kl:np.float64(0.0008126581537908351) - rollout_corr/rollout_ppl:np.float64(1.6110614440403879) - rollout_corr/rollout_log_ppl:np.float64(0.46403314505005255) - rollout_corr/log_ppl_diff:np.float64(0.000974982394836843) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017385364044457674) - rollout_corr/log_ppl_diff_max:np.float64(0.01138317584991455) - rollout_corr/log_ppl_diff_min:np.float64(-0.004374325275421143) - rollout_corr/ppl_ratio:np.float64(1.0009776803199202) - rollout_corr/chi2_token:np.float64(0.0012213955633342266) - rollout_corr/chi2_seq:np.float64(0.3682185474317521) - actor/pg_loss:np.float64(0.0047809198731556535) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00044549755830303184) - actor/ppo_kl:np.float64(9.73689185599369e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.828125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.828125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6015625) - self_distillation/token_reweight_w_mean:np.float64(8763.06257886719) - self_distillation/token_reweight_w_std:np.float64(183308.58723631455) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9998282515443861) - self_distillation/token_reweight_w_clip_frac:np.float64(0.10694929157034494) - self_distillation/empty_target_batch:np.float64(0.171875) - actor/grad_norm:np.float64(0.27595510706305504) - perf/mfu/actor:np.float64(0.053128164780379857) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(200.69163131713867) - actor/lr:np.float64(1e-06) - training/global_step:82 - training/epoch:1 - critic/score/mean:0.6015625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6015625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0036035440862178802 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0036035440862178802 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:428.31640625 - response_length/max
|
||||
step:83 - global_seqlen/min:17994 - global_seqlen/max:23375 - global_seqlen/minmax_diff:5381 - global_seqlen/balanced_min:21160 - global_seqlen/balanced_max:21170 - global_seqlen/mean:21165.25 - actor/entropy:0.4580913484096527 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1788254976272583 - training/rollout_probs_diff_mean:0.005244065076112747 - training/rollout_probs_diff_std:0.011341297999024391 - training/rollout_actor_probs_pearson_corr:0.9992040395736694 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.7578125 - self_distillation/correct_sample_fraction:0.49609375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7578125 - rollout_corr/rollout_is_mean:1.0000450611114502 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6948878765106201 - rollout_corr/rollout_is_min:0.5044571757316589 - rollout_corr/rollout_is_std:0.03826860338449478 - rollout_corr/rollout_is_eff_sample_size:0.9985378931829969 - rollout_corr/rollout_is_seq_mean:0.9999917149543762 - rollout_corr/rollout_is_seq_std:0.0024205187801271677 - rollout_corr/rollout_is_seq_max:1.007947325706482 - rollout_corr/rollout_is_seq_min:0.9913325309753418 - rollout_corr/rollout_is_seq_max_deviation:0.008667469024658203 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6229573977179825) - rollout_corr/training_log_ppl:np.float64(0.4732308021048084) - rollout_corr/kl:np.float64(0.0009288024647275961) - rollout_corr/k3_kl:np.float64(0.0008689218968811474) - rollout_corr/rollout_ppl:np.float64(1.6214703912846744) - rollout_corr/rollout_log_ppl:np.float64(0.47230199666228145) - rollout_corr/log_ppl_diff:np.float64(0.0009288054425269365) - rollout_corr/log_ppl_abs_diff:np.float64(0.002043650601990521) - rollout_corr/log_ppl_diff_max:np.float64(0.01018601655960083) - rollout_corr/log_ppl_diff_min:np.float64(-0.0065596699714660645) - rollout_corr/ppl_ratio:np.float64(1.0009324771817774) - rollout_corr/chi2_token:np.float64(0.0016192824114114046) - rollout_corr/chi2_seq:np.float64(0.7770539077464491) - actor/pg_loss:np.float64(0.005605395417660475) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005950954305262712) - actor/ppo_kl:np.float64(0.00014756420982209306) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.7578125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.7578125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.49609375) - self_distillation/token_reweight_w_mean:np.float64(43651.46375838155) - self_distillation/token_reweight_w_std:np.float64(663337.3462570815) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.001920053968206) - self_distillation/token_reweight_w_clip_frac:np.float64(0.12145704132854007) - self_distillation/empty_target_batch:np.float64(0.2421875) - actor/grad_norm:np.float64(0.37171755731105804) - perf/mfu/actor:np.float64(0.051307628909952094) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(200.40789794921875) - actor/lr:np.float64(1e-06) - training/global_step:83 - training/epoch:1 - critic/score/mean:0.49609375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.49609375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0006963750929571688 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0006963750929571688 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:362.5078125 - response_length/max:903.0 - respons
|
||||
step:84 - global_seqlen/min:16326 - global_seqlen/max:23634 - global_seqlen/minmax_diff:7308 - global_seqlen/balanced_min:19956 - global_seqlen/balanced_max:19966 - global_seqlen/mean:19962.0 - actor/entropy:0.4355461001396179 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23105639219284058 - training/rollout_probs_diff_mean:0.0052682701498270035 - training/rollout_probs_diff_std:0.011572654359042645 - training/rollout_actor_probs_pearson_corr:0.9991452097892761 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.8671875 - self_distillation/correct_sample_fraction:0.6875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8671875 - rollout_corr/rollout_is_mean:0.9999979138374329 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.7088184356689453 - rollout_corr/rollout_is_min:0.6109715700149536 - rollout_corr/rollout_is_std:0.038198452442884445 - rollout_corr/rollout_is_eff_sample_size:0.9985430042300492 - rollout_corr/rollout_is_seq_mean:0.999947190284729 - rollout_corr/rollout_is_seq_std:0.0022522625513374805 - rollout_corr/rollout_is_seq_max:1.0068590641021729 - rollout_corr/rollout_is_seq_min:0.9928377866744995 - rollout_corr/rollout_is_seq_max_deviation:0.007162213325500488 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5662654708139598) - rollout_corr/training_log_ppl:np.float64(0.4385948544368148) - rollout_corr/kl:np.float64(0.0010095660052041877) - rollout_corr/k3_kl:np.float64(0.0008661396312845682) - rollout_corr/rollout_ppl:np.float64(1.5646833986975253) - rollout_corr/rollout_log_ppl:np.float64(0.4375852865341585) - rollout_corr/log_ppl_diff:np.float64(0.001009567902656272) - rollout_corr/log_ppl_abs_diff:np.float64(0.002128819382051006) - rollout_corr/log_ppl_diff_max:np.float64(0.010426610708236694) - rollout_corr/log_ppl_diff_min:np.float64(-0.0056244730949401855) - rollout_corr/ppl_ratio:np.float64(1.0010133925825357) - rollout_corr/chi2_token:np.float64(0.0014718861784785986) - rollout_corr/chi2_seq:np.float64(0.8280399544164538) - actor/pg_loss:np.float64(0.005005670711398125) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003053363407161669) - actor/ppo_kl:np.float64(0.00018772196818872544) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8671875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8671875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6875) - self_distillation/token_reweight_w_mean:np.float64(1562.93544895784) - self_distillation/token_reweight_w_std:np.float64(29233.69878470525) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9984272990841419) - self_distillation/token_reweight_w_clip_frac:np.float64(0.09582373814191669) - self_distillation/empty_target_batch:np.float64(0.1328125) - actor/grad_norm:np.float64(0.27201734483242035) - perf/mfu/actor:np.float64(0.04843784064306516) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(200.41874313354492) - actor/lr:np.float64(1e-06) - training/global_step:84 - training/epoch:1 - critic/score/mean:0.6875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0011461590183898807 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0011461590183898807 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:369.78125 - response_length/max:898.0 - response_length/min:147.0 -
|
||||
step:85 - global_seqlen/min:18150 - global_seqlen/max:26353 - global_seqlen/minmax_diff:8203 - global_seqlen/balanced_min:22089 - global_seqlen/balanced_max:22097 - global_seqlen/mean:22095.125 - actor/entropy:0.5015257596969604 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.22547149658203125 - training/rollout_probs_diff_mean:0.005411901045590639 - training/rollout_probs_diff_std:0.011227205395698547 - training/rollout_actor_probs_pearson_corr:0.9992418885231018 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.75 - self_distillation/correct_sample_fraction:0.71484375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.75 - rollout_corr/rollout_is_mean:1.0000245571136475 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.9998200514237396e-05 - rollout_corr/rollout_is_max:1.8690942525863647 - rollout_corr/rollout_is_min:0.3950064182281494 - rollout_corr/rollout_is_std:0.038822196424007416 - rollout_corr/rollout_is_eff_sample_size:0.9984952239250111 - rollout_corr/rollout_is_seq_mean:1.0000362396240234 - rollout_corr/rollout_is_seq_std:0.0021999897435307503 - rollout_corr/rollout_is_seq_max:1.0113379955291748 - rollout_corr/rollout_is_seq_min:0.9936926364898682 - rollout_corr/rollout_is_seq_max_deviation:0.011337995529174805 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6718262596987188) - rollout_corr/training_log_ppl:np.float64(0.5036372784525156) - rollout_corr/kl:np.float64(0.0008671445817096668) - rollout_corr/k3_kl:np.float64(0.0008546028941509576) - rollout_corr/rollout_ppl:np.float64(1.670389695558697) - rollout_corr/rollout_log_ppl:np.float64(0.5027701300568879) - rollout_corr/log_ppl_diff:np.float64(0.000867148395627737) - rollout_corr/log_ppl_abs_diff:np.float64(0.00184359191916883) - rollout_corr/log_ppl_diff_max:np.float64(0.007470607757568359) - rollout_corr/log_ppl_diff_min:np.float64(-0.005036354064941406) - rollout_corr/ppl_ratio:np.float64(1.000870036194101) - rollout_corr/chi2_token:np.float64(0.001706940121948719) - rollout_corr/chi2_seq:np.float64(0.803199248155579) - actor/pg_loss:np.float64(0.0017579543637111783) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(4.799221642315388e-05) - actor/ppo_kl:np.float64(0.00011437771347289072) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.75) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.75) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.71484375) - self_distillation/token_reweight_w_mean:np.float64(531.5800662350375) - self_distillation/token_reweight_w_std:np.float64(6761.905646110186) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.998828696552664) - self_distillation/token_reweight_w_clip_frac:np.float64(0.02915754570858553) - self_distillation/empty_target_batch:np.float64(0.25) - actor/grad_norm:np.float64(0.16114498674869537) - perf/mfu/actor:np.float64(0.052797853671240076) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(200.15685653686523) - actor/lr:np.float64(1e-06) - training/global_step:85 - training/epoch:1 - critic/score/mean:0.71484375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.71484375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0009824115550145507 - critic/advantages/max:0.5 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0009824115550145507 - critic/returns/max:0.5 - critic/returns/min:-0.875 - response_length/mean:390.66015625 - response_length/max:984.0 - response_length/min:1
|
||||
step:86 - global_seqlen/min:19172 - global_seqlen/max:23707 - global_seqlen/minmax_diff:4535 - global_seqlen/balanced_min:21178 - global_seqlen/balanced_max:21185 - global_seqlen/mean:21182.625 - actor/entropy:0.4417911767959595 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2772989869117737 - training/rollout_probs_diff_mean:0.004951337352395058 - training/rollout_probs_diff_std:0.011015241034328938 - training/rollout_actor_probs_pearson_corr:0.999230682849884 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80078125 - self_distillation/correct_sample_fraction:0.609375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.80078125 - rollout_corr/rollout_is_mean:1.0001946687698364 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.0157130418519955e-05 - rollout_corr/rollout_is_max:1.913230538368225 - rollout_corr/rollout_is_min:0.48568272590637207 - rollout_corr/rollout_is_std:0.03722003847360611 - rollout_corr/rollout_is_eff_sample_size:0.9986170608525367 - rollout_corr/rollout_is_seq_mean:1.000307321548462 - rollout_corr/rollout_is_seq_std:0.0024071792140603065 - rollout_corr/rollout_is_seq_max:1.0088176727294922 - rollout_corr/rollout_is_seq_min:0.9944450259208679 - rollout_corr/rollout_is_seq_max_deviation:0.008817672729492188 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.608684377744794) - rollout_corr/training_log_ppl:np.float64(0.4590825633786153) - rollout_corr/kl:np.float64(0.0005738776003347823) - rollout_corr/k3_kl:np.float64(0.0008272058389025005) - rollout_corr/rollout_ppl:np.float64(1.6077406979165971) - rollout_corr/rollout_log_ppl:np.float64(0.45850868528941646) - rollout_corr/log_ppl_diff:np.float64(0.0005738780891988426) - rollout_corr/log_ppl_abs_diff:np.float64(0.0020339481707196683) - rollout_corr/log_ppl_diff_max:np.float64(0.008264422416687012) - rollout_corr/log_ppl_diff_min:np.float64(-0.009043440222740173) - rollout_corr/ppl_ratio:np.float64(1.00057721324265) - rollout_corr/chi2_token:np.float64(0.0023612575605511665) - rollout_corr/chi2_seq:np.float64(1.4658026506658643) - actor/pg_loss:np.float64(0.00446285936050117) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004066334154231299) - actor/ppo_kl:np.float64(0.00014794760728875644) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.80078125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.80078125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.609375) - self_distillation/token_reweight_w_mean:np.float64(132.06133578903973) - self_distillation/token_reweight_w_std:np.float64(2865.135968480725) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9988646688871086) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1047386541031301) - self_distillation/empty_target_batch:np.float64(0.19921875) - actor/grad_norm:np.float64(0.297544926404953) - perf/mfu/actor:np.float64(0.05125659080467514) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(200.21656799316406) - actor/lr:np.float64(1e-06) - training/global_step:86 - training/epoch:1 - critic/score/mean:0.609375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.609375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.011623566970229149 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.011623566970229149 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:384.58203125 - response_length/max:958.0
|
||||
step:87 - global_seqlen/min:19881 - global_seqlen/max:22805 - global_seqlen/minmax_diff:2924 - global_seqlen/balanced_min:21579 - global_seqlen/balanced_max:21581 - global_seqlen/mean:21580.5 - actor/entropy:0.4654737114906311 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3751887083053589 - training/rollout_probs_diff_mean:0.0050046672113239765 - training/rollout_probs_diff_std:0.010888464748859406 - training/rollout_actor_probs_pearson_corr:0.9992407560348511 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.86328125 - self_distillation/correct_sample_fraction:0.62109375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.86328125 - rollout_corr/rollout_is_mean:1.0000450611114502 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.0308853234164417e-05 - rollout_corr/rollout_is_max:1.854941487312317 - rollout_corr/rollout_is_min:0.3242693543434143 - rollout_corr/rollout_is_std:0.03734952583909035 - rollout_corr/rollout_is_eff_sample_size:0.998607075049263 - rollout_corr/rollout_is_seq_mean:1.0000066757202148 - rollout_corr/rollout_is_seq_std:0.00233814911916852 - rollout_corr/rollout_is_seq_max:1.0092549324035645 - rollout_corr/rollout_is_seq_min:0.9934799671173096 - rollout_corr/rollout_is_seq_max_deviation:0.009254932403564453 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6343936491757631) - rollout_corr/training_log_ppl:np.float64(0.47929042315809056) - rollout_corr/kl:np.float64(0.0010372451878630784) - rollout_corr/k3_kl:np.float64(0.00085206899655077) - rollout_corr/rollout_ppl:np.float64(1.6326408791355789) - rollout_corr/rollout_log_ppl:np.float64(0.4782531792880036) - rollout_corr/log_ppl_diff:np.float64(0.001037243870086968) - rollout_corr/log_ppl_abs_diff:np.float64(0.002030018251389265) - rollout_corr/log_ppl_diff_max:np.float64(0.011031702160835266) - rollout_corr/log_ppl_diff_min:np.float64(-0.006544649600982666) - rollout_corr/ppl_ratio:np.float64(1.0010407408699393) - rollout_corr/chi2_token:np.float64(0.0013177578803151846) - rollout_corr/chi2_seq:np.float64(0.7552365991286933) - actor/pg_loss:np.float64(0.006197207723744214) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.000917987231787265) - actor/ppo_kl:np.float64(0.00031247709554804715) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.86328125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.86328125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.62109375) - self_distillation/token_reweight_w_mean:np.float64(8798.193575591547) - self_distillation/token_reweight_w_std:np.float64(137248.53065139498) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0011895855423063) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1116473093861714) - self_distillation/empty_target_batch:np.float64(0.13671875) - actor/grad_norm:np.float64(0.3543427586555481) - perf/mfu/actor:np.float64(0.052530135237403436) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(200.04389572143555) - actor/lr:np.float64(1e-06) - training/global_step:87 - training/epoch:1 - critic/score/mean:0.62109375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.62109375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.004306523595005274 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.004306523595005274 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:378.921875 - response_length/max:813.0
|
||||
step:88 - global_seqlen/min:18137 - global_seqlen/max:23978 - global_seqlen/minmax_diff:5841 - global_seqlen/balanced_min:21618 - global_seqlen/balanced_max:21626 - global_seqlen/mean:21623.75 - actor/entropy:0.4558359980583191 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24135497212409973 - training/rollout_probs_diff_mean:0.005039794836193323 - training/rollout_probs_diff_std:0.01103699766099453 - training/rollout_actor_probs_pearson_corr:0.9992441534996033 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.84765625 - self_distillation/correct_sample_fraction:0.58984375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.84765625 - rollout_corr/rollout_is_mean:0.9999773502349854 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.879797284549568e-05 - rollout_corr/rollout_is_max:1.6429718732833862 - rollout_corr/rollout_is_min:0.34945860505104065 - rollout_corr/rollout_is_std:0.03734952583909035 - rollout_corr/rollout_is_eff_sample_size:0.9986068372944764 - rollout_corr/rollout_is_seq_mean:0.9999490976333618 - rollout_corr/rollout_is_seq_std:0.00212614587508142 - rollout_corr/rollout_is_seq_max:1.007338047027588 - rollout_corr/rollout_is_seq_min:0.9925395846366882 - rollout_corr/rollout_is_seq_max_deviation:0.007460415363311768 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6032785079441965) - rollout_corr/training_log_ppl:np.float64(0.46080948115559295) - rollout_corr/kl:np.float64(0.0011248015301781322) - rollout_corr/k3_kl:np.float64(0.0008873714502897201) - rollout_corr/rollout_ppl:np.float64(1.601451831869781) - rollout_corr/rollout_log_ppl:np.float64(0.4596846750937402) - rollout_corr/log_ppl_diff:np.float64(0.0011248060618527234) - rollout_corr/log_ppl_abs_diff:np.float64(0.002022003580350429) - rollout_corr/log_ppl_diff_max:np.float64(0.012501776218414307) - rollout_corr/log_ppl_diff_min:np.float64(-0.007802844047546387) - rollout_corr/ppl_ratio:np.float64(1.0011287301313132) - rollout_corr/chi2_token:np.float64(0.0012870903592556715) - rollout_corr/chi2_seq:np.float64(0.6421570053789765) - actor/pg_loss:np.float64(0.004205821780487895) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004930502600473119) - actor/ppo_kl:np.float64(0.00033077471513121015) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.84765625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.84765625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.58984375) - self_distillation/token_reweight_w_mean:np.float64(21107.031607169192) - self_distillation/token_reweight_w_std:np.float64(309542.6647549386) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0036525861360133) - self_distillation/token_reweight_w_clip_frac:np.float64(0.08997728512622416) - self_distillation/empty_target_batch:np.float64(0.15234375) - actor/grad_norm:np.float64(0.29456714540719986) - perf/mfu/actor:np.float64(0.05240681242673272) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(200.06198120117188) - actor/lr:np.float64(1e-06) - training/global_step:88 - training/epoch:1 - critic/score/mean:0.58984375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.58984375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0007283487357199192 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0007283487357199192 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:406.9296875 - response_length/m
|
||||
step:89 - global_seqlen/min:20192 - global_seqlen/max:26260 - global_seqlen/minmax_diff:6068 - global_seqlen/balanced_min:23178 - global_seqlen/balanced_max:23186 - global_seqlen/mean:23183.75 - actor/entropy:0.4386472702026367 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7019415497779846 - training/rollout_probs_diff_mean:0.0047370451502501965 - training/rollout_probs_diff_std:0.010689225047826767 - training/rollout_actor_probs_pearson_corr:0.999261736869812 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.86328125 - self_distillation/correct_sample_fraction:0.59765625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.86328125 - rollout_corr/rollout_is_mean:0.9999846816062927 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.8036542314803228e-05 - rollout_corr/rollout_is_max:1.5554907321929932 - rollout_corr/rollout_is_min:0.03433068096637726 - rollout_corr/rollout_is_std:0.0357397198677063 - rollout_corr/rollout_is_eff_sample_size:0.9987241830413971 - rollout_corr/rollout_is_seq_mean:0.9999765157699585 - rollout_corr/rollout_is_seq_std:0.0020081165712326765 - rollout_corr/rollout_is_seq_max:1.0105236768722534 - rollout_corr/rollout_is_seq_min:0.9940011501312256 - rollout_corr/rollout_is_seq_max_deviation:0.010523676872253418 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6203870358876884) - rollout_corr/training_log_ppl:np.float64(0.47031728448928334) - rollout_corr/kl:np.float64(0.0010279387461658018) - rollout_corr/k3_kl:np.float64(0.0008464122184363987) - rollout_corr/rollout_ppl:np.float64(1.6186228357255459) - rollout_corr/rollout_log_ppl:np.float64(0.4692893385363277) - rollout_corr/log_ppl_diff:np.float64(0.0010279459529556334) - rollout_corr/log_ppl_abs_diff:np.float64(0.00189788214629516) - rollout_corr/log_ppl_diff_max:np.float64(0.0077957212924957275) - rollout_corr/log_ppl_diff_min:np.float64(-0.004509687423706055) - rollout_corr/ppl_ratio:np.float64(1.0010310297366232) - rollout_corr/chi2_token:np.float64(0.0013277053367346525) - rollout_corr/chi2_seq:np.float64(0.5938672970514745) - actor/pg_loss:np.float64(0.0077381625305861235) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0008514890976130118) - actor/ppo_kl:np.float64(0.0002891868470378256) - actor/pg_clipfrac_lower:np.float64(8.172070920409169e-06) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.86328125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.86328125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.59765625) - self_distillation/token_reweight_w_mean:np.float64(47.27642107871361) - self_distillation/token_reweight_w_std:np.float64(801.2714838348329) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000362384133041) - self_distillation/token_reweight_w_clip_frac:np.float64(0.13113477459410205) - self_distillation/empty_target_batch:np.float64(0.13671875) - actor/grad_norm:np.float64(0.3602307364344597) - perf/mfu/actor:np.float64(0.05579510012189607) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(197.08685684204102) - actor/lr:np.float64(1e-06) - training/global_step:89 - training/epoch:1 - critic/score/mean:0.59765625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.59765625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0041562956757843494 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0041562956757843494 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:433.148437
|
||||
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 90}
|
||||
validation generation end
|
||||
[prompt] system
|
||||
|
||||
Given a question and four options, please select the right answer. Respond in the following format:
|
||||
<reasoning>
|
||||
...
|
||||
</reasoning>
|
||||
<answer>
|
||||
...
|
||||
</answer>
|
||||
|
||||
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
|
||||
user
|
||||
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||||
|
||||
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||||
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||||
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||||
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||||
Please reason step by step.
|
||||
assistant
|
||||
<think>
|
||||
|
||||
</think>
|
||||
|
||||
|
||||
[response] <reasoning>
|
||||
To determine the correct reactant for the synthesis of the product **"Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1"**, we need to analyze the structure of the product and compare it with the options provided.
|
||||
|
||||
The product has the following key features:
|
||||
- A nitrogen-containing ring (likely a pyrrole or similar structure).
|
||||
- A **chlorine-substituted benzene ring** (from the **Cl** in **"ccc(Cl)"**).
|
||||
- A **trifluoromethyl group** (from **"C(F)(F)F"**).
|
||||
- A **nitrogen-containing bridge** connecting the ring to the rest of the molecule.
|
||||
|
||||
Now, let’s evaluate each option:
|
||||
|
||||
**Option A**:
|
||||
`O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1`
|
||||
- Contains a **chlorine-substituted benzene ring**.
|
||||
- Contains a **trifluoromethyl group**.
|
||||
- Contains a **nitrogen-containing bridge** with the correct connectivity.
|
||||
- This matches the structure of the product and appears to be a direct precursor.
|
||||
|
||||
**Option B**:
|
||||
`O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1`
|
||||
- Contains **chlorine atoms**, but the **trifluoromethyl group** is missing.
|
||||
- The **nitrogen-containing bridge** is not properly connected to the **trifluoromethyl** group.
|
||||
- This does not match the product structure.
|
||||
|
||||
**Option C**:
|
||||
`O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1`
|
||||
- Contains a **chlorine-substituted benzene ring** and a **trifluoromethyl group**.
|
||||
- However, the **nitrogen-containing bridge** is not properly positioned and is connected via an **oxygen**, which is not part of the product.
|
||||
- This is not a correct match.
|
||||
|
||||
**Option D**:
|
||||
`O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F`
|
||||
- Contains a **trifluoromethyl group**.
|
||||
- However, it **lacks the chlorine-substituted benzene ring** and the **correct nitrogen-containing bridge**.
|
||||
- This also does not match the product structure.
|
||||
|
||||
**Conclusion**:
|
||||
Only **Option A** contains all the necessary functional groups and the correct connectivity to serve as the only correct reactant in the synthesis of the given product.
|
||||
|
||||
</reasoning>
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
[ground_truth] A
|
||||
[score] 1.0
|
||||
[acc] 1.0
|
||||
[pred] A
|
||||
[incorrect_format] 1
|
||||
[feedback]
|
||||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_90
|
||||
step:90 - global_seqlen/min:18865 - global_seqlen/max:24932 - global_seqlen/minmax_diff:6067 - global_seqlen/balanced_min:22815 - global_seqlen/balanced_max:22831 - global_seqlen/mean:22826.625 - actor/entropy:0.40077415108680725 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5775288939476013 - training/rollout_probs_diff_mean:0.0045609320513904095 - training/rollout_probs_diff_std:0.010716044344007969 - training/rollout_actor_probs_pearson_corr:0.9992610216140747 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.85546875 - self_distillation/correct_sample_fraction:0.62109375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.85546875 - rollout_corr/rollout_is_mean:0.9999971985816956 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:8.970459930424113e-06 - rollout_corr/rollout_is_max:1.8681867122650146 - rollout_corr/rollout_is_min:0.2344389706850052 - rollout_corr/rollout_is_std:0.03548699617385864 - rollout_corr/rollout_is_eff_sample_size:0.9987421380652595 - rollout_corr/rollout_is_seq_mean:1.0000590085983276 - rollout_corr/rollout_is_seq_std:0.0020531753543764353 - rollout_corr/rollout_is_seq_max:1.0109810829162598 - rollout_corr/rollout_is_seq_min:0.9949836730957031 - rollout_corr/rollout_is_seq_max_deviation:0.010981082916259766 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5583982807584107) - rollout_corr/training_log_ppl:np.float64(0.43078241116018035) - rollout_corr/kl:np.float64(0.0008912494752149058) - rollout_corr/k3_kl:np.float64(0.000791456644748223) - rollout_corr/rollout_ppl:np.float64(1.5569715378805995) - rollout_corr/rollout_log_ppl:np.float64(0.42989116214448586) - rollout_corr/log_ppl_diff:np.float64(0.0008912490156944841) - rollout_corr/log_ppl_abs_diff:np.float64(0.001720420696074143) - rollout_corr/log_ppl_diff_max:np.float64(0.01338726282119751) - rollout_corr/log_ppl_diff_min:np.float64(-0.00902646780014038) - rollout_corr/ppl_ratio:np.float64(1.0008942331187427) - rollout_corr/chi2_token:np.float64(0.0014041787944734097) - rollout_corr/chi2_seq:np.float64(0.752974865725264) - actor/pg_loss:np.float64(0.004781530937179923) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005209917633237637) - actor/ppo_kl:np.float64(0.0002642208678480529) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.85546875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.85546875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.62109375) - self_distillation/token_reweight_w_mean:np.float64(619.1751045542769) - self_distillation/token_reweight_w_std:np.float64(7680.31531262747) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.999175232835114) - self_distillation/token_reweight_w_clip_frac:np.float64(0.1011948958330322) - self_distillation/empty_target_batch:np.float64(0.14453125) - actor/grad_norm:np.float64(0.260063748806715) - perf/mfu/actor:np.float64(0.05435728539450956) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(196.9617691040039) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7092261904761905) - val-aux/sciknoweval/reward/std@16:np.float64(0.1437505073441559) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7665238095238094) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.11153257282472556) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6511952380952379) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.13026803458061142) - val-aux/sciknoweval/reward/maj@2/mean:np.flo
|
||||
step:91 - global_seqlen/min:20110 - global_seqlen/max:28143 - global_seqlen/minmax_diff:8033 - global_seqlen/balanced_min:22954 - global_seqlen/balanced_max:22958 - global_seqlen/mean:22955.625 - actor/entropy:0.4345111846923828 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.318805456161499 - training/rollout_probs_diff_mean:0.00483555207028985 - training/rollout_probs_diff_std:0.010917664505541325 - training/rollout_actor_probs_pearson_corr:0.9992064833641052 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8046875 - self_distillation/correct_sample_fraction:0.640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8046875 - rollout_corr/rollout_is_mean:1.0000874996185303 - rollout_corr/rollout_is_ratio_fraction_high:9.058217074198183e-06 - rollout_corr/rollout_is_ratio_fraction_low:1.8116434148396365e-05 - rollout_corr/rollout_is_max:2.0618367195129395 - rollout_corr/rollout_is_min:0.27237245440483093 - rollout_corr/rollout_is_std:0.037339948117733 - rollout_corr/rollout_is_eff_sample_size:0.9986079071919081 - rollout_corr/rollout_is_seq_mean:1.0000126361846924 - rollout_corr/rollout_is_seq_std:0.0018744589760899544 - rollout_corr/rollout_is_seq_max:1.0062435865402222 - rollout_corr/rollout_is_seq_min:0.9926995038986206 - rollout_corr/rollout_is_seq_max_deviation:0.0073004961013793945 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5809713881462812) - rollout_corr/training_log_ppl:np.float64(0.4482057260465808) - rollout_corr/kl:np.float64(0.0007782714664905654) - rollout_corr/k3_kl:np.float64(0.0007742295367734187) - rollout_corr/rollout_ppl:np.float64(1.5796857569366693) - rollout_corr/rollout_log_ppl:np.float64(0.4474274521926418) - rollout_corr/log_ppl_diff:np.float64(0.0007782738539390266) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016364274197258055) - rollout_corr/log_ppl_diff_max:np.float64(0.007638216018676758) - rollout_corr/log_ppl_diff_min:np.float64(-0.005470573902130127) - rollout_corr/ppl_ratio:np.float64(1.0007805426139385) - rollout_corr/chi2_token:np.float64(0.001570558873936534) - rollout_corr/chi2_seq:np.float64(0.9185826345346868) - actor/pg_loss:np.float64(0.004598198342137039) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.000316593827392353) - actor/ppo_kl:np.float64(5.939174411295767e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8046875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8046875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.640625) - self_distillation/token_reweight_w_mean:np.float64(7563.319094766863) - self_distillation/token_reweight_w_std:np.float64(150023.32301567192) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.999886890873313) - self_distillation/token_reweight_w_clip_frac:np.float64(0.08767912839539349) - self_distillation/empty_target_batch:np.float64(0.1953125) - actor/grad_norm:np.float64(0.2757617086172104) - perf/mfu/actor:np.float64(0.054246744251757004) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(201.1460475921631) - actor/lr:np.float64(1e-06) - training/global_step:91 - training/epoch:1 - critic/score/mean:0.640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0008684565545991063 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0008684565545991063 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:431.23828125 - response_l
|
||||
step:92 - global_seqlen/min:20808 - global_seqlen/max:26065 - global_seqlen/minmax_diff:5257 - global_seqlen/balanced_min:23344 - global_seqlen/balanced_max:23359 - global_seqlen/mean:23353.875 - actor/entropy:0.43259748816490173 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2583375573158264 - training/rollout_probs_diff_mean:0.004716021008789539 - training/rollout_probs_diff_std:0.010604353621602058 - training/rollout_actor_probs_pearson_corr:0.9992642402648926 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.828125 - self_distillation/correct_sample_fraction:0.609375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.828125 - rollout_corr/rollout_is_mean:1.0000042915344238 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:8.831503691908438e-06 - rollout_corr/rollout_is_max:1.6297050714492798 - rollout_corr/rollout_is_min:0.3157358169555664 - rollout_corr/rollout_is_std:0.03597741574048996 - rollout_corr/rollout_is_eff_sample_size:0.9987072987735173 - rollout_corr/rollout_is_seq_mean:0.9999913573265076 - rollout_corr/rollout_is_seq_std:0.0019688971806317568 - rollout_corr/rollout_is_seq_max:1.0057662725448608 - rollout_corr/rollout_is_seq_min:0.9920455813407898 - rollout_corr/rollout_is_seq_max_deviation:0.007954418659210205 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5944940163753927) - rollout_corr/training_log_ppl:np.float64(0.4538156447233632) - rollout_corr/kl:np.float64(0.0008374510896800302) - rollout_corr/k3_kl:np.float64(0.000751384576176406) - rollout_corr/rollout_ppl:np.float64(1.5930784032680094) - rollout_corr/rollout_log_ppl:np.float64(0.45297819224651903) - rollout_corr/log_ppl_diff:np.float64(0.0008374524768441916) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017089889151975513) - rollout_corr/log_ppl_diff_max:np.float64(0.008830234408378601) - rollout_corr/log_ppl_diff_min:np.float64(-0.005124390125274658) - rollout_corr/ppl_ratio:np.float64(1.0008400096558034) - rollout_corr/chi2_token:np.float64(0.0013158204965293407) - rollout_corr/chi2_seq:np.float64(0.8615647666156292) - actor/pg_loss:np.float64(0.00415497412905097) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.000329681670109494) - actor/ppo_kl:np.float64(0.000143634846812013) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.828125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.828125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.609375) - self_distillation/token_reweight_w_mean:np.float64(2081.352001107065) - self_distillation/token_reweight_w_std:np.float64(46560.20298913785) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0010203949641436) - self_distillation/token_reweight_w_clip_frac:np.float64(0.09130151267163455) - self_distillation/empty_target_batch:np.float64(0.171875) - actor/grad_norm:np.float64(0.25993651896715164) - perf/mfu/actor:np.float64(0.05560204307595709) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(200.7599754333496) - actor/lr:np.float64(1e-06) - training/global_step:92 - training/epoch:1 - critic/score/mean:0.609375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.609375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.011590245179831982 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.011590245179831982 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:442.30859375 - response_length/max:1140.0 - respon
|
||||
step:93 - global_seqlen/min:20104 - global_seqlen/max:26536 - global_seqlen/minmax_diff:6432 - global_seqlen/balanced_min:22165 - global_seqlen/balanced_max:22172 - global_seqlen/mean:22170.375 - actor/entropy:0.4773474633693695 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18463870882987976 - training/rollout_probs_diff_mean:0.005192984826862812 - training/rollout_probs_diff_std:0.01093792449682951 - training/rollout_actor_probs_pearson_corr:0.9992780685424805 - self_distillation/success_group_fraction:1.0 - self_distillation/success_sample_fraction:0.9921875 - self_distillation/correct_sample_fraction:0.83203125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.9921875 - rollout_corr/rollout_is_mean:1.0000450611114502 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.661463737487793 - rollout_corr/rollout_is_min:0.516041100025177 - rollout_corr/rollout_is_std:0.037531014531850815 - rollout_corr/rollout_is_eff_sample_size:0.9985935232071051 - rollout_corr/rollout_is_seq_mean:1.0000596046447754 - rollout_corr/rollout_is_seq_std:0.0020964141003787518 - rollout_corr/rollout_is_seq_max:1.0067566633224487 - rollout_corr/rollout_is_seq_min:0.9933486580848694 - rollout_corr/rollout_is_seq_max_deviation:0.0067566633224487305 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6435860991477966) - rollout_corr/training_log_ppl:np.float64(0.48362978905788623) - rollout_corr/kl:np.float64(0.000619826348660979) - rollout_corr/k3_kl:np.float64(0.0007837708870965798) - rollout_corr/rollout_ppl:np.float64(1.6425714306533337) - rollout_corr/rollout_log_ppl:np.float64(0.48300995968747884) - rollout_corr/log_ppl_diff:np.float64(0.0006198293704073876) - rollout_corr/log_ppl_abs_diff:np.float64(0.001762611762387678) - rollout_corr/log_ppl_diff_max:np.float64(0.009479671716690063) - rollout_corr/log_ppl_diff_min:np.float64(-0.005926311016082764) - rollout_corr/ppl_ratio:np.float64(1.0006224988028407) - rollout_corr/chi2_token:np.float64(0.0019295907113701105) - rollout_corr/chi2_seq:np.float64(1.5755573513451964) - actor/pg_loss:np.float64(0.004148329375311732) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002517454886401538) - actor/ppo_kl:np.float64(-5.375289406117645e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.9921875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.9921875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.83203125) - self_distillation/token_reweight_w_mean:np.float64(17544.475337664597) - self_distillation/token_reweight_w_std:np.float64(333058.533660894) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0001345220953226) - self_distillation/token_reweight_w_clip_frac:np.float64(0.08209344942588359) - self_distillation/empty_target_batch:np.float64(0.0078125) - actor/grad_norm:np.float64(0.2505064271390438) - perf/mfu/actor:np.float64(0.05288764926923402) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(200.60185623168945) - actor/lr:np.float64(1e-06) - training/global_step:93 - training/epoch:1 - critic/score/mean:0.83203125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.83203125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00024851804482750595 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00024851804482750595 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:426.35546875 - response_length/max:917.0 - response_
|
||||
step:94 - global_seqlen/min:20450 - global_seqlen/max:26778 - global_seqlen/minmax_diff:6328 - global_seqlen/balanced_min:23452 - global_seqlen/balanced_max:23461 - global_seqlen/mean:23457.625 - actor/entropy:0.45124438405036926 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.47588256001472473 - training/rollout_probs_diff_mean:0.004685731139034033 - training/rollout_probs_diff_std:0.010418391786515713 - training/rollout_actor_probs_pearson_corr:0.999316930770874 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.875 - self_distillation/correct_sample_fraction:0.76171875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.875 - rollout_corr/rollout_is_mean:0.9999880194664001 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.4401948394370265e-05 - rollout_corr/rollout_is_max:1.8073612451553345 - rollout_corr/rollout_is_min:0.04583345353603363 - rollout_corr/rollout_is_std:0.03505437821149826 - rollout_corr/rollout_is_eff_sample_size:0.9987726987625546 - rollout_corr/rollout_is_seq_mean:0.9999796152114868 - rollout_corr/rollout_is_seq_std:0.0017362802755087614 - rollout_corr/rollout_is_seq_max:1.0060876607894897 - rollout_corr/rollout_is_seq_min:0.9951460361480713 - rollout_corr/rollout_is_seq_max_deviation:0.006087660789489746 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.600627651438117) - rollout_corr/training_log_ppl:np.float64(0.4611963724019006) - rollout_corr/kl:np.float64(0.0007944866175790821) - rollout_corr/k3_kl:np.float64(0.000699135587751698) - rollout_corr/rollout_ppl:np.float64(1.599346871022135) - rollout_corr/rollout_log_ppl:np.float64(0.46040188503684476) - rollout_corr/log_ppl_diff:np.float64(0.0007944873650558293) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015184025396592915) - rollout_corr/log_ppl_diff_max:np.float64(0.007160961627960205) - rollout_corr/log_ppl_diff_min:np.float64(-0.004008620977401733) - rollout_corr/ppl_ratio:np.float64(1.0007963897660375) - rollout_corr/chi2_token:np.float64(0.0012088804505765438) - rollout_corr/chi2_seq:np.float64(0.3547051209025085) - actor/pg_loss:np.float64(0.0030129868537187576) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00016773132483649533) - actor/ppo_kl:np.float64(0.00010856679642934353) - actor/pg_clipfrac_lower:np.float64(7.659314178454224e-06) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.76171875) - self_distillation/token_reweight_w_mean:np.float64(4465.609730498167) - self_distillation/token_reweight_w_std:np.float64(81083.046648157) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9998130949679762) - self_distillation/token_reweight_w_clip_frac:np.float64(0.04856777528766543) - self_distillation/empty_target_batch:np.float64(0.125) - actor/grad_norm:np.float64(0.18460461869835854) - perf/mfu/actor:np.float64(0.05546533287041886) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(197.40380096435547) - actor/lr:np.float64(1e-06) - training/global_step:94 - training/epoch:1 - critic/score/mean:0.76171875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.76171875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.005895104259252548 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.005895104259252548 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:480.23828125 - response_length/max:104
|
||||
step:95 - global_seqlen/min:18176 - global_seqlen/max:26822 - global_seqlen/minmax_diff:8646 - global_seqlen/balanced_min:23320 - global_seqlen/balanced_max:23330 - global_seqlen/mean:23326.75 - actor/entropy:0.42938148975372314 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4102376401424408 - training/rollout_probs_diff_mean:0.0046544442884624004 - training/rollout_probs_diff_std:0.010648963041603565 - training/rollout_actor_probs_pearson_corr:0.9992654323577881 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.8984375 - self_distillation/correct_sample_fraction:0.68359375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8984375 - rollout_corr/rollout_is_mean:1.0000568628311157 - rollout_corr/rollout_is_ratio_fraction_high:1.6773175957496278e-05 - rollout_corr/rollout_is_ratio_fraction_low:8.386587978748139e-06 - rollout_corr/rollout_is_max:2.396075487136841 - rollout_corr/rollout_is_min:0.3959798514842987 - rollout_corr/rollout_is_std:0.035107050091028214 - rollout_corr/rollout_is_eff_sample_size:0.9987691312695665 - rollout_corr/rollout_is_seq_mean:1.0000056028366089 - rollout_corr/rollout_is_seq_std:0.0018295958871021867 - rollout_corr/rollout_is_seq_max:1.0062776803970337 - rollout_corr/rollout_is_seq_min:0.9954047203063965 - rollout_corr/rollout_is_seq_max_deviation:0.006277680397033691 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5648417291231453) - rollout_corr/training_log_ppl:np.float64(0.4370654333033599) - rollout_corr/kl:np.float64(0.0006150017093133897) - rollout_corr/k3_kl:np.float64(0.0007087535628897967) - rollout_corr/rollout_ppl:np.float64(1.5638857381418347) - rollout_corr/rollout_log_ppl:np.float64(0.43645042926073074) - rollout_corr/log_ppl_diff:np.float64(0.0006150040426291525) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015464308089576662) - rollout_corr/log_ppl_diff_max:np.float64(0.005993485450744629) - rollout_corr/log_ppl_diff_min:np.float64(-0.004650503396987915) - rollout_corr/ppl_ratio:np.float64(1.000617001671344) - rollout_corr/chi2_token:np.float64(0.0016058182809501886) - rollout_corr/chi2_seq:np.float64(0.5998094449751079) - actor/pg_loss:np.float64(0.005259391153231263) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00024053280617408745) - actor/ppo_kl:np.float64(-4.200763549278008e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8984375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8984375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.68359375) - self_distillation/token_reweight_w_mean:np.float64(142.92776992870495) - self_distillation/token_reweight_w_std:np.float64(3127.1786301988177) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0005555916577578) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0809866230993066) - self_distillation/empty_target_batch:np.float64(0.1015625) - actor/grad_norm:np.float64(0.24522369727492332) - perf/mfu/actor:np.float64(0.05459364258201367) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(197.3312644958496) - actor/lr:np.float64(1e-06) - training/global_step:95 - training/epoch:1 - critic/score/mean:0.68359375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.68359375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.017624415457248688 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.017624415457248688 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:465.7734375
|
||||
step:96 - global_seqlen/min:20301 - global_seqlen/max:28288 - global_seqlen/minmax_diff:7987 - global_seqlen/balanced_min:23030 - global_seqlen/balanced_max:23033 - global_seqlen/mean:23031.375 - actor/entropy:0.44385668635368347 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2772989869117737 - training/rollout_probs_diff_mean:0.004754771012812853 - training/rollout_probs_diff_std:0.010701586492359638 - training/rollout_actor_probs_pearson_corr:0.99928879737854 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - self_distillation/correct_sample_fraction:0.67578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8359375 - rollout_corr/rollout_is_mean:1.000027060508728 - rollout_corr/rollout_is_ratio_fraction_high:8.62165597936837e-06 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.084261417388916 - rollout_corr/rollout_is_min:0.5124618411064148 - rollout_corr/rollout_is_std:0.036241523921489716 - rollout_corr/rollout_is_eff_sample_size:0.9986881560335953 - rollout_corr/rollout_is_seq_mean:1.0000017881393433 - rollout_corr/rollout_is_seq_std:0.0018420576816424727 - rollout_corr/rollout_is_seq_max:1.0063093900680542 - rollout_corr/rollout_is_seq_min:0.9944624304771423 - rollout_corr/rollout_is_seq_max_deviation:0.006309390068054199 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.598726301919669) - rollout_corr/training_log_ppl:np.float64(0.45704233669675887) - rollout_corr/kl:np.float64(0.0006188384301958338) - rollout_corr/k3_kl:np.float64(0.0007431540022935224) - rollout_corr/rollout_ppl:np.float64(1.5976339941844344) - rollout_corr/rollout_log_ppl:np.float64(0.45642349659465253) - rollout_corr/log_ppl_diff:np.float64(0.0006188401021063328) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015413255896419287) - rollout_corr/log_ppl_diff_max:np.float64(0.0064958930015563965) - rollout_corr/log_ppl_diff_min:np.float64(-0.0055953264236450195) - rollout_corr/ppl_ratio:np.float64(1.0006208247505128) - rollout_corr/chi2_token:np.float64(0.0018268234562128782) - rollout_corr/chi2_seq:np.float64(1.2716746137011796) - actor/pg_loss:np.float64(0.004606917384080589) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00021080635951875593) - actor/ppo_kl:np.float64(-7.161125205357166e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.8359375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8359375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.67578125) - self_distillation/token_reweight_w_mean:np.float64(99.39291401347145) - self_distillation/token_reweight_w_std:np.float64(2908.174926977954) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9999677997548133) - self_distillation/token_reweight_w_clip_frac:np.float64(0.08357357300701551) - self_distillation/empty_target_batch:np.float64(0.1640625) - actor/grad_norm:np.float64(0.23777736723423004) - perf/mfu/actor:np.float64(0.05523982050510799) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(197.2464942932129) - actor/lr:np.float64(1e-06) - training/global_step:96 - training/epoch:1 - critic/score/mean:0.67578125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.67578125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.003513324772939086 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.003513324772939086 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:453.07421875 - response_length/
|
||||
step:97 - global_seqlen/min:21184 - global_seqlen/max:28019 - global_seqlen/minmax_diff:6835 - global_seqlen/balanced_min:24523 - global_seqlen/balanced_max:24538 - global_seqlen/mean:24532.875 - actor/entropy:0.43073078989982605 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5830099582672119 - training/rollout_probs_diff_mean:0.004653396084904671 - training/rollout_probs_diff_std:0.010732254013419151 - training/rollout_actor_probs_pearson_corr:0.9992837905883789 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.765625 - self_distillation/correct_sample_fraction:0.5625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.765625 - rollout_corr/rollout_is_mean:0.999879002571106 - rollout_corr/rollout_is_ratio_fraction_high:8.271640581369866e-06 - rollout_corr/rollout_is_ratio_fraction_low:1.654328116273973e-05 - rollout_corr/rollout_is_max:2.069552421569824 - rollout_corr/rollout_is_min:0.06639717519283295 - rollout_corr/rollout_is_std:0.03590112552046776 - rollout_corr/rollout_is_eff_sample_size:0.9987124115549366 - rollout_corr/rollout_is_seq_mean:0.9998928308486938 - rollout_corr/rollout_is_seq_std:0.001641205046325922 - rollout_corr/rollout_is_seq_max:1.008506178855896 - rollout_corr/rollout_is_seq_min:0.9953399300575256 - rollout_corr/rollout_is_seq_max_deviation:0.008506178855895996 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5909629054367542) - rollout_corr/training_log_ppl:np.float64(0.45360937030636705) - rollout_corr/kl:np.float64(0.0007517823871872054) - rollout_corr/k3_kl:np.float64(0.0007286852571155578) - rollout_corr/rollout_ppl:np.float64(1.5898262090049684) - rollout_corr/rollout_log_ppl:np.float64(0.45285758542013355) - rollout_corr/log_ppl_diff:np.float64(0.0007517848862335086) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015345667488873005) - rollout_corr/log_ppl_diff_max:np.float64(0.00910806655883789) - rollout_corr/log_ppl_diff_min:np.float64(-0.007346630096435547) - rollout_corr/ppl_ratio:np.float64(1.0007539454381913) - rollout_corr/chi2_token:np.float64(0.00142996059730649) - rollout_corr/chi2_seq:np.float64(0.9373617491219193) - actor/pg_loss:np.float64(0.0030622705817222595) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003035150989489921) - actor/ppo_kl:np.float64(-6.880035599365897e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.765625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.765625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.5625) - self_distillation/token_reweight_w_mean:np.float64(482.64709699410014) - self_distillation/token_reweight_w_std:np.float64(7838.689288595924) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0027755422051996) - self_distillation/token_reweight_w_clip_frac:np.float64(0.06985412852372974) - self_distillation/empty_target_batch:np.float64(0.234375) - actor/grad_norm:np.float64(0.2353973463177681) - perf/mfu/actor:np.float64(0.058007253459953285) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(197.27971267700195) - actor/lr:np.float64(1e-06) - training/global_step:97 - training/epoch:1 - critic/score/mean:0.5625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0036426237784326077 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0036426237784326077 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:472.24609375 - response_length/max:
|
||||
step:98 - global_seqlen/min:21081 - global_seqlen/max:28015 - global_seqlen/minmax_diff:6934 - global_seqlen/balanced_min:24041 - global_seqlen/balanced_max:24240 - global_seqlen/mean:24067.875 - actor/entropy:0.4481050968170166 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3950059413909912 - training/rollout_probs_diff_mean:0.0048210457898676395 - training/rollout_probs_diff_std:0.010794337838888168 - training/rollout_actor_probs_pearson_corr:0.9992629289627075 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.875 - self_distillation/correct_sample_fraction:0.609375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.875 - rollout_corr/rollout_is_mean:1.000078797340393 - rollout_corr/rollout_is_ratio_fraction_high:3.446166738285683e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.7230833691428415e-05 - rollout_corr/rollout_is_max:4.591516971588135 - rollout_corr/rollout_is_min:0.39532703161239624 - rollout_corr/rollout_is_std:0.03653962165117264 - rollout_corr/rollout_is_eff_sample_size:0.9986668739730418 - rollout_corr/rollout_is_seq_mean:1.000028133392334 - rollout_corr/rollout_is_seq_std:0.0019362398888915777 - rollout_corr/rollout_is_seq_max:1.0084097385406494 - rollout_corr/rollout_is_seq_min:0.9942269325256348 - rollout_corr/rollout_is_seq_max_deviation:0.008409738540649414 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.616344683803618) - rollout_corr/training_log_ppl:np.float64(0.4696293716551736) - rollout_corr/kl:np.float64(0.0008239057563663721) - rollout_corr/k3_kl:np.float64(0.000780692749458467) - rollout_corr/rollout_ppl:np.float64(1.6150271990336478) - rollout_corr/rollout_log_ppl:np.float64(0.46880546608008444) - rollout_corr/log_ppl_diff:np.float64(0.0008239055750891566) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015701879747211933) - rollout_corr/log_ppl_diff_max:np.float64(0.007200688123703003) - rollout_corr/log_ppl_diff_min:np.float64(-0.004303097724914551) - rollout_corr/ppl_ratio:np.float64(1.0008261122275144) - rollout_corr/chi2_token:np.float64(0.001507394015789032) - rollout_corr/chi2_seq:np.float64(0.24129036953672767) - actor/pg_loss:np.float64(0.003733934834599495) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.000350107799476973) - actor/ppo_kl:np.float64(0.00011479198082042785) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.609375) - self_distillation/token_reweight_w_mean:np.float64(2963.129227875499) - self_distillation/token_reweight_w_std:np.float64(42459.57496579323) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0034989113919437) - self_distillation/token_reweight_w_clip_frac:np.float64(0.09955895229359157) - self_distillation/empty_target_batch:np.float64(0.125) - actor/grad_norm:np.float64(0.2265518605709076) - perf/mfu/actor:np.float64(0.057423452397621935) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(197.2193260192871) - actor/lr:np.float64(1e-06) - training/global_step:98 - training/epoch:1 - critic/score/mean:0.609375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.609375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.008424799889326096 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.008424799889326096 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:453.40234375 - response_length/max:1533.0 -
|
||||
step:99 - global_seqlen/min:19692 - global_seqlen/max:25383 - global_seqlen/minmax_diff:5691 - global_seqlen/balanced_min:23698 - global_seqlen/balanced_max:23716 - global_seqlen/mean:23712.375 - actor/entropy:0.4742501676082611 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4091079831123352 - training/rollout_probs_diff_mean:0.004931129980832338 - training/rollout_probs_diff_std:0.010619550943374634 - training/rollout_actor_probs_pearson_corr:0.9992989301681519 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.765625 - self_distillation/correct_sample_fraction:0.578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.765625 - rollout_corr/rollout_is_mean:0.9998912215232849 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:9.113195119425654e-06 - rollout_corr/rollout_is_max:1.7866209745407104 - rollout_corr/rollout_is_min:0.43442291021347046 - rollout_corr/rollout_is_std:0.03628097102046013 - rollout_corr/rollout_is_eff_sample_size:0.9986851836242739 - rollout_corr/rollout_is_seq_mean:0.9998824596405029 - rollout_corr/rollout_is_seq_std:0.00225405883975327 - rollout_corr/rollout_is_seq_max:1.0066941976547241 - rollout_corr/rollout_is_seq_min:0.9925881028175354 - rollout_corr/rollout_is_seq_max_deviation:0.0074118971824646 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6540217595174909) - rollout_corr/training_log_ppl:np.float64(0.49091246334137395) - rollout_corr/kl:np.float64(0.0009071645049343324) - rollout_corr/k3_kl:np.float64(0.0007826964022683569) - rollout_corr/rollout_ppl:np.float64(1.6525658327154815) - rollout_corr/rollout_log_ppl:np.float64(0.4900052981101908) - rollout_corr/log_ppl_diff:np.float64(0.0009071652311831713) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018514589173719287) - rollout_corr/log_ppl_diff_max:np.float64(0.008251607418060303) - rollout_corr/log_ppl_diff_min:np.float64(-0.005766570568084717) - rollout_corr/ppl_ratio:np.float64(1.0009100551251322) - rollout_corr/chi2_token:np.float64(0.001357766566798091) - rollout_corr/chi2_seq:np.float64(1.0302318546455353) - actor/pg_loss:np.float64(0.004117259988561273) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00030947469667808036) - actor/ppo_kl:np.float64(9.188319657127408e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.765625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.765625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.578125) - self_distillation/token_reweight_w_mean:np.float64(190.36794193601236) - self_distillation/token_reweight_w_std:np.float64(4398.700665255601) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0010327426716685) - self_distillation/token_reweight_w_clip_frac:np.float64(0.09465871116844937) - self_distillation/empty_target_batch:np.float64(0.234375) - actor/grad_norm:np.float64(0.2390700727701187) - perf/mfu/actor:np.float64(0.055725058096870825) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(197.29999542236328) - actor/lr:np.float64(1e-06) - training/global_step:99 - training/epoch:1 - critic/score/mean:0.578125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.578125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.006170772016048431 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.006170772016048431 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:428.63671875 - response_length/max:1227.0 - resp
|
||||
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 100}
|
||||
validation generation end
|
||||
[prompt] system
|
||||
|
||||
Given a question and four options, please select the right answer. Respond in the following format:
|
||||
<reasoning>
|
||||
...
|
||||
</reasoning>
|
||||
<answer>
|
||||
...
|
||||
</answer>
|
||||
|
||||
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
|
||||
user
|
||||
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||||
|
||||
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||||
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||||
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||||
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||||
Please reason step by step.
|
||||
assistant
|
||||
<think>
|
||||
|
||||
</think>
|
||||
|
||||
|
||||
[response] <reasoning>
|
||||
To determine the correct reactant for the synthesis of the product **"Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1"**, we need to analyze the structure of the product and compare it with the structures of the given options.
|
||||
|
||||
The product contains:
|
||||
- A **nitrogen-containing ring** (likely an imine or azole ring),
|
||||
- A **chlorine atom** attached to a benzene ring,
|
||||
- A **trifluoromethyl group** (C(F)(F)F),
|
||||
- A **nitrogen atom** connected to a carbon chain containing a **chlorinated benzene ring** and a **fluorinated group**.
|
||||
|
||||
Now, let’s examine the options:
|
||||
|
||||
**Option A:**
|
||||
- Contains a **nitrogen-containing ring** (likely a pyrrole or similar),
|
||||
- A **chlorinated benzene ring** (with Cl),
|
||||
- A **trifluoromethyl group** (C(F)(F)F),
|
||||
- A **nitrogen atom** connected to a carbon chain that includes the **chlorinated benzene** and **fluorinated group**.
|
||||
|
||||
This matches the structure of the target product very closely, including all the key functional groups and their positions.
|
||||
|
||||
**Option B:**
|
||||
- Contains a **chlorine** and **nitrogen**, but the **fluorinated group** is not present in the correct position, and the **nitrogen ring** is not properly structured.
|
||||
|
||||
**Option C:**
|
||||
- Contains a **fluorinated group**, but the **chlorine** is attached to an **oxygen**, not directly to the benzene ring. The **fluorinated group** is in a different position than in the product.
|
||||
|
||||
**Option D:**
|
||||
- Contains a **fluorinated group** and a **chlorinated benzene ring**, but the **nitrogen-containing ring** is not present in the correct structure.
|
||||
|
||||
Only **Option A** contains all the required functional groups in the correct positions to serve as the **only correct reactant** for the synthesis of the given product.
|
||||
|
||||
</reasoning>
|
||||
<answer>
|
||||
A
|
||||
</answer>
|
||||
[ground_truth] A
|
||||
[score] 1.0
|
||||
[acc] 1.0
|
||||
[pred] A
|
||||
[incorrect_format] 1
|
||||
[feedback]
|
||||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_100
|
||||
step:100 - global_seqlen/min:23741 - global_seqlen/max:31057 - global_seqlen/minmax_diff:7316 - global_seqlen/balanced_min:24739 - global_seqlen/balanced_max:31757 - global_seqlen/mean:25617.625 - actor/entropy:0.4436495304107666 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24008721113204956 - training/rollout_probs_diff_mean:0.004401609301567078 - training/rollout_probs_diff_std:0.009758861735463142 - training/rollout_actor_probs_pearson_corr:0.9993872046470642 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.91796875 - self_distillation/correct_sample_fraction:0.69921875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.91796875 - rollout_corr/rollout_is_mean:1.0000593662261963 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.7518070936203003 - rollout_corr/rollout_is_min:0.5589723587036133 - rollout_corr/rollout_is_std:0.03378680348396301 - rollout_corr/rollout_is_eff_sample_size:0.9988599913647978 - rollout_corr/rollout_is_seq_mean:1.0000724792480469 - rollout_corr/rollout_is_seq_std:0.0018252491718158126 - rollout_corr/rollout_is_seq_max:1.0057400465011597 - rollout_corr/rollout_is_seq_min:0.9935486912727356 - rollout_corr/rollout_is_seq_max_deviation:0.006451308727264404 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6480820523574948) - rollout_corr/training_log_ppl:np.float64(0.4857633382116546) - rollout_corr/kl:np.float64(0.0007351505751955756) - rollout_corr/k3_kl:np.float64(0.0006855184932135217) - rollout_corr/rollout_ppl:np.float64(1.6468097120523453) - rollout_corr/rollout_log_ppl:np.float64(0.48502818830229444) - rollout_corr/log_ppl_diff:np.float64(0.0007351499093601888) - rollout_corr/log_ppl_abs_diff:np.float64(0.001427958528893214) - rollout_corr/log_ppl_diff_max:np.float64(0.007762610912322998) - rollout_corr/log_ppl_diff_min:np.float64(-0.004005610942840576) - rollout_corr/ppl_ratio:np.float64(1.0007370123639703) - rollout_corr/chi2_token:np.float64(0.0012815699446946383) - rollout_corr/chi2_seq:np.float64(0.31913869245909154) - actor/pg_loss:np.float64(0.003367550321854651) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003017752374034899) - actor/ppo_kl:np.float64(0.00018170115079385596) - actor/pg_clipfrac_lower:np.float64(0.0) - self_distillation/token_reweight_lambda:np.float64(0.5) - self_distillation/token_reweight_eps_w:np.float64(0.20000000000000004) - self_distillation/token_reweight_active_token_fraction:np.float64(0.91796875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.91796875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.69921875) - self_distillation/token_reweight_w_mean:np.float64(1570.6249493323267) - self_distillation/token_reweight_w_std:np.float64(40484.064735737746) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.001023317920044) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0961508322507143) - self_distillation/empty_target_batch:np.float64(0.08203125) - actor/grad_norm:np.float64(0.2478880025446415) - perf/mfu/actor:np.float64(0.05716674172282474) - perf/max_memory_allocated_gb:np.float64(134.57307815551758) - perf/max_memory_reserved_gb:np.float64(149.07421875) - perf/cpu_memory_used_gb:np.float64(197.12733459472656) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7065476190476191) - val-aux/sciknoweval/reward/std@16:np.float64(0.14963967195739616) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7672380952380953) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.12522806043858783) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6460952380952382) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.12793485078945213) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0
|
||||
("Final validation metrics: {'val-aux/sciknoweval/reward/mean@16': "
|
||||
"np.float64(0.7065476190476191), 'val-aux/sciknoweval/reward/std@16': "
|
||||
"np.float64(0.14963967195739616), 'val-aux/sciknoweval/reward/best@2/mean': "
|
||||
"np.float64(0.7672380952380953), 'val-aux/sciknoweval/reward/best@2/std': "
|
||||
"np.float64(0.12522806043858783), 'val-aux/sciknoweval/reward/worst@2/mean': "
|
||||
"np.float64(0.6460952380952382), 'val-aux/sciknoweval/reward/worst@2/std': "
|
||||
"np.float64(0.12793485078945213), 'val-aux/sciknoweval/reward/maj@2/mean': "
|
||||
"np.float64(0.7068333333333333), 'val-aux/sciknoweval/reward/maj@2/std': "
|
||||
"np.float64(0.14978365629442147), 'val-aux/sciknoweval/reward/best@4/mean': "
|
||||
"np.float64(0.8158333333333333), 'val-aux/sciknoweval/reward/best@4/std': "
|
||||
"np.float64(0.09352166658220648), 'val-aux/sciknoweval/reward/worst@4/mean': "
|
||||
"np.float64(0.5964190476190476), 'val-aux/sciknoweval/reward/worst@4/std': "
|
||||
"np.float64(0.09674390853009297), 'val-aux/sciknoweval/reward/maj@4/mean': "
|
||||
"np.float64(0.7151904761904762), 'val-aux/sciknoweval/reward/maj@4/std': "
|
||||
"np.float64(0.1197973344827459), 'val-aux/sciknoweval/reward/best@8/mean': "
|
||||
"np.float64(0.8520904761904763), 'val-aux/sciknoweval/reward/best@8/std': "
|
||||
"np.float64(0.06467821428517558), 'val-aux/sciknoweval/reward/worst@8/mean': "
|
||||
"np.float64(0.5596142857142857), 'val-aux/sciknoweval/reward/worst@8/std': "
|
||||
"np.float64(0.06871965760347537), 'val-aux/sciknoweval/reward/maj@8/mean': "
|
||||
"np.float64(0.7202857142857142), 'val-aux/sciknoweval/reward/maj@8/std': "
|
||||
"np.float64(0.08852864730736695), 'val-aux/sciknoweval/reward/best@16/mean': "
|
||||
"np.float64(0.8776952380952381), 'val-aux/sciknoweval/reward/best@16/std': "
|
||||
"np.float64(0.0403808452308748), 'val-aux/sciknoweval/reward/worst@16/mean': "
|
||||
"np.float64(0.5314904761904762), 'val-aux/sciknoweval/reward/worst@16/std': "
|
||||
"np.float64(0.04527152807209723), 'val-aux/sciknoweval/reward/maj@16/mean': "
|
||||
"np.float64(0.7233190476190475), 'val-aux/sciknoweval/reward/maj@16/std': "
|
||||
"np.float64(0.06487826170016521), 'val-aux/sciknoweval/score/mean@16': "
|
||||
"np.float64(0.7065476190476191), 'val-aux/sciknoweval/score/std@16': "
|
||||
"np.float64(0.14963967195739616), 'val-aux/sciknoweval/score/best@2/mean': "
|
||||
"np.float64(0.7672380952380953), 'val-aux/sciknoweval/score/best@2/std': "
|
||||
"np.float64(0.12522806043858783), 'val-aux/sciknoweval/score/worst@2/mean': "
|
||||
"np.float64(0.6460952380952382), 'val-aux/sciknoweval/score/worst@2/std': "
|
||||
"np.float64(0.12793485078945213), 'val-aux/sciknoweval/score/maj@2/mean': "
|
||||
"np.float64(0.7068333333333333), 'val-aux/sciknoweval/score/maj@2/std': "
|
||||
"np.float64(0.14978365629442147), 'val-aux/sciknoweval/score/best@4/mean': "
|
||||
"np.float64(0.8158333333333333), 'val-aux/sciknoweval/score/best@4/std': "
|
||||
"np.float64(0.09352166658220648), 'val-aux/sciknoweval/score/worst@4/mean': "
|
||||
"np.float64(0.5964190476190476), 'val-aux/sciknoweval/score/worst@4/std': "
|
||||
"np.float64(0.09674390853009297), 'val-aux/sciknoweval/score/maj@4/mean': "
|
||||
"np.float64(0.7151904761904762), 'val-aux/sciknoweval/score/maj@4/std': "
|
||||
"np.float64(0.1197973344827459), 'val-aux/sciknoweval/score/best@8/mean': "
|
||||
"np.float64(0.8520904761904763), 'val-aux/sciknoweval/score/best@8/std': "
|
||||
"np.float64(0.06467821428517558), 'val-aux/sciknoweval/score/worst@8/mean': "
|
||||
"np.float64(0.5596142857142857), 'val-aux/sciknoweval/score/worst@8/std': "
|
||||
"np.float64(0.06871965760347537), 'val-aux/sciknoweval/score/maj@8/mean': "
|
||||
"np.float64(0.7202857142857142), 'val-aux/sciknoweval/score/maj@8/std': "
|
||||
"np.float64(0.08852864730736695), 'val-aux/sciknoweval/score/best@16/mean': "
|
||||
"np.float64(0.8776952380952381), 'val-aux/sciknoweval/score/best@16/std': "
|
||||
"np.float64(0.0403808452308748), 'val-aux/sciknoweval/score/worst@16/mean': "
|
||||
"np.float64(0.5314904761904762), 'val-aux/sciknoweval/score/worst@16/std': "
|
||||
"np.float64(0.04527152807209723), 'val-aux/sciknoweval/score/maj@16/mean': "
|
||||
"np.float64(0.7233190476190475), 'val-aux/sciknoweval/score/maj@16/std': "
|
||||
"np.float64(0.06487826170016521), 'val-core/sciknoweval/acc/mean@16': "
|
||||
"np.float64(0.7065476190476191), 'val-aux/sciknoweval/acc/std@16': "
|
||||
"np.float64(0.14963967195739616), 'val-aux/sciknoweval/acc/best@2/mean': "
|
||||
"np.float64(0.7672380952380953), 'val-aux/sciknoweval/acc/best@2/std': "
|
||||
"np.float64(0.12522806043858783), 'val-aux/sciknoweval/acc/worst@2/mean': "
|
||||
"np.float64(0.6460952380952382), 'val-aux/sciknoweval/acc/worst@2/std': "
|
||||
"np.float64(0.12793485078945213), 'val-aux/sciknoweval/acc/maj@2/mean': "
|
||||
"np.float64(0.7068333333333333), 'val-aux/sciknoweval/acc/maj@2/std': "
|
||||
"np.float64(0.14978365629442147), 'val-aux/sciknoweval/acc/best@4/mean': "
|
||||
"np.float64(0.8158333333333333), 'val-aux/sciknoweval/acc/best@4/std': "
|
||||
"np.float64(0.09352166658220648), 'val-aux/sciknoweval/acc/worst@4/mean': "
|
||||
"np.float64(0.5964190476190476), 'val-aux/sciknoweval/acc/worst@4/std': "
|
||||
"np.float64(0.09674390853009297), 'val-aux/sciknoweval/acc/maj@4/mean': "
|
||||
"np.float64(0.7151904761904762), 'val-aux/sciknoweval/acc/maj@4/std': "
|
||||
"np.float64(0.1197973344827459), 'val-aux/sciknoweval/acc/best@8/mean': "
|
||||
"np.float64(0.8520904761904763), 'val-aux/sciknoweval/acc/best@8/std': "
|
||||
"np.float64(0.06467821428517558), 'val-aux/sciknoweval/acc/worst@8/mean': "
|
||||
"np.float64(0.5596142857142857), 'val-aux/sciknoweval/acc/worst@8/std': "
|
||||
"np.float64(0.06871965760347537), 'val-aux/sciknoweval/acc/maj@8/mean': "
|
||||
"np.float64(0.7202857142857142), 'val-aux/sciknoweval/acc/maj@8/std': "
|
||||
"np.float64(0.08852864730736695), 'val-core/sciknoweval/acc/best@16/mean': "
|
||||
"np.float64(0.8776952380952381), 'val-core/sciknoweval/acc/best@16/std': "
|
||||
"np.float64(0.0403808452308748), 'val-aux/sciknoweval/acc/worst@16/mean': "
|
||||
"np.float64(0.5314904761904762), 'val-aux/sciknoweval/acc/worst@16/std': "
|
||||
"np.float64(0.04527152807209723), 'val-core/sciknoweval/acc/maj@16/mean': "
|
||||
"np.float64(0.7233190476190475), 'val-core/sciknoweval/acc/maj@16/std': "
|
||||
'np.float64(0.06487826170016521), '
|
||||
"'val-aux/sciknoweval/incorrect_format/mean@16': "
|
||||
'np.float64(0.9982142857142857), '
|
||||
"'val-aux/sciknoweval/incorrect_format/std@16': "
|
||||
'np.float64(0.006916041689656103), '
|
||||
"'val-aux/sciknoweval/incorrect_format/best@2/mean': "
|
||||
'np.float64(0.999904761904762), '
|
||||
"'val-aux/sciknoweval/incorrect_format/best@2/std': "
|
||||
'np.float64(0.0016180983939162673), '
|
||||
"'val-aux/sciknoweval/incorrect_format/worst@2/mean': "
|
||||
'np.float64(0.9966190476190476), '
|
||||
"'val-aux/sciknoweval/incorrect_format/worst@2/std': "
|
||||
'np.float64(0.009226171602153282), '
|
||||
"'val-aux/sciknoweval/incorrect_format/maj@2/mean': "
|
||||
'np.float64(0.9982380952380953), '
|
||||
"'val-aux/sciknoweval/incorrect_format/maj@2/std': "
|
||||
'np.float64(0.006862312067166463), '
|
||||
"'val-aux/sciknoweval/incorrect_format/best@4/mean': np.float64(1.0), "
|
||||
"'val-aux/sciknoweval/incorrect_format/best@4/std': np.float64(0.0), "
|
||||
"'val-aux/sciknoweval/incorrect_format/worst@4/mean': "
|
||||
'np.float64(0.9934619047619048), '
|
||||
"'val-aux/sciknoweval/incorrect_format/worst@4/std': "
|
||||
'np.float64(0.012001753208129275), '
|
||||
"'val-aux/sciknoweval/incorrect_format/maj@4/mean': "
|
||||
'np.float64(0.9993142857142857), '
|
||||
"'val-aux/sciknoweval/incorrect_format/maj@4/std': "
|
||||
'np.float64(0.0043365672892029085), '
|
||||
"'val-aux/sciknoweval/incorrect_format/best@8/mean': np.float64(1.0), "
|
||||
"'val-aux/sciknoweval/incorrect_format/best@8/std': np.float64(0.0), "
|
||||
"'val-aux/sciknoweval/incorrect_format/worst@8/mean': "
|
||||
'np.float64(0.988647619047619), '
|
||||
"'val-aux/sciknoweval/incorrect_format/worst@8/std': "
|
||||
'np.float64(0.01397823789672266), '
|
||||
"'val-aux/sciknoweval/incorrect_format/maj@8/mean': "
|
||||
'np.float64(0.9998380952380953), '
|
||||
"'val-aux/sciknoweval/incorrect_format/maj@8/std': "
|
||||
'np.float64(0.0019314678388328032), '
|
||||
"'val-aux/sciknoweval/incorrect_format/best@16/mean': np.float64(1.0), "
|
||||
"'val-aux/sciknoweval/incorrect_format/best@16/std': np.float64(0.0), "
|
||||
"'val-aux/sciknoweval/incorrect_format/worst@16/mean': "
|
||||
'np.float64(0.9817619047619048), '
|
||||
"'val-aux/sciknoweval/incorrect_format/worst@16/std': "
|
||||
'np.float64(0.013705261025670075), '
|
||||
"'val-aux/sciknoweval/incorrect_format/maj@16/mean': "
|
||||
'np.float64(0.9999714285714286), '
|
||||
"'val-aux/sciknoweval/incorrect_format/maj@16/std': "
|
||||
"np.float64(0.0006015850374731164), 'val-aux/num_turns/min': np.int32(2), "
|
||||
"'val-aux/num_turns/max': np.int32(2), 'val-aux/num_turns/mean': "
|
||||
'np.float64(2.0)}')
|
||||
5789
artifacts/queue.log
Normal file
5789
artifacts/queue.log
Normal file
File diff suppressed because one or more lines are too long
286
artifacts/requirements.txt
Normal file
286
artifacts/requirements.txt
Normal file
@@ -0,0 +1,286 @@
|
||||
colorama==0.4.6
|
||||
psutil==7.1.3
|
||||
wheel==0.47.0
|
||||
pip==26.1.2
|
||||
mpmath==1.3.0
|
||||
typing_extensions==4.15.0
|
||||
pillow==12.2.0
|
||||
nvidia-cuda-runtime-cu12==12.8.90
|
||||
nvidia-cuda-nvrtc-cu12==12.8.93
|
||||
nvidia-cusparse-cu12==12.5.8.93
|
||||
nvidia-cufft-cu12==11.3.3.83
|
||||
nvidia-cuda-cupti-cu12==12.8.90
|
||||
nvidia-cublas-cu12==12.8.4.1
|
||||
llguidance==1.3.0
|
||||
anthropic==0.71.0
|
||||
nodeenv==1.10.0
|
||||
networkx==3.6.1
|
||||
MarkupSafe==3.0.3
|
||||
filelock==3.29.0
|
||||
nvidia-cudnn-cu12==9.10.2.21
|
||||
transformers==4.57.1
|
||||
Jinja2==3.1.6
|
||||
nvidia-cusolver-cu12==11.7.3.90
|
||||
torch==2.9.0
|
||||
rich-toolkit==0.20.1
|
||||
z3-solver==4.15.4.0
|
||||
word2number==1.1
|
||||
pytz==2026.2
|
||||
pylatexenc==2.10
|
||||
py-spy==0.4.2
|
||||
opencensus-context==0.1.3
|
||||
distlib==0.4.3
|
||||
colorful==0.5.8
|
||||
antlr4-python3-runtime==4.9.3
|
||||
zipp==4.1.0
|
||||
xxhash==3.7.0
|
||||
wrapt==2.2.1
|
||||
Werkzeug==3.1.8
|
||||
urllib3==2.7.0
|
||||
tzdata==2026.2
|
||||
typing-inspection==0.4.2
|
||||
tqdm==4.68.2
|
||||
tensorboard-data-server==0.7.2
|
||||
smmap==5.0.3
|
||||
six==1.17.0
|
||||
safetensors==0.8.0
|
||||
rpds-py==2026.5.1
|
||||
regex==2026.5.9
|
||||
pyzmq==27.1.0
|
||||
PyYAML==6.0.3
|
||||
Pygments==2.20.0
|
||||
pydantic_core==2.46.4
|
||||
pycparser==3.0
|
||||
pybind11==3.0.1
|
||||
pyasn1==0.6.3
|
||||
pyarrow==22.0.0
|
||||
psutil==7.1.3
|
||||
protobuf==6.33.6
|
||||
propcache==0.5.2
|
||||
prometheus_client==0.25.0
|
||||
pluggy==1.6.0
|
||||
platformdirs==4.10.0
|
||||
httpcore==1.0.9
|
||||
packaging==25.0
|
||||
orjson==3.11.9
|
||||
opentelemetry-api==1.42.1
|
||||
numpy==2.1.0
|
||||
multidict==6.7.1
|
||||
msgpack==1.2.0
|
||||
Markdown==3.10.2
|
||||
iniconfig==2.3.0
|
||||
idna==3.18
|
||||
identify==2.6.19
|
||||
hf-xet==1.5.1
|
||||
h11==0.16.0
|
||||
grpcio==1.81.1
|
||||
gitdb==4.0.12
|
||||
fsspec==2025.10.0
|
||||
frozenlist==1.8.0
|
||||
dill==0.4.0
|
||||
codetiming==1.4.0
|
||||
cloudpickle==3.1.2
|
||||
click==8.4.1
|
||||
charset-normalizer==3.4.7
|
||||
cfgv==3.5.0
|
||||
certifi==2026.5.20
|
||||
attrs==26.1.0
|
||||
annotated-types==0.7.0
|
||||
annotated-doc==0.0.4
|
||||
aiohappyeyeballs==2.6.2
|
||||
absl-py==2.4.0
|
||||
yarl==1.24.2
|
||||
uvicorn==0.40.0
|
||||
tensorboard==2.20.0
|
||||
smart_open==7.6.1
|
||||
sentry-sdk==2.62.0
|
||||
requests==2.34.2
|
||||
referencing==0.37.0
|
||||
pyvers==0.1.0
|
||||
python-discovery==1.4.2
|
||||
python-dateutil==2.9.0.post0
|
||||
pytest==9.1.0
|
||||
pydantic==2.13.4
|
||||
pyasn1_modules==0.4.2
|
||||
proto-plus==1.28.0
|
||||
nvidia-nvjitlink-cu12==12.8.93
|
||||
nvidia-curand-cu12==10.3.9.90
|
||||
omegaconf==2.3.1
|
||||
multiprocess==0.70.18
|
||||
latex2sympy2_extended==1.10.2
|
||||
latex2sympy2==1.5.4
|
||||
googleapis-common-protos==1.75.0
|
||||
cffi==2.0.0
|
||||
anyio==4.13.0
|
||||
aiosignal==1.4.0
|
||||
virtualenv==21.5.0
|
||||
starlette==0.50.0
|
||||
pytest-rerunfailures==16.3
|
||||
pytest-asyncio==1.4.0
|
||||
pandas==2.3.3
|
||||
nvidia-nccl-cu12==2.27.5
|
||||
math-verify==0.8.0
|
||||
jsonschema-specifications==2025.9.1
|
||||
hydra-core==1.3.2
|
||||
huggingface_hub==0.36.2
|
||||
httpx==0.28.1
|
||||
GitPython==3.1.50
|
||||
cryptography==49.0.0
|
||||
aiohttp==3.14.1
|
||||
wandb==0.23.1
|
||||
torchdata==0.11.0
|
||||
numba==0.61.2
|
||||
tensordict==0.10.0
|
||||
pre_commit==4.6.0
|
||||
liger_kernel==0.8.0
|
||||
jsonschema==4.26.0
|
||||
google-auth==2.54.0
|
||||
fastapi==0.127.0
|
||||
aiohttp-cors==0.8.1
|
||||
accelerate==1.12.0
|
||||
llvmlite==0.44.0
|
||||
google-api-core==2.29.0
|
||||
datasets==4.4.2
|
||||
peft==0.18.0
|
||||
opencensus==0.11.4
|
||||
verl==0.7.0.dev0
|
||||
einops==0.8.2
|
||||
py-cpuinfo==9.0.0
|
||||
nvidia-cusparselt-cu12==0.7.1
|
||||
websockets==16.0
|
||||
uvloop==0.22.1
|
||||
sniffio==1.3.1
|
||||
shellingham==1.5.4
|
||||
sentencepiece==0.2.1
|
||||
scipy==1.17.1
|
||||
rignore==0.7.6
|
||||
python-multipart==0.0.32
|
||||
python-json-logger==4.1.0
|
||||
python-dotenv==1.2.2
|
||||
pycountry==26.2.16
|
||||
partial-json-parser==0.2.1.1.post7
|
||||
opentelemetry-semantic-conventions-ai==0.4.13
|
||||
opencv-python-headless==4.13.0.92
|
||||
ninja==1.13.0
|
||||
nest-asyncio==1.6.0
|
||||
msgspec==0.21.1
|
||||
mdurl==0.1.2
|
||||
lark==1.2.2
|
||||
jiter==0.15.0
|
||||
interegular==0.3.3
|
||||
importlib_metadata==8.0.0
|
||||
httptools==0.8.0
|
||||
fastar==0.11.0
|
||||
dnspython==2.8.0
|
||||
distro==1.9.0
|
||||
diskcache==5.6.3
|
||||
detect-installer==0.1.0
|
||||
Deprecated==1.3.1
|
||||
cuda-pathfinder==1.5.5
|
||||
cachetools==7.1.4
|
||||
blake3==1.0.8
|
||||
astor==0.8.1
|
||||
airportsdata==20260315
|
||||
watchfiles==1.2.0
|
||||
tiktoken==0.13.0
|
||||
tilelang==0.1.9
|
||||
markdown-it-py==4.2.0
|
||||
gguf==0.19.0
|
||||
email-validator==2.3.0
|
||||
cuda-tile==1.3.0
|
||||
cupy-cuda12x==14.1.1
|
||||
nvidia-ml-py==13.610.43
|
||||
rich==15.0.0
|
||||
pydantic-settings==2.14.1
|
||||
pydantic-extra-types==2.11.1
|
||||
prometheus-fastapi-instrumentator==7.1.0
|
||||
supervisor==4.3.0
|
||||
tokenspeed-mla==0.1.2
|
||||
outlines_core==0.2.11
|
||||
typer==0.26.7
|
||||
quack-kernels==0.5.0
|
||||
nvidia-nvvm==13.2.78
|
||||
torch_c_dlpack_ext==0.1.5
|
||||
mistral_common==1.11.3
|
||||
fastapi-cloud-cli==0.20.0
|
||||
fastapi-cli==0.0.24
|
||||
cuda-toolkit==13.0.2
|
||||
tabulate==0.10.0
|
||||
sympy==1.14.0
|
||||
setuptools==79.0.1
|
||||
pybase64==1.4.3
|
||||
nvidia-cufile-cu12==1.13.1.3
|
||||
humming-kernels==0.1.4
|
||||
setproctitle==1.3.7
|
||||
triton==3.5.0
|
||||
nvidia-nvtx-cu12==12.8.90
|
||||
nvidia-cusparselt-cu13==0.8.0
|
||||
tokenspeed-triton==3.7.10.post20260531
|
||||
PyJWT==2.13.0
|
||||
pyelftools==0.33
|
||||
nvidia-nvshmem-cu12==3.3.20
|
||||
nvidia-nvtx==13.0.85
|
||||
nvidia-nvshmem-cu13==3.4.5
|
||||
nvidia-nvjitlink==13.0.88
|
||||
nvidia-nccl-cu13==2.28.9
|
||||
nvidia-curand==10.4.0.35
|
||||
nvidia-cufile==1.15.1.6
|
||||
nvidia-cudnn-frontend==1.25.0
|
||||
nvidia-cuda-runtime==13.0.96
|
||||
nvidia-cuda-nvrtc==13.0.88
|
||||
nvidia-cuda-cupti==13.0.85
|
||||
nvidia-cuda-crt==13.3.33
|
||||
nvidia-cuda-cccl==13.3.3.3.1
|
||||
nvidia-cublas==13.1.0.3
|
||||
ml_dtypes==0.5.4
|
||||
loguru==0.7.3
|
||||
jmespath==1.1.0
|
||||
ijson==3.5.0
|
||||
httpx-sse==0.4.3
|
||||
docstring_parser==0.18.0
|
||||
depyf==0.20.0
|
||||
cuda-core==1.0.1
|
||||
cuda-bindings==13.3.1
|
||||
cbor2==6.1.2
|
||||
apache-tvm-ffi==0.1.9
|
||||
mcp==1.27.2
|
||||
opentelemetry-semantic-conventions==0.63b1
|
||||
opentelemetry-proto==1.42.1
|
||||
nvidia-cusparse==12.6.3.3
|
||||
nvidia-cufft==12.0.0.61
|
||||
nvidia-cudnn-cu13==9.19.0.56
|
||||
nvidia-cuda-nvcc==13.2.78
|
||||
cuda-python==13.3.1
|
||||
fastsafetensors==0.3.2
|
||||
tokenizers==0.22.2
|
||||
sse-starlette==3.4.4
|
||||
opentelemetry-sdk==1.42.1
|
||||
nvidia-cutlass-dsl==4.5.2
|
||||
opentelemetry-exporter-otlp-proto-common==1.42.1
|
||||
openai-harmony==0.0.8
|
||||
openai==2.41.1
|
||||
nvidia-cutlass-dsl-libs-base==4.5.2
|
||||
nvidia-cusolver==12.0.4.66
|
||||
nvidia-cuda-tileiras==13.2.78
|
||||
nvidia-cutlass-dsl-libs-cu13==4.5.2
|
||||
lm-format-enforcer==0.11.3
|
||||
ray==2.53.0
|
||||
model-hosting-container-standards==0.1.15
|
||||
opentelemetry-exporter-otlp-proto-http==1.42.1
|
||||
opentelemetry-exporter-otlp-proto-grpc==1.42.1
|
||||
opentelemetry-exporter-otlp==1.42.1
|
||||
opentelemetry-exporter-prometheus==0.63b1
|
||||
xgrammar==0.1.27
|
||||
torchvision==0.24.0
|
||||
torchaudio==2.9.0
|
||||
flashinfer-python==0.5.3
|
||||
compressed-tensors==0.12.2
|
||||
vllm==0.12.0
|
||||
flash_attn==2.8.3
|
||||
flashinfer-cubin==0.5.3
|
||||
pyparsing==3.3.2
|
||||
kiwisolver==1.5.0
|
||||
fonttools==4.63.0
|
||||
cycler==0.12.1
|
||||
contourpy==1.3.3
|
||||
matplotlib==3.11.0
|
||||
109
artifacts/wandb-metadata.json
Normal file
109
artifacts/wandb-metadata.json
Normal file
@@ -0,0 +1,109 @@
|
||||
{
|
||||
"os": "Linux-6.8.0-71-generic-x86_64-with-glibc2.36",
|
||||
"python": "CPython 3.12.13",
|
||||
"startedAt": "2026-07-03T01:43:31.670132Z",
|
||||
"args": [
|
||||
"--node-ip-address=198.19.44.212",
|
||||
"--node-manager-port=37959",
|
||||
"--object-store-name=/tmp/ray_q3g_chemistry_rlsd_tr_Qwen3_8B/session_2026-07-03_01-41-16_482264_2253022/sockets/plasma_store",
|
||||
"--raylet-name=/tmp/ray_q3g_chemistry_rlsd_tr_Qwen3_8B/session_2026-07-03_01-41-16_482264_2253022/sockets/raylet",
|
||||
"--redis-address=None",
|
||||
"--metrics-agent-port=57708",
|
||||
"--logging-rotate-bytes=536870912",
|
||||
"--logging-rotate-backup-count=5",
|
||||
"--runtime-env-agent-port=62827",
|
||||
"--gcs-address=198.19.44.212:50038",
|
||||
"--session-name=session_2026-07-03_01-41-16_482264_2253022",
|
||||
"--temp-dir=/tmp/ray_q3g_chemistry_rlsd_tr_Qwen3_8B",
|
||||
"--webui=",
|
||||
"--cluster-id=3afe38b5af2b40574604f3bc47f74eb7b868ca5b8931243913020512",
|
||||
"--startup-token=32",
|
||||
"--worker-launch-time-ms=1783042879915",
|
||||
"--node-id=5b00d08a465ef9eb566888683a3c2531820d4deaed97e3957024d102",
|
||||
"--runtime-env-hash=-1062963280"
|
||||
],
|
||||
"program": "/mnt/mole/SDPO/L2T/.venv/lib/python3.12/site-packages/ray/_private/workers/default_worker.py",
|
||||
"codePath": ".venv/lib/python3.12/site-packages/ray/_private/workers/default_worker.py",
|
||||
"codePathLocal": ".venv/lib/python3.12/site-packages/ray/_private/workers/default_worker.py",
|
||||
"git": {
|
||||
"remote": "https://github.com/jungseongryong/L2T.git",
|
||||
"commit": "ee6a667700697069b93db636aa937f970c1fb57a"
|
||||
},
|
||||
"email": "jungsr1116@cau.ac.kr",
|
||||
"root": "/mnt/mole/SDPO/L2T",
|
||||
"host": "mole-workspace-rw",
|
||||
"executable": "/mnt/mole/SDPO/L2T/.venv/bin/python",
|
||||
"cpu_count": 64,
|
||||
"cpu_count_logical": 128,
|
||||
"gpu": "NVIDIA H200",
|
||||
"gpu_count": 8,
|
||||
"disk": {
|
||||
"/": {
|
||||
"total": "46086056050688",
|
||||
"used": "24804622532608"
|
||||
}
|
||||
},
|
||||
"memory": {
|
||||
"total": "2163980390400"
|
||||
},
|
||||
"gpu_nvidia": [
|
||||
{
|
||||
"name": "NVIDIA H200",
|
||||
"memoryTotal": "150754820096",
|
||||
"cudaCores": 16896,
|
||||
"architecture": "Hopper",
|
||||
"uuid": "GPU-c0451621-891c-7169-976d-71d81958db34"
|
||||
},
|
||||
{
|
||||
"name": "NVIDIA H200",
|
||||
"memoryTotal": "150754820096",
|
||||
"cudaCores": 16896,
|
||||
"architecture": "Hopper",
|
||||
"uuid": "GPU-e2a29dfa-7dad-6cd7-fb05-79d0060f27f0"
|
||||
},
|
||||
{
|
||||
"name": "NVIDIA H200",
|
||||
"memoryTotal": "150754820096",
|
||||
"cudaCores": 16896,
|
||||
"architecture": "Hopper",
|
||||
"uuid": "GPU-d69bb741-f2a7-a375-7f56-5c0d09d797b2"
|
||||
},
|
||||
{
|
||||
"name": "NVIDIA H200",
|
||||
"memoryTotal": "150754820096",
|
||||
"cudaCores": 16896,
|
||||
"architecture": "Hopper",
|
||||
"uuid": "GPU-1bbbea1e-7541-3126-1bfc-51b43a67577e"
|
||||
},
|
||||
{
|
||||
"name": "NVIDIA H200",
|
||||
"memoryTotal": "150754820096",
|
||||
"cudaCores": 16896,
|
||||
"architecture": "Hopper",
|
||||
"uuid": "GPU-6e0055af-452d-3f46-ce21-0c71e347a608"
|
||||
},
|
||||
{
|
||||
"name": "NVIDIA H200",
|
||||
"memoryTotal": "150754820096",
|
||||
"cudaCores": 16896,
|
||||
"architecture": "Hopper",
|
||||
"uuid": "GPU-317fdb5b-b280-71d8-398e-a843ad897437"
|
||||
},
|
||||
{
|
||||
"name": "NVIDIA H200",
|
||||
"memoryTotal": "150754820096",
|
||||
"cudaCores": 16896,
|
||||
"architecture": "Hopper",
|
||||
"uuid": "GPU-f19d8161-d2dd-49a1-dcb8-5c341b5e4514"
|
||||
},
|
||||
{
|
||||
"name": "NVIDIA H200",
|
||||
"memoryTotal": "150754820096",
|
||||
"cudaCores": 16896,
|
||||
"architecture": "Hopper",
|
||||
"uuid": "GPU-961f5ebf-4bf8-6855-828b-c065af57b87c"
|
||||
}
|
||||
],
|
||||
"cudaVersion": "13.0",
|
||||
"writerId": "9v6x6pyjy21pdf97xdjkbtz80s5krcxu"
|
||||
}
|
||||
Reference in New Issue
Block a user