1099 lines
428 KiB
Plaintext
1099 lines
428 KiB
Plaintext
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/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-4B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/latest_checkpointed_iteration.txt
|
||
Training from scratch
|
||
Training Progress: 9%|▉ | 9/100 [05:01<47:05, 31.05s/it]
|
||
step:1 - global_seqlen/min:16884 - global_seqlen/max:25427 - global_seqlen/minmax_diff:8543 - global_seqlen/balanced_min:21909 - global_seqlen/balanced_max:21911 - global_seqlen/mean:21909.875 - actor/entropy:0.05220453813672066 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4939250648021698 - training/rollout_probs_diff_mean:0.0028897912707179785 - training/rollout_probs_diff_std:0.015141941606998444 - training/rollout_actor_probs_pearson_corr:0.9923052787780762 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.68359375 - 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:0.68359375 - rollout_corr/rollout_is_mean:1.0001649856567383 - rollout_corr/rollout_is_ratio_fraction_high:0.00010997201752616093 - rollout_corr/rollout_is_ratio_fraction_low:0.0004398880701046437 - rollout_corr/rollout_is_max:8.644091606140137 - rollout_corr/rollout_is_min:0.15896040201187134 - rollout_corr/rollout_is_std:0.03886362910270691 - rollout_corr/rollout_is_eff_sample_size:0.9984918961139008 - rollout_corr/rollout_is_seq_mean:1.0001929998397827 - rollout_corr/rollout_is_seq_std:0.0036600278690457344 - rollout_corr/rollout_is_seq_max:1.0304529666900635 - rollout_corr/rollout_is_seq_min:0.9936718940734863 - rollout_corr/rollout_is_seq_max_deviation:0.030452966690063477 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.052628115285188) - rollout_corr/training_log_ppl:np.float64(0.05033751486917026) - rollout_corr/kl:np.float64(0.0009047110073332476) - rollout_corr/k3_kl:np.float64(0.00109770550371735) - rollout_corr/rollout_ppl:np.float64(1.0516652716323733) - rollout_corr/rollout_log_ppl:np.float64(0.049432803332820185) - rollout_corr/log_ppl_diff:np.float64(0.000904711536350078) - rollout_corr/log_ppl_abs_diff:np.float64(0.002053696260190918) - rollout_corr/log_ppl_diff_max:np.float64(0.010051384568214417) - rollout_corr/log_ppl_diff_min:np.float64(-0.008277039974927902) - rollout_corr/ppl_ratio:np.float64(1.0009085375349969) - rollout_corr/chi2_token:np.float64(0.004085964523255825) - rollout_corr/chi2_seq:np.float64(1.1025378371123224) - actor/pg_loss:np.float64(0.0008221748284995556) - 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(0.68359375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.68359375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.55078125) - self_distillation/token_reweight_w_mean:np.float64(386915.0456223127) - self_distillation/token_reweight_w_std:np.float64(4613742.325882526) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0013521341606975) - self_distillation/token_reweight_w_clip_frac:np.float64(0.027686280984198675) - self_distillation/empty_target_batch:np.float64(0.31640625) - actor/grad_norm:np.float64(0.7696246653795242) - perf/mfu/actor:np.float64(0.0319981890644053) - perf/max_memory_allocated_gb:np.float64(116.94682025909424) - perf/max_memory_reserved_gb:np.float64(121.123046875) - perf/cpu_memory_used_gb:np.float64(155.23672485351562) - actor/lr:np.float64(0.0) - training/global_step:1 - 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:-0.002170420018956065 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.002170420018956065 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:319.68359375 - response_length/max:623.0 - respon
|
||
step:2 - global_seqlen/min:19173 - global_seqlen/max:25878 - global_seqlen/minmax_diff:6705 - global_seqlen/balanced_min:21916 - global_seqlen/balanced_max:21918 - global_seqlen/mean:21917.375 - actor/entropy:0.04673546180129051 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.603319525718689 - training/rollout_probs_diff_mean:0.0025578797794878483 - training/rollout_probs_diff_std:0.014482971280813217 - training/rollout_actor_probs_pearson_corr:0.9922713041305542 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65234375 - 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:0.65234375 - rollout_corr/rollout_is_mean:1.0001020431518555 - rollout_corr/rollout_is_ratio_fraction_high:0.00011651540989987552 - rollout_corr/rollout_is_ratio_fraction_low:0.00025892312987707555 - rollout_corr/rollout_is_max:2.880552053451538 - rollout_corr/rollout_is_min:0.2698138356208801 - rollout_corr/rollout_is_std:0.036167439073324203 - rollout_corr/rollout_is_eff_sample_size:0.9986938631090916 - rollout_corr/rollout_is_seq_mean:1.0001484155654907 - rollout_corr/rollout_is_seq_std:0.0026204257737845182 - rollout_corr/rollout_is_seq_max:1.0204964876174927 - rollout_corr/rollout_is_seq_min:0.9947845935821533 - rollout_corr/rollout_is_seq_max_deviation:0.020496487617492676 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0471329153515399) - rollout_corr/training_log_ppl:np.float64(0.0452757393468346) - rollout_corr/kl:np.float64(0.0005820094276507426) - rollout_corr/k3_kl:np.float64(0.0007234128699593612) - rollout_corr/rollout_ppl:np.float64(1.046513618901372) - rollout_corr/rollout_log_ppl:np.float64(0.044693729219943634) - rollout_corr/log_ppl_diff:np.float64(0.0005820101268909639) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016724726065149298) - rollout_corr/log_ppl_diff_max:np.float64(0.0078116655349731445) - rollout_corr/log_ppl_diff_min:np.float64(-0.013105824589729309) - rollout_corr/ppl_ratio:np.float64(1.0005849157460034) - rollout_corr/chi2_token:np.float64(0.0018090438097715378) - rollout_corr/chi2_seq:np.float64(0.5638161306269467) - actor/pg_loss:np.float64(0.0029225065372884274) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005952740175416693) - actor/ppo_kl:np.float64(8.465423888992518e-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(0.65234375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.65234375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.5078125) - self_distillation/token_reweight_w_mean:np.float64(538980.418017009) - self_distillation/token_reweight_w_std:np.float64(5349243.185795557) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0007042260840535) - self_distillation/token_reweight_w_clip_frac:np.float64(0.03253644743381301) - self_distillation/empty_target_batch:np.float64(0.34765625) - actor/grad_norm:np.float64(0.5853021740913391) - perf/mfu/actor:np.float64(0.0395066146021501) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.123046875) - perf/cpu_memory_used_gb:np.float64(148.72659301757812) - actor/lr:np.float64(1e-07) - training/global_step:2 - 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:0.002992180408909917 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.002992180408909917 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:301.73046875
|
||
step:3 - global_seqlen/min:19090 - global_seqlen/max:24914 - global_seqlen/minmax_diff:5824 - global_seqlen/balanced_min:21992 - global_seqlen/balanced_max:21997 - global_seqlen/mean:21995.0 - actor/entropy:0.05087893828749657 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4744122624397278 - training/rollout_probs_diff_mean:0.0026655818801373243 - training/rollout_probs_diff_std:0.014628268778324127 - training/rollout_actor_probs_pearson_corr:0.9926422834396362 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65625 - 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:0.65625 - rollout_corr/rollout_is_mean:0.999974250793457 - rollout_corr/rollout_is_ratio_fraction_high:0.00014163576997816563 - rollout_corr/rollout_is_ratio_fraction_low:0.00027039553970098495 - rollout_corr/rollout_is_max:2.957324981689453 - rollout_corr/rollout_is_min:0.16658133268356323 - rollout_corr/rollout_is_std:0.03663085773587227 - rollout_corr/rollout_is_eff_sample_size:0.9986597405244322 - rollout_corr/rollout_is_seq_mean:0.9999274611473083 - rollout_corr/rollout_is_seq_std:0.002721959026530385 - rollout_corr/rollout_is_seq_max:1.0112687349319458 - rollout_corr/rollout_is_seq_min:0.9901095032691956 - rollout_corr/rollout_is_seq_max_deviation:0.0112687349319458 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0487183942459524) - rollout_corr/training_log_ppl:np.float64(0.04701724427377485) - rollout_corr/kl:np.float64(0.0009062165111648479) - rollout_corr/k3_kl:np.float64(0.0008099392354239399) - rollout_corr/rollout_ppl:np.float64(1.0477627934888005) - rollout_corr/rollout_log_ppl:np.float64(0.04611102748094709) - rollout_corr/log_ppl_diff:np.float64(0.0009062167928277631) - rollout_corr/log_ppl_abs_diff:np.float64(0.00218963389761484) - rollout_corr/log_ppl_diff_max:np.float64(0.011827714741230011) - rollout_corr/log_ppl_diff_min:np.float64(-0.009073961526155472) - rollout_corr/ppl_ratio:np.float64(1.000910591101274) - rollout_corr/chi2_token:np.float64(0.0014679760206490755) - rollout_corr/chi2_seq:np.float64(0.9060272467322648) - actor/pg_loss:np.float64(0.0008821621304377913) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00032237121331490926) - actor/ppo_kl:np.float64(4.3753385401856804e-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(0.65625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.65625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6171875) - self_distillation/token_reweight_w_mean:np.float64(166842.85989560653) - self_distillation/token_reweight_w_std:np.float64(1549589.6575031993) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9989383574575186) - self_distillation/token_reweight_w_clip_frac:np.float64(0.01831802512606373) - self_distillation/empty_target_batch:np.float64(0.34375) - actor/grad_norm:np.float64(0.3902961015701294) - perf/mfu/actor:np.float64(0.03959429188243401) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.140625) - perf/cpu_memory_used_gb:np.float64(145.1807632446289) - actor/lr:np.float64(2e-07) - training/global_step:3 - 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:0.0022404203191399574 - critic/advantages/max:0.375 - critic/advantages/min:-0.875 - critic/returns/mean:0.0022404203191399574 - critic/returns/max:0.375 - critic/returns/min:-0.875 - response_length/mean:303.375 - response_length/max:83
|
||
step:4 - global_seqlen/min:17677 - global_seqlen/max:23520 - global_seqlen/minmax_diff:5843 - global_seqlen/balanced_min:21128 - global_seqlen/balanced_max:21131 - global_seqlen/mean:21129.875 - actor/entropy:0.04886283725500107 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.39975839853286743 - training/rollout_probs_diff_mean:0.0025600639637559652 - training/rollout_probs_diff_std:0.01360520999878645 - training/rollout_actor_probs_pearson_corr:0.9934918284416199 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71484375 - 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:0.71484375 - rollout_corr/rollout_is_mean:1.0000934600830078 - rollout_corr/rollout_is_ratio_fraction_high:0.0001256018440471962 - rollout_corr/rollout_is_ratio_fraction_low:0.00023724792117718607 - rollout_corr/rollout_is_max:2.7177038192749023 - rollout_corr/rollout_is_min:0.00025454876595176756 - rollout_corr/rollout_is_std:0.03572804480791092 - rollout_corr/rollout_is_eff_sample_size:0.998725372095864 - rollout_corr/rollout_is_seq_mean:1.0000848770141602 - rollout_corr/rollout_is_seq_std:0.002455302746966481 - rollout_corr/rollout_is_seq_max:1.008711814880371 - rollout_corr/rollout_is_seq_min:0.9932835698127747 - rollout_corr/rollout_is_seq_max_deviation:0.008711814880371094 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0504189697094262) - rollout_corr/training_log_ppl:np.float64(0.04830419022800925) - rollout_corr/kl:np.float64(0.0007874526963513873) - rollout_corr/k3_kl:np.float64(0.0008414268101475386) - rollout_corr/rollout_ppl:np.float64(1.0495754312723875) - rollout_corr/rollout_log_ppl:np.float64(0.04751673756254604) - rollout_corr/log_ppl_diff:np.float64(0.0007874526654632064) - rollout_corr/log_ppl_abs_diff:np.float64(0.0021492367868631845) - rollout_corr/log_ppl_diff_max:np.float64(0.02507532387971878) - rollout_corr/log_ppl_diff_min:np.float64(-0.0065584443509578705) - rollout_corr/ppl_ratio:np.float64(1.0007923119701445) - rollout_corr/chi2_token:np.float64(0.0016987675335258245) - rollout_corr/chi2_seq:np.float64(0.4790643185842782) - actor/pg_loss:np.float64(0.0018603979842737317) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004774848684974131) - actor/ppo_kl:np.float64(6.891565731592664e-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(0.71484375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71484375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.62109375) - self_distillation/token_reweight_w_mean:np.float64(258505.98163826996) - self_distillation/token_reweight_w_std:np.float64(2970903.3340365184) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0009436684194952) - self_distillation/token_reweight_w_clip_frac:np.float64(0.023301916036871262) - self_distillation/empty_target_batch:np.float64(0.28515625) - actor/grad_norm:np.float64(0.3927927166223526) - perf/mfu/actor:np.float64(0.038632921844974494) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.140625) - perf/cpu_memory_used_gb:np.float64(144.30906677246094) - actor/lr:np.float64(3e-07) - training/global_step:4 - 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:0.0001727025373838842 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0001727025373838842 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:279
|
||
step:5 - global_seqlen/min:19375 - global_seqlen/max:25073 - global_seqlen/minmax_diff:5698 - global_seqlen/balanced_min:23071 - global_seqlen/balanced_max:23081 - global_seqlen/mean:23079.125 - actor/entropy:0.04640193283557892 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.932929515838623 - training/rollout_probs_diff_mean:0.0027079873252660036 - training/rollout_probs_diff_std:0.016817985102534294 - training/rollout_actor_probs_pearson_corr:0.9894599318504333 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.6875 - 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:0.6875 - rollout_corr/rollout_is_mean:1.000158429145813 - rollout_corr/rollout_is_ratio_fraction_high:0.0001317286369157955 - rollout_corr/rollout_is_ratio_fraction_low:0.0004071612493135035 - rollout_corr/rollout_is_max:9.990342140197754 - rollout_corr/rollout_is_min:0.060471389442682266 - rollout_corr/rollout_is_std:0.0379825085401535 - rollout_corr/rollout_is_eff_sample_size:0.9985597640769998 - rollout_corr/rollout_is_seq_mean:1.0000975131988525 - rollout_corr/rollout_is_seq_std:0.0027155140414834023 - rollout_corr/rollout_is_seq_max:1.022322416305542 - rollout_corr/rollout_is_seq_min:0.9924696683883667 - rollout_corr/rollout_is_seq_max_deviation:0.022322416305541992 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0472423760220408) - rollout_corr/training_log_ppl:np.float64(0.04542937656697177) - rollout_corr/kl:np.float64(0.0008827663722907886) - rollout_corr/k3_kl:np.float64(0.0008958050244363136) - rollout_corr/rollout_ppl:np.float64(1.046309506520629) - rollout_corr/rollout_log_ppl:np.float64(0.044546609537064796) - rollout_corr/log_ppl_diff:np.float64(0.0008827670299069723) - rollout_corr/log_ppl_abs_diff:np.float64(0.0020228019693604438) - rollout_corr/log_ppl_diff_max:np.float64(0.017049439251422882) - rollout_corr/log_ppl_diff_min:np.float64(-0.009903965517878532) - rollout_corr/ppl_ratio:np.float64(1.0008867790456861) - rollout_corr/chi2_token:np.float64(0.0024128060322254896) - rollout_corr/chi2_seq:np.float64(0.8199321425054222) - actor/pg_loss:np.float64(0.002238532295450568) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005521543926079175) - actor/ppo_kl:np.float64(0.00012161997225490495) - actor/pg_clipfrac_lower:np.float64(1.10658638732275e-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(0.6875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.6875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.55078125) - self_distillation/token_reweight_w_mean:np.float64(315907.14576262655) - self_distillation/token_reweight_w_std:np.float64(4066129.582987148) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000939420890063) - self_distillation/token_reweight_w_clip_frac:np.float64(0.01852622180012986) - self_distillation/empty_target_batch:np.float64(0.3125) - actor/grad_norm:np.float64(0.4212125167250633) - perf/mfu/actor:np.float64(0.042256150204857096) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.1796875) - perf/cpu_memory_used_gb:np.float64(143.5656852722168) - actor/lr:np.float64(4e-07) - training/global_step:5 - 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:0.010946949012577534 - critic/advantages/max:0.75 - critic/advantages/min:-0.75 - critic/returns/mean:0.010946949012577534 - critic/returns/max:0.75 - critic/returns/min:-0.75 - response_length/mean:326.19140625 - respon
|
||
step:6 - global_seqlen/min:20841 - global_seqlen/max:25836 - global_seqlen/minmax_diff:4995 - global_seqlen/balanced_min:22901 - global_seqlen/balanced_max:22911 - global_seqlen/mean:22908.0 - actor/entropy:0.061381950974464417 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7882620096206665 - training/rollout_probs_diff_mean:0.00288434699177742 - training/rollout_probs_diff_std:0.014909966848790646 - training/rollout_actor_probs_pearson_corr:0.9936090111732483 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.78125 - self_distillation/correct_sample_fraction:0.625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.78125 - rollout_corr/rollout_is_mean:0.9998575448989868 - rollout_corr/rollout_is_ratio_fraction_high:0.00012235708709340543 - rollout_corr/rollout_is_ratio_fraction_low:0.00039154267869889736 - rollout_corr/rollout_is_max:4.272493362426758 - rollout_corr/rollout_is_min:0.1879345029592514 - rollout_corr/rollout_is_std:0.037344738841056824 - rollout_corr/rollout_is_eff_sample_size:0.9986068372944764 - rollout_corr/rollout_is_seq_mean:0.9998612999916077 - rollout_corr/rollout_is_seq_std:0.0022746948525309563 - rollout_corr/rollout_is_seq_max:1.008756399154663 - rollout_corr/rollout_is_seq_min:0.9944114685058594 - rollout_corr/rollout_is_seq_max_deviation:0.008756399154663086 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0594791187904775) - rollout_corr/training_log_ppl:np.float64(0.05643353109553573) - rollout_corr/kl:np.float64(0.0009431499511398655) - rollout_corr/k3_kl:np.float64(0.0008027483744008634) - rollout_corr/rollout_ppl:np.float64(1.058475960046053) - rollout_corr/rollout_log_ppl:np.float64(0.055490380680566886) - rollout_corr/log_ppl_diff:np.float64(0.0009431504149688408) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019810996163869277) - rollout_corr/log_ppl_diff_max:np.float64(0.013158190995454788) - rollout_corr/log_ppl_diff_min:np.float64(-0.0052896589040756226) - rollout_corr/ppl_ratio:np.float64(1.0009466442279518) - rollout_corr/chi2_token:np.float64(0.0013544457033276558) - rollout_corr/chi2_seq:np.float64(0.35554804536513984) - actor/pg_loss:np.float64(0.0034894959535449743) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0007761314168419631) - actor/ppo_kl:np.float64(-1.1544899273019382e-05) - actor/pg_clipfrac_lower:np.float64(1.1322464160912205e-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(0.78125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.78125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.625) - self_distillation/token_reweight_w_mean:np.float64(327357.1017849911) - self_distillation/token_reweight_w_std:np.float64(3991956.410076243) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0000990864355117) - self_distillation/token_reweight_w_clip_frac:np.float64(0.03530408094229642) - self_distillation/empty_target_batch:np.float64(0.21875) - actor/grad_norm:np.float64(0.6698948740959167) - perf/mfu/actor:np.float64(0.041629396460479307) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.1796875) - perf/cpu_memory_used_gb:np.float64(142.73164749145508) - actor/lr:np.float64(5e-07) - training/global_step:6 - training/epoch:0 - critic/score/mean:0.625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.002935040509328246 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.002935040509328246 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:319.25 - response_lengt
|
||
step:7 - global_seqlen/min:20578 - global_seqlen/max:23542 - global_seqlen/minmax_diff:2964 - global_seqlen/balanced_min:21776 - global_seqlen/balanced_max:21780 - global_seqlen/mean:21777.875 - actor/entropy:0.041082244366407394 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7632356882095337 - training/rollout_probs_diff_mean:0.0023401619400829077 - training/rollout_probs_diff_std:0.013588822446763515 - training/rollout_actor_probs_pearson_corr:0.9919775128364563 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - 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:0.8125 - rollout_corr/rollout_is_mean:0.9999773502349854 - rollout_corr/rollout_is_ratio_fraction_high:8.138130942825228e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00017632618255447596 - rollout_corr/rollout_is_max:4.427189826965332 - rollout_corr/rollout_is_min:0.31313708424568176 - rollout_corr/rollout_is_std:0.03326589614152908 - rollout_corr/rollout_is_eff_sample_size:0.9988942465813453 - rollout_corr/rollout_is_seq_mean:1.0000526905059814 - rollout_corr/rollout_is_seq_std:0.002266512019559741 - rollout_corr/rollout_is_seq_max:1.010648488998413 - rollout_corr/rollout_is_seq_min:0.9942504167556763 - rollout_corr/rollout_is_seq_max_deviation:0.010648488998413086 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0387283051386476) - rollout_corr/training_log_ppl:np.float64(0.03770313827953942) - rollout_corr/kl:np.float64(0.0005458609889821275) - rollout_corr/k3_kl:np.float64(0.0007262877969829162) - rollout_corr/rollout_ppl:np.float64(1.0381519650109112) - rollout_corr/rollout_log_ppl:np.float64(0.0371572768290207) - rollout_corr/log_ppl_diff:np.float64(0.0005458614505187143) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018243853737658355) - rollout_corr/log_ppl_diff_max:np.float64(0.009508870542049408) - rollout_corr/log_ppl_diff_min:np.float64(-0.005496785044670105) - rollout_corr/ppl_ratio:np.float64(1.000548951793462) - rollout_corr/chi2_token:np.float64(0.0021165036596357822) - rollout_corr/chi2_seq:np.float64(0.8815795592963696) - actor/pg_loss:np.float64(0.002434899564832449) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006158806254461524) - actor/ppo_kl:np.float64(-3.5115794112527965e-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(0.8125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.8125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.66796875) - self_distillation/token_reweight_w_mean:np.float64(338665.25113054714) - self_distillation/token_reweight_w_std:np.float64(4103084.2462062733) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0004773365799338) - self_distillation/token_reweight_w_clip_frac:np.float64(0.02430391594680259) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.5386742353439331) - perf/mfu/actor:np.float64(0.04007456025987775) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.1796875) - perf/cpu_memory_used_gb:np.float64(142.17034912109375) - actor/lr:np.float64(6e-07) - training/global_step:7 - 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:0.0008477219962514937 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.0008477219962514937 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:287.99609375 - response_length
|
||
step:8 - global_seqlen/min:21846 - global_seqlen/max:28118 - global_seqlen/minmax_diff:6272 - global_seqlen/balanced_min:23821 - global_seqlen/balanced_max:23824 - global_seqlen/mean:23822.75 - actor/entropy:0.03920365869998932 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.655200719833374 - training/rollout_probs_diff_mean:0.002379731275141239 - training/rollout_probs_diff_std:0.0142827108502388 - training/rollout_actor_probs_pearson_corr:0.9913021922111511 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.6875 - 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:0.6875 - rollout_corr/rollout_is_mean:1.000138521194458 - rollout_corr/rollout_is_ratio_fraction_high:0.00015401748532895 - rollout_corr/rollout_is_ratio_fraction_low:0.0002369499852648005 - rollout_corr/rollout_is_max:3.0959160327911377 - rollout_corr/rollout_is_min:0.16313594579696655 - rollout_corr/rollout_is_std:0.035375967621803284 - rollout_corr/rollout_is_eff_sample_size:0.9987502239821981 - rollout_corr/rollout_is_seq_mean:1.0001851320266724 - rollout_corr/rollout_is_seq_std:0.002349885180592537 - rollout_corr/rollout_is_seq_max:1.0165387392044067 - rollout_corr/rollout_is_seq_min:0.9932057857513428 - rollout_corr/rollout_is_seq_max_deviation:0.01653873920440674 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0392427039332688) - rollout_corr/training_log_ppl:np.float64(0.038196586596313864) - rollout_corr/kl:np.float64(0.0005487694252828135) - rollout_corr/k3_kl:np.float64(0.0007983955072852211) - rollout_corr/rollout_ppl:np.float64(1.0386743173003197) - rollout_corr/rollout_log_ppl:np.float64(0.03764781677227802) - rollout_corr/log_ppl_diff:np.float64(0.0005487698240358441) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015752802414681355) - rollout_corr/log_ppl_diff_max:np.float64(0.010580610483884811) - rollout_corr/log_ppl_diff_min:np.float64(-0.007809683680534363) - rollout_corr/ppl_ratio:np.float64(1.0005512910429388) - rollout_corr/chi2_token:np.float64(0.0031555849127471447) - rollout_corr/chi2_seq:np.float64(0.5169845761265606) - actor/pg_loss:np.float64(0.0010190074099227786) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003858940949612588) - actor/ppo_kl:np.float64(3.8873301086539414e-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(0.6875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.6875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.62109375) - self_distillation/token_reweight_w_mean:np.float64(118259.53067091317) - self_distillation/token_reweight_w_std:np.float64(1720886.6347599835) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.999833413399756) - self_distillation/token_reweight_w_clip_frac:np.float64(0.013415048630122328) - self_distillation/empty_target_batch:np.float64(0.3125) - actor/grad_norm:np.float64(0.2754215709865093) - perf/mfu/actor:np.float64(0.04353131948916253) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.220703125) - perf/cpu_memory_used_gb:np.float64(141.9685821533203) - actor/lr:np.float64(7e-07) - training/global_step:8 - 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:0.0012128876987844706 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.0012128876987844706 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:329.7109375 - response_length/max:
|
||
step:9 - global_seqlen/min:20344 - global_seqlen/max:24766 - global_seqlen/minmax_diff:4422 - global_seqlen/balanced_min:23476 - global_seqlen/balanced_max:23477 - global_seqlen/mean:23476.5 - actor/entropy:0.03558506444096565 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9514046311378479 - training/rollout_probs_diff_mean:0.0024439506232738495 - training/rollout_probs_diff_std:0.016063909977674484 - training/rollout_actor_probs_pearson_corr:0.9874861240386963 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71484375 - 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:0.71484375 - rollout_corr/rollout_is_mean:1.0000462532043457 - rollout_corr/rollout_is_ratio_fraction_high:0.00014843401731923223 - rollout_corr/rollout_is_ratio_fraction_low:0.00033397655352018774 - rollout_corr/rollout_is_max:9.368364334106445 - rollout_corr/rollout_is_min:0.04748138040304184 - rollout_corr/rollout_is_std:0.03585793450474739 - rollout_corr/rollout_is_eff_sample_size:0.9987156219328152 - rollout_corr/rollout_is_seq_mean:1.0000134706497192 - rollout_corr/rollout_is_seq_std:0.002829483011737466 - rollout_corr/rollout_is_seq_max:1.026300072669983 - rollout_corr/rollout_is_seq_min:0.9934424161911011 - rollout_corr/rollout_is_seq_max_deviation:0.02630007266998291 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0366761242039502) - rollout_corr/training_log_ppl:np.float64(0.03581141377253516) - rollout_corr/kl:np.float64(0.0009130834553232603) - rollout_corr/k3_kl:np.float64(0.0009266735434110274) - rollout_corr/rollout_ppl:np.float64(1.035728181246668) - rollout_corr/rollout_log_ppl:np.float64(0.03489832976265461) - rollout_corr/log_ppl_diff:np.float64(0.0009130840098805493) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019995503262180137) - rollout_corr/log_ppl_diff_max:np.float64(0.018973413854837418) - rollout_corr/log_ppl_diff_min:np.float64(-0.004774460569024086) - rollout_corr/ppl_ratio:np.float64(1.000917831202969) - rollout_corr/chi2_token:np.float64(0.0017166594043374062) - rollout_corr/chi2_seq:np.float64(0.4378669585566968) - actor/pg_loss:np.float64(0.0019036654848605394) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004998436570531339) - actor/ppo_kl:np.float64(8.825525457956118e-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(0.71484375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71484375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.62109375) - self_distillation/token_reweight_w_mean:np.float64(311737.4288784778) - self_distillation/token_reweight_w_std:np.float64(3446273.8661390827) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.999256769195199) - self_distillation/token_reweight_w_clip_frac:np.float64(0.02237211080500856) - self_distillation/empty_target_batch:np.float64(0.28515625) - actor/grad_norm:np.float64(0.5329340994358063) - perf/mfu/actor:np.float64(0.04305836727668846) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.220703125) - perf/cpu_memory_used_gb:np.float64(141.86572265625) - actor/lr:np.float64(8e-07) - training/global_step:9 - 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:-0.0016698826802894473 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0016698826802894473 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:315.796875 -
|
||
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
|
||
Given the following crystal structure parameters for the material RbLa9(IrO6)4 (Material ID: mp-560657), calculate the volume of the unit cell (in ų).
|
||
|
||
Lattice Parameters:
|
||
|
||
a = 7.82 Å
|
||
b = 7.82 Å
|
||
c = 17.88 Å
|
||
α = 90.00º
|
||
β = 90.00º
|
||
γ = 90.00º
|
||
|
||
|
||
|
||
A: 1025.67 ų
|
||
|
||
B: 1094.31 ų
|
||
|
||
C: 1200.45 ų
|
||
|
||
D: 1150.78 ų
|
||
|
||
Please reason step by step.
|
||
assistant
|
||
<think>
|
||
|
||
</think>
|
||
|
||
|
||
[response] <reasoning>
|
||
The volume of a unit cell in a crystal structure can be calculated using the formula:
|
||
|
||
$$ V = a \times b \times c \times \sin(\alpha) \times \sin(\beta) \times \sin(\gamma) $$
|
||
|
||
However, since all angles (α, β, γ) are 90.00º, their sines are all 1. Therefore, the formula simplifies to:
|
||
|
||
$$ V = a \times b \times c $$
|
||
|
||
Given:
|
||
- $ a = 7.82 $ Å
|
||
- $ b = 7.82 $ Å
|
||
- $ c = 17.88 $ Å
|
||
|
||
Substituting the values:
|
||
|
||
$$ V = 7.82 \times 7.82 \times 17.88 $$
|
||
|
||
First, calculate $ 7.82 \times 7.82 $:
|
||
|
||
$$ 7.82 \times 7.82 = 61.1524 $$
|
||
|
||
Then multiply by 17.88:
|
||
|
||
$$ 61.1524 \times 17.88 \approx 1094.31 $$
|
||
|
||
Thus, the volume of the unit cell is approximately 1094.31 ų.
|
||
</reasoning>
|
||
|
||
<answer>
|
||
B
|
||
</answer>
|
||
[ground_truth] B
|
||
[score] 1.0
|
||
[acc] 1.0
|
||
[pred] B
|
||
[incorrect_format] 1
|
||
[feedback]
|
||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-4B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_10
|
||
step:10 - global_seqlen/min:20683 - global_seqlen/max:25751 - global_seqlen/minmax_diff:5068 - global_seqlen/balanced_min:23251 - global_seqlen/balanced_max:23253 - global_seqlen/mean:23252.5 - actor/entropy:0.04874306544661522 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5441901683807373 - training/rollout_probs_diff_mean:0.0026878858916461468 - training/rollout_probs_diff_std:0.014487761072814465 - training/rollout_actor_probs_pearson_corr:0.9926190972328186 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71875 - 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:0.71875 - rollout_corr/rollout_is_mean:0.9999591708183289 - rollout_corr/rollout_is_ratio_fraction_high:0.00016809951921459287 - rollout_corr/rollout_is_ratio_fraction_low:0.0003722203546203673 - rollout_corr/rollout_is_max:5.771607875823975 - rollout_corr/rollout_is_min:0.23010946810245514 - rollout_corr/rollout_is_std:0.03833707422018051 - rollout_corr/rollout_is_eff_sample_size:0.9985320690303704 - rollout_corr/rollout_is_seq_mean:1.0000007152557373 - rollout_corr/rollout_is_seq_std:0.003206924069672823 - rollout_corr/rollout_is_seq_max:1.0279439687728882 - rollout_corr/rollout_is_seq_min:0.9925442934036255 - rollout_corr/rollout_is_seq_max_deviation:0.027943968772888184 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.048803678713739) - rollout_corr/training_log_ppl:np.float64(0.04719654364271264) - rollout_corr/kl:np.float64(0.0009922661836050395) - rollout_corr/k3_kl:np.float64(0.0009911187928040022) - rollout_corr/rollout_ppl:np.float64(1.0477598579600453) - rollout_corr/rollout_log_ppl:np.float64(0.0462042771250708) - rollout_corr/log_ppl_diff:np.float64(0.0009922665176418377) - rollout_corr/log_ppl_abs_diff:np.float64(0.0020914054821332684) - rollout_corr/log_ppl_diff_max:np.float64(0.013107702136039734) - rollout_corr/log_ppl_diff_min:np.float64(-0.0085487961769104) - rollout_corr/ppl_ratio:np.float64(1.0009965328499675) - rollout_corr/chi2_token:np.float64(0.002523774513974786) - rollout_corr/chi2_seq:np.float64(0.6901092522311956) - actor/pg_loss:np.float64(0.001417467137798667) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004598129285113828) - actor/ppo_kl:np.float64(8.220813472037403e-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(0.71875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6328125) - self_distillation/token_reweight_w_mean:np.float64(172479.47193897865) - self_distillation/token_reweight_w_std:np.float64(2368412.845991519) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9999711983837187) - self_distillation/token_reweight_w_clip_frac:np.float64(0.018695179103815462) - self_distillation/empty_target_batch:np.float64(0.28125) - actor/grad_norm:np.float64(0.4023687466979027) - perf/mfu/actor:np.float64(0.04252209567923851) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.220703125) - perf/cpu_memory_used_gb:np.float64(141.84356307983398) - actor/lr:np.float64(9e-07) - val-aux/sciknoweval/reward/mean@16:np.float64(0.6529255319148937) - val-aux/sciknoweval/reward/std@16:np.float64(0.12541629540699634) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7040851063829787) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.09844906330661858) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6004680851063829) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.11257283375619549) - val-aux/sciknoweval/reward/maj@2/m
|
||
step:11 - global_seqlen/min:21959 - global_seqlen/max:24923 - global_seqlen/minmax_diff:2964 - global_seqlen/balanced_min:23564 - global_seqlen/balanced_max:23566 - global_seqlen/mean:23565.25 - actor/entropy:0.04310470446944237 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.725770115852356 - training/rollout_probs_diff_mean:0.0022230474278330803 - training/rollout_probs_diff_std:0.013385867699980736 - training/rollout_actor_probs_pearson_corr:0.9929795861244202 - self_distillation/success_group_fraction:0.59375 - self_distillation/success_sample_fraction:0.5859375 - 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:0.5859375 - rollout_corr/rollout_is_mean:1.0000650882720947 - rollout_corr/rollout_is_ratio_fraction_high:0.00010156867210753262 - rollout_corr/rollout_is_ratio_fraction_low:0.00024827898596413434 - rollout_corr/rollout_is_max:2.600924015045166 - rollout_corr/rollout_is_min:0.21203798055648804 - rollout_corr/rollout_is_std:0.032979775220155716 - rollout_corr/rollout_is_eff_sample_size:0.9989137540740176 - rollout_corr/rollout_is_seq_mean:1.0000079870224 - rollout_corr/rollout_is_seq_std:0.0018282465171068907 - rollout_corr/rollout_is_seq_max:1.0058741569519043 - rollout_corr/rollout_is_seq_min:0.99213707447052 - rollout_corr/rollout_is_seq_max_deviation:0.00786292552947998 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.04387602256611) - rollout_corr/training_log_ppl:np.float64(0.04242584969688323) - rollout_corr/kl:np.float64(0.0006134525538925573) - rollout_corr/k3_kl:np.float64(0.0007203111015670061) - rollout_corr/rollout_ppl:np.float64(1.043228325434029) - rollout_corr/rollout_log_ppl:np.float64(0.04181239714671392) - rollout_corr/log_ppl_diff:np.float64(0.0006134525501693133) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016334005340468138) - rollout_corr/log_ppl_diff_max:np.float64(0.010004661977291107) - rollout_corr/log_ppl_diff_min:np.float64(-0.004983656108379364) - rollout_corr/ppl_ratio:np.float64(1.0006159681361169) - rollout_corr/chi2_token:np.float64(0.0017383343074470758) - rollout_corr/chi2_seq:np.float64(0.5499970901291817) - actor/pg_loss:np.float64(0.0008666964713484049) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003046969727620308) - actor/ppo_kl:np.float64(3.5258178954222785e-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(0.5859375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.5859375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.48046875) - self_distillation/token_reweight_w_mean:np.float64(184026.21509693912) - self_distillation/token_reweight_w_std:np.float64(2792648.148677664) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0008155833929777) - self_distillation/token_reweight_w_clip_frac:np.float64(0.014882214221870527) - self_distillation/empty_target_batch:np.float64(0.4140625) - actor/grad_norm:np.float64(0.21856637299060822) - perf/mfu/actor:np.float64(0.04311457286710895) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.220703125) - perf/cpu_memory_used_gb:np.float64(144.24859619140625) - 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:0.006627355702221394 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.006627355702221394 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:346.1328125 - res
|
||
step:12 - global_seqlen/min:20706 - global_seqlen/max:24659 - global_seqlen/minmax_diff:3953 - global_seqlen/balanced_min:23407 - global_seqlen/balanced_max:23571 - global_seqlen/mean:23429.625 - actor/entropy:0.048118073493242264 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5477700233459473 - training/rollout_probs_diff_mean:0.0025818103458732367 - training/rollout_probs_diff_std:0.014044265262782574 - training/rollout_actor_probs_pearson_corr:0.9929377436637878 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80078125 - 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:0.80078125 - rollout_corr/rollout_is_mean:0.9999388456344604 - rollout_corr/rollout_is_ratio_fraction_high:9.061562013812363e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0004077702760696411 - rollout_corr/rollout_is_max:2.6784048080444336 - rollout_corr/rollout_is_min:0.13162191212177277 - rollout_corr/rollout_is_std:0.03624316677451134 - rollout_corr/rollout_is_eff_sample_size:0.9986879182401984 - rollout_corr/rollout_is_seq_mean:0.9999709129333496 - rollout_corr/rollout_is_seq_std:0.0021563658956438303 - rollout_corr/rollout_is_seq_max:1.0135271549224854 - rollout_corr/rollout_is_seq_min:0.9925267696380615 - rollout_corr/rollout_is_seq_max_deviation:0.013527154922485352 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0476039084605873) - rollout_corr/training_log_ppl:np.float64(0.04610199074159027) - rollout_corr/kl:np.float64(0.0009057767044708953) - rollout_corr/k3_kl:np.float64(0.0008639268171179992) - rollout_corr/rollout_ppl:np.float64(1.0466419314034283) - rollout_corr/rollout_log_ppl:np.float64(0.0451962135030044) - rollout_corr/log_ppl_diff:np.float64(0.0009057772385858698) - rollout_corr/log_ppl_abs_diff:np.float64(0.0020111543035454815) - rollout_corr/log_ppl_diff_max:np.float64(0.009585428982973099) - rollout_corr/log_ppl_diff_min:np.float64(-0.008911248296499252) - rollout_corr/ppl_ratio:np.float64(1.000909530557692) - rollout_corr/chi2_token:np.float64(0.0016000857576727867) - rollout_corr/chi2_seq:np.float64(0.8010658347047865) - actor/pg_loss:np.float64(0.0020140354754403234) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0008085124986791925) - actor/ppo_kl:np.float64(0.00010460224394570616) - 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(0.61328125) - self_distillation/token_reweight_w_mean:np.float64(478855.19562057476) - self_distillation/token_reweight_w_std:np.float64(5985952.214456502) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0002396942581981) - self_distillation/token_reweight_w_clip_frac:np.float64(0.030733069819689263) - self_distillation/empty_target_batch:np.float64(0.19921875) - actor/grad_norm:np.float64(0.40229838341474533) - perf/mfu/actor:np.float64(0.042521912948479346) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.494140625) - perf/cpu_memory_used_gb:np.float64(144.2531280517578) - actor/lr:np.float64(1e-06) - training/global_step:12 - 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:0.0012176473392173648 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0012176473392173648 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:3
|
||
step:13 - global_seqlen/min:22261 - global_seqlen/max:24593 - global_seqlen/minmax_diff:2332 - global_seqlen/balanced_min:23464 - global_seqlen/balanced_max:23466 - global_seqlen/mean:23465.25 - actor/entropy:0.040193114429712296 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.39976027607917786 - training/rollout_probs_diff_mean:0.002327559981495142 - training/rollout_probs_diff_std:0.013716135174036026 - training/rollout_actor_probs_pearson_corr:0.9918900728225708 - self_distillation/success_group_fraction:0.625 - self_distillation/success_sample_fraction:0.62109375 - self_distillation/correct_sample_fraction:0.51953125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.62109375 - rollout_corr/rollout_is_mean:1.0001914501190186 - rollout_corr/rollout_is_ratio_fraction_high:0.0001476632314734161 - rollout_corr/rollout_is_ratio_fraction_low:0.0002707159146666527 - rollout_corr/rollout_is_max:2.9459996223449707 - rollout_corr/rollout_is_min:0.25255417823791504 - rollout_corr/rollout_is_std:0.03525106608867645 - rollout_corr/rollout_is_eff_sample_size:0.9987594991657962 - rollout_corr/rollout_is_seq_mean:1.0001559257507324 - rollout_corr/rollout_is_seq_std:0.002297038910910487 - rollout_corr/rollout_is_seq_max:1.0091631412506104 - rollout_corr/rollout_is_seq_min:0.9913782477378845 - rollout_corr/rollout_is_seq_max_deviation:0.009163141250610352 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0394970341585577) - rollout_corr/training_log_ppl:np.float64(0.03850088814215269) - rollout_corr/kl:np.float64(0.0007251129889169761) - rollout_corr/k3_kl:np.float64(0.0006720473286687678) - rollout_corr/rollout_ppl:np.float64(1.0387373333796859) - rollout_corr/rollout_log_ppl:np.float64(0.037775775064801564) - rollout_corr/log_ppl_diff:np.float64(0.000725113077351125) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018534923983679619) - rollout_corr/log_ppl_diff_max:np.float64(0.012402225285768509) - rollout_corr/log_ppl_diff_min:np.float64(-0.005973838269710541) - rollout_corr/ppl_ratio:np.float64(1.0007284900639206) - rollout_corr/chi2_token:np.float64(0.0012406953610479832) - rollout_corr/chi2_seq:np.float64(1.1534826629795134) - actor/pg_loss:np.float64(0.0009124996140599251) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002558234486969013) - actor/ppo_kl:np.float64(0.00024511930615922495) - 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(0.62109375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.62109375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.51953125) - self_distillation/token_reweight_w_mean:np.float64(293645.92851710087) - self_distillation/token_reweight_w_std:np.float64(2976888.573851983) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0000577387399971) - self_distillation/token_reweight_w_clip_frac:np.float64(0.014476351476332638) - self_distillation/empty_target_batch:np.float64(0.37890625) - actor/grad_norm:np.float64(0.33747463673353195) - perf/mfu/actor:np.float64(0.04315192332532921) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.494140625) - perf/cpu_memory_used_gb:np.float64(144.07625579833984) - actor/lr:np.float64(1e-06) - training/global_step:13 - training/epoch:0 - critic/score/mean:0.51953125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.51953125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0010813255794346333 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0010813255794346333 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:31
|
||
step:14 - global_seqlen/min:18580 - global_seqlen/max:25429 - global_seqlen/minmax_diff:6849 - global_seqlen/balanced_min:22314 - global_seqlen/balanced_max:22321 - global_seqlen/mean:22318.875 - actor/entropy:0.046363964676856995 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7826330661773682 - training/rollout_probs_diff_mean:0.0022935778833925724 - training/rollout_probs_diff_std:0.013024184852838516 - training/rollout_actor_probs_pearson_corr:0.9938106536865234 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.84375 - self_distillation/correct_sample_fraction:0.7890625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.84375 - rollout_corr/rollout_is_mean:1.0002796649932861 - rollout_corr/rollout_is_ratio_fraction_high:0.0001367223885608837 - rollout_corr/rollout_is_ratio_fraction_low:0.000198868932784535 - rollout_corr/rollout_is_max:3.9407498836517334 - rollout_corr/rollout_is_min:0.12595464289188385 - rollout_corr/rollout_is_std:0.034748680889606476 - rollout_corr/rollout_is_eff_sample_size:0.9987944610216803 - rollout_corr/rollout_is_seq_mean:1.0002996921539307 - rollout_corr/rollout_is_seq_std:0.0024711701553314924 - rollout_corr/rollout_is_seq_max:1.015850305557251 - rollout_corr/rollout_is_seq_min:0.9948000311851501 - rollout_corr/rollout_is_seq_max_deviation:0.015850305557250977 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0458477013744414) - rollout_corr/training_log_ppl:np.float64(0.0445621470498736) - rollout_corr/kl:np.float64(0.0007553838675917746) - rollout_corr/k3_kl:np.float64(0.0007893300585628538) - rollout_corr/rollout_ppl:np.float64(1.0450514145195484) - rollout_corr/rollout_log_ppl:np.float64(0.04380676286928065) - rollout_corr/log_ppl_diff:np.float64(0.0007553841805929551) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019627007295639487) - rollout_corr/log_ppl_diff_max:np.float64(0.007131712511181831) - rollout_corr/log_ppl_diff_min:np.float64(-0.006304092705249786) - rollout_corr/ppl_ratio:np.float64(1.0007586732972413) - rollout_corr/chi2_token:np.float64(0.001714975107461214) - rollout_corr/chi2_seq:np.float64(0.7236321275122464) - actor/pg_loss:np.float64(0.0008898881496861577) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003190960019310296) - actor/ppo_kl:np.float64(0.0003813289751271043) - 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(0.84375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.84375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.7890625) - self_distillation/token_reweight_w_mean:np.float64(175243.05724862264) - self_distillation/token_reweight_w_std:np.float64(1823834.6288890508) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0001764104235917) - self_distillation/token_reweight_w_clip_frac:np.float64(0.011729434947483242) - self_distillation/empty_target_batch:np.float64(0.15625) - actor/grad_norm:np.float64(0.306204192340374) - perf/mfu/actor:np.float64(0.04095481831414559) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.494140625) - perf/cpu_memory_used_gb:np.float64(144.13796997070312) - actor/lr:np.float64(1e-06) - training/global_step:14 - training/epoch:0 - critic/score/mean:0.7890625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7890625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.009107575751841068 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.009107575751841068 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:314.27734375 - response_le
|
||
step:15 - global_seqlen/min:20380 - global_seqlen/max:25231 - global_seqlen/minmax_diff:4851 - global_seqlen/balanced_min:23489 - global_seqlen/balanced_max:23499 - global_seqlen/mean:23496.25 - actor/entropy:0.04732014238834381 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.583829402923584 - training/rollout_probs_diff_mean:0.0023648759815841913 - training/rollout_probs_diff_std:0.01362634263932705 - training/rollout_actor_probs_pearson_corr:0.9935072660446167 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.7421875 - 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:0.7421875 - rollout_corr/rollout_is_mean:1.000017762184143 - rollout_corr/rollout_is_ratio_fraction_high:7.937048212625086e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00026078871451318264 - rollout_corr/rollout_is_max:3.049744129180908 - rollout_corr/rollout_is_min:0.09667123854160309 - rollout_corr/rollout_is_std:0.035375967621803284 - rollout_corr/rollout_is_eff_sample_size:0.9987501050707064 - rollout_corr/rollout_is_seq_mean:1.000028133392334 - rollout_corr/rollout_is_seq_std:0.002376183634623885 - rollout_corr/rollout_is_seq_max:1.0091273784637451 - rollout_corr/rollout_is_seq_min:0.9939660429954529 - rollout_corr/rollout_is_seq_max_deviation:0.009127378463745117 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0488063767552376) - rollout_corr/training_log_ppl:np.float64(0.047080223926968756) - rollout_corr/kl:np.float64(0.0009711041951301347) - rollout_corr/k3_kl:np.float64(0.0009545547938358823) - rollout_corr/rollout_ppl:np.float64(1.0477678030729294) - rollout_corr/rollout_log_ppl:np.float64(0.04610911908275739) - rollout_corr/log_ppl_diff:np.float64(0.0009711048442113679) - rollout_corr/log_ppl_abs_diff:np.float64(0.002235111896879971) - rollout_corr/log_ppl_diff_max:np.float64(0.014504663646221161) - rollout_corr/log_ppl_diff_min:np.float64(-0.0077231526374816895) - rollout_corr/ppl_ratio:np.float64(1.0009760304819793) - rollout_corr/chi2_token:np.float64(0.002052654279395938) - rollout_corr/chi2_seq:np.float64(1.0901849407237023) - actor/pg_loss:np.float64(0.0008747054962441325) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.000255831333106471) - actor/ppo_kl:np.float64(0.000288242356635493) - actor/pg_clipfrac_lower:np.float64(5.612428140011616e-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(0.7421875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.7421875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.671875) - self_distillation/token_reweight_w_mean:np.float64(126971.70744604967) - self_distillation/token_reweight_w_std:np.float64(1583958.2053749266) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0001212873030454) - self_distillation/token_reweight_w_clip_frac:np.float64(0.013588594767497852) - self_distillation/empty_target_batch:np.float64(0.2578125) - actor/grad_norm:np.float64(0.2117880955338478) - perf/mfu/actor:np.float64(0.04322908957344593) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(144.1304931640625) - actor/lr:np.float64(1e-06) - training/global_step:15 - 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:-0.0024647368118166924 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0024647368118166924 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:344.
|
||
step:16 - global_seqlen/min:20902 - global_seqlen/max:26195 - global_seqlen/minmax_diff:5293 - global_seqlen/balanced_min:23678 - global_seqlen/balanced_max:23682 - global_seqlen/mean:23680.125 - actor/entropy:0.052205078303813934 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5838291645050049 - training/rollout_probs_diff_mean:0.0023996636737138033 - training/rollout_probs_diff_std:0.013024316169321537 - training/rollout_actor_probs_pearson_corr:0.9943579435348511 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.77734375 - self_distillation/correct_sample_fraction:0.70703125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.77734375 - rollout_corr/rollout_is_mean:1.0000131130218506 - rollout_corr/rollout_is_ratio_fraction_high:9.346558363176882e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00017524797294754535 - rollout_corr/rollout_is_max:3.5453145503997803 - rollout_corr/rollout_is_min:0.32816171646118164 - rollout_corr/rollout_is_std:0.03295627236366272 - rollout_corr/rollout_is_eff_sample_size:0.9989150625306705 - rollout_corr/rollout_is_seq_mean:1.000024676322937 - rollout_corr/rollout_is_seq_std:0.00202034879475832 - rollout_corr/rollout_is_seq_max:1.0080302953720093 - rollout_corr/rollout_is_seq_min:0.9921471476554871 - rollout_corr/rollout_is_seq_max_deviation:0.008030295372009277 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0531891910359263) - rollout_corr/training_log_ppl:np.float64(0.0511211731009098) - rollout_corr/kl:np.float64(0.0006379728563885578) - rollout_corr/k3_kl:np.float64(0.0006680498469080476) - rollout_corr/rollout_ppl:np.float64(1.0525041660293937) - rollout_corr/rollout_log_ppl:np.float64(0.050483200158851105) - rollout_corr/log_ppl_diff:np.float64(0.0006379729420586955) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017212007733178325) - rollout_corr/log_ppl_diff_max:np.float64(0.012218199670314789) - rollout_corr/log_ppl_diff_min:np.float64(-0.008264167234301567) - rollout_corr/ppl_ratio:np.float64(1.000640834448859) - rollout_corr/chi2_token:np.float64(0.0014848981518298388) - rollout_corr/chi2_seq:np.float64(0.45312996907159686) - actor/pg_loss:np.float64(0.0012663383968174458) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00036590168474504026) - actor/ppo_kl:np.float64(8.058341451189222e-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(0.77734375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.77734375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.70703125) - self_distillation/token_reweight_w_mean:np.float64(145555.51019261987) - self_distillation/token_reweight_w_std:np.float64(1861382.0687996645) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9992707001511008) - self_distillation/token_reweight_w_clip_frac:np.float64(0.022003042249707505) - self_distillation/empty_target_batch:np.float64(0.22265625) - actor/grad_norm:np.float64(0.2867910601198673) - perf/mfu/actor:np.float64(0.04326631512717655) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(144.0705108642578) - actor/lr:np.float64(1e-06) - training/global_step:16 - training/epoch:0 - critic/score/mean:0.70703125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.70703125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0011858446523547173 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0011858446523547173 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:3
|
||
step:17 - global_seqlen/min:19849 - global_seqlen/max:25601 - global_seqlen/minmax_diff:5752 - global_seqlen/balanced_min:22440 - global_seqlen/balanced_max:22447 - global_seqlen/mean:22445.25 - actor/entropy:0.057849083095788956 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6970136761665344 - training/rollout_probs_diff_mean:0.002556848106905818 - training/rollout_probs_diff_std:0.013555520214140415 - training/rollout_actor_probs_pearson_corr:0.9945673942565918 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65625 - 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:0.65625 - rollout_corr/rollout_is_mean:1.0001786947250366 - rollout_corr/rollout_is_ratio_fraction_high:0.00012861074355896562 - rollout_corr/rollout_is_ratio_fraction_low:0.00023149934713728726 - rollout_corr/rollout_is_max:2.902588129043579 - rollout_corr/rollout_is_min:0.17272382974624634 - rollout_corr/rollout_is_std:0.03630724921822548 - rollout_corr/rollout_is_eff_sample_size:0.9986839946654996 - rollout_corr/rollout_is_seq_mean:1.0002074241638184 - rollout_corr/rollout_is_seq_std:0.002327009802684188 - rollout_corr/rollout_is_seq_max:1.0122935771942139 - rollout_corr/rollout_is_seq_min:0.993233323097229 - rollout_corr/rollout_is_seq_max_deviation:0.012293577194213867 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0583903319202363) - rollout_corr/training_log_ppl:np.float64(0.055446361937356414) - rollout_corr/kl:np.float64(0.0006691735518495889) - rollout_corr/k3_kl:np.float64(0.0008160383540030125) - rollout_corr/rollout_ppl:np.float64(1.0576571812853217) - rollout_corr/rollout_log_ppl:np.float64(0.05477718745532911) - rollout_corr/log_ppl_diff:np.float64(0.0006691744820273016) - rollout_corr/log_ppl_abs_diff:np.float64(0.002002931298193289) - rollout_corr/log_ppl_diff_max:np.float64(0.016655150800943375) - rollout_corr/log_ppl_diff_min:np.float64(-0.008791331201791763) - rollout_corr/ppl_ratio:np.float64(1.0006732817273587) - rollout_corr/chi2_token:np.float64(0.0020205182954669) - rollout_corr/chi2_seq:np.float64(1.0497447636444122) - actor/pg_loss:np.float64(0.0015107695944607258) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00045274926287675044) - actor/ppo_kl:np.float64(0.00018547102608934551) - 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(0.65625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.65625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.578125) - self_distillation/token_reweight_w_mean:np.float64(183102.81868288037) - self_distillation/token_reweight_w_std:np.float64(1977929.5838652337) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0002353701274842) - self_distillation/token_reweight_w_clip_frac:np.float64(0.020067830417247023) - self_distillation/empty_target_batch:np.float64(0.34375) - actor/grad_norm:np.float64(0.3122555688023567) - perf/mfu/actor:np.float64(0.04100007467048859) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(144.06307220458984) - actor/lr:np.float64(1e-06) - training/global_step:17 - training/epoch:0 - 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:0.001212156261317432 - critic/advantages/max:0.75 - critic/advantages/min:-0.75 - critic/returns/mean:0.001212156261317432 - critic/returns/max:0.75 - critic/returns/min:-0.75 - response_length/mean:303.7265625 - response_length/ma
|
||
step:18 - global_seqlen/min:20824 - global_seqlen/max:23587 - global_seqlen/minmax_diff:2763 - global_seqlen/balanced_min:22174 - global_seqlen/balanced_max:22178 - global_seqlen/mean:22176.375 - actor/entropy:0.04556930065155029 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5623742341995239 - training/rollout_probs_diff_mean:0.0023348708637058735 - training/rollout_probs_diff_std:0.013050748035311699 - training/rollout_actor_probs_pearson_corr:0.9937483072280884 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.7109375 - 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:0.7109375 - rollout_corr/rollout_is_mean:1.000004768371582 - rollout_corr/rollout_is_ratio_fraction_high:9.006343316286802e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00015439445269294083 - rollout_corr/rollout_is_max:2.489572763442993 - rollout_corr/rollout_is_min:0.08543625473976135 - rollout_corr/rollout_is_std:0.034692030400037766 - rollout_corr/rollout_is_eff_sample_size:0.9987979097727459 - rollout_corr/rollout_is_seq_mean:0.9999890327453613 - rollout_corr/rollout_is_seq_std:0.0021977166179567575 - rollout_corr/rollout_is_seq_max:1.0082364082336426 - rollout_corr/rollout_is_seq_min:0.9947015643119812 - rollout_corr/rollout_is_seq_max_deviation:0.008236408233642578 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0492625422775745) - rollout_corr/training_log_ppl:np.float64(0.04721992137547204) - rollout_corr/kl:np.float64(0.0008239883364355993) - rollout_corr/k3_kl:np.float64(0.0007315587119585132) - rollout_corr/rollout_ppl:np.float64(1.0484029022045434) - rollout_corr/rollout_log_ppl:np.float64(0.046395932438372256) - rollout_corr/log_ppl_diff:np.float64(0.0008239889370997844) - rollout_corr/log_ppl_abs_diff:np.float64(0.002035233240803791) - rollout_corr/log_ppl_diff_max:np.float64(0.009614132344722748) - rollout_corr/log_ppl_diff_min:np.float64(-0.006796486675739288) - rollout_corr/ppl_ratio:np.float64(1.0008276372682303) - rollout_corr/chi2_token:np.float64(0.0012551543768495321) - rollout_corr/chi2_seq:np.float64(0.4554172637872398) - actor/pg_loss:np.float64(0.0013457525055855513) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005007227500755107) - actor/ppo_kl:np.float64(0.00018828859354158567) - 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(0.609375) - self_distillation/token_reweight_w_mean:np.float64(568884.8094898432) - self_distillation/token_reweight_w_std:np.float64(5111404.572583297) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0010124356485903) - self_distillation/token_reweight_w_clip_frac:np.float64(0.02188170356384944) - self_distillation/empty_target_batch:np.float64(0.2890625) - actor/grad_norm:np.float64(0.3384881727397442) - perf/mfu/actor:np.float64(0.04055189458228217) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(144.11115264892578) - actor/lr:np.float64(1e-06) - training/global_step:18 - training/epoch:0 - 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:0.0028788133058696985 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0028788133058696985 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:303.60546875 -
|
||
step:19 - global_seqlen/min:20557 - global_seqlen/max:27231 - global_seqlen/minmax_diff:6674 - global_seqlen/balanced_min:23834 - global_seqlen/balanced_max:23840 - global_seqlen/mean:23837.375 - actor/entropy:0.05351593345403671 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6858960390090942 - training/rollout_probs_diff_mean:0.0023528658784925938 - training/rollout_probs_diff_std:0.012376116588711739 - training/rollout_actor_probs_pearson_corr:0.9950661659240723 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.65625 - 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:0.9999329447746277 - rollout_corr/rollout_is_ratio_fraction_high:5.4138919949764386e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00028152239974588156 - rollout_corr/rollout_is_max:2.5501341819763184 - rollout_corr/rollout_is_min:0.03525986894965172 - rollout_corr/rollout_is_std:0.033763863146305084 - rollout_corr/rollout_is_eff_sample_size:0.9988611807426664 - rollout_corr/rollout_is_seq_mean:0.9999514818191528 - rollout_corr/rollout_is_seq_std:0.002110761357471347 - rollout_corr/rollout_is_seq_max:1.0080809593200684 - rollout_corr/rollout_is_seq_min:0.991990327835083 - rollout_corr/rollout_is_seq_max_deviation:0.00808095932006836 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.053273911587894) - rollout_corr/training_log_ppl:np.float64(0.051362776504902286) - rollout_corr/kl:np.float64(0.0007405043419939084) - rollout_corr/k3_kl:np.float64(0.0007356684092343357) - rollout_corr/rollout_ppl:np.float64(1.0524831032380462) - rollout_corr/rollout_log_ppl:np.float64(0.05062227204325609) - rollout_corr/log_ppl_diff:np.float64(0.0007405044616461964) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016553804071008926) - rollout_corr/log_ppl_diff_max:np.float64(0.011551260948181152) - rollout_corr/log_ppl_diff_min:np.float64(-0.005798947066068649) - rollout_corr/ppl_ratio:np.float64(1.000743430107832) - rollout_corr/chi2_token:np.float64(0.0014110791962593794) - rollout_corr/chi2_seq:np.float64(0.35151986847631633) - actor/pg_loss:np.float64(0.0021186874946579337) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006218356556928484) - actor/ppo_kl:np.float64(4.354648220283863e-05) - actor/pg_clipfrac_lower:np.float64(1.5024038475530688e-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(0.65625) - self_distillation/token_reweight_w_mean:np.float64(256372.4387400134) - self_distillation/token_reweight_w_std:np.float64(3430843.094189947) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000731071922928) - self_distillation/token_reweight_w_clip_frac:np.float64(0.02942131183226593) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.3311096951365471) - perf/mfu/actor:np.float64(0.043486210929261346) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(144.10540008544922) - actor/lr:np.float64(1e-06) - training/global_step:19 - training/epoch:0 - critic/score/mean:0.65625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.65625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0031359968706965446 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.0031359968706965446 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:360.76171875 - 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': 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
|
||
Given the following crystal structure parameters for the material RbLa9(IrO6)4 (Material ID: mp-560657), calculate the volume of the unit cell (in ų).
|
||
|
||
Lattice Parameters:
|
||
|
||
a = 7.82 Å
|
||
b = 7.82 Å
|
||
c = 17.88 Å
|
||
α = 90.00º
|
||
β = 90.00º
|
||
γ = 90.00º
|
||
|
||
|
||
|
||
A: 1025.67 ų
|
||
|
||
B: 1094.31 ų
|
||
|
||
C: 1200.45 ų
|
||
|
||
D: 1150.78 ų
|
||
|
||
Please reason step by step.
|
||
assistant
|
||
<think>
|
||
|
||
</think>
|
||
|
||
|
||
[response] <reasoning>
|
||
The volume $ V $ of a unit cell in a crystal structure can be calculated using the formula:
|
||
|
||
$$
|
||
V = a \times b \times c \times \sqrt{1 + 2\cos\alpha\cos\beta\cos\gamma - \cos^2\alpha - \cos^2\beta - \cos^2\gamma}
|
||
$$
|
||
|
||
However, since all the angles $ \alpha, \beta, \gamma $ are 90.00º, the formula simplifies because $ \cos(90^\circ) = 0 $. This makes the formula reduce to:
|
||
|
||
$$
|
||
V = a \times b \times c
|
||
$$
|
||
|
||
Given:
|
||
- $ a = 7.82 $ Å
|
||
- $ b = 7.82 $ Å
|
||
- $ c = 17.88 $ Å
|
||
|
||
$$
|
||
V = 7.82 \times 7.82 \times 17.88
|
||
$$
|
||
|
||
First, calculate $ 7.82 \times 7.82 $:
|
||
|
||
$$
|
||
7.82 \times 7.82 = 61.1524
|
||
$$
|
||
|
||
Then multiply by 17.88:
|
||
|
||
$$
|
||
61.1524 \times 17.88 \approx 1094.31
|
||
$$
|
||
|
||
Thus, the volume of the unit cell is approximately 1094.31 ų.
|
||
</reasoning>
|
||
<answer>
|
||
B
|
||
</answer>
|
||
[ground_truth] B
|
||
[score] 1.0
|
||
[acc] 1.0
|
||
[pred] B
|
||
[incorrect_format] 1
|
||
[feedback]
|
||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-4B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_20
|
||
step:20 - global_seqlen/min:20705 - global_seqlen/max:27047 - global_seqlen/minmax_diff:6342 - global_seqlen/balanced_min:23525 - global_seqlen/balanced_max:23528 - global_seqlen/mean:23526.875 - actor/entropy:0.05020154267549515 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9398549199104309 - training/rollout_probs_diff_mean:0.002406689804047346 - training/rollout_probs_diff_std:0.012952032499015331 - training/rollout_actor_probs_pearson_corr:0.9944028854370117 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71484375 - 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:0.71484375 - rollout_corr/rollout_is_mean:0.9998804330825806 - rollout_corr/rollout_is_ratio_fraction_high:8.183019235730171e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00016366038471460342 - rollout_corr/rollout_is_max:3.036044120788574 - rollout_corr/rollout_is_min:6.201778160175309e-05 - rollout_corr/rollout_is_std:0.03356378525495529 - rollout_corr/rollout_is_eff_sample_size:0.9988745019683117 - rollout_corr/rollout_is_seq_mean:0.9997984170913696 - rollout_corr/rollout_is_seq_std:0.002129780361428857 - rollout_corr/rollout_is_seq_max:1.006217122077942 - rollout_corr/rollout_is_seq_min:0.9912189245223999 - rollout_corr/rollout_is_seq_max_deviation:0.008781075477600098 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0514160860329866) - rollout_corr/training_log_ppl:np.float64(0.04974506429698522) - rollout_corr/kl:np.float64(0.000963009055914732) - rollout_corr/k3_kl:np.float64(0.0007886385717021938) - rollout_corr/rollout_ppl:np.float64(1.0503967669792473) - rollout_corr/rollout_log_ppl:np.float64(0.04878205476416042) - rollout_corr/log_ppl_diff:np.float64(0.000963009532824799) - rollout_corr/log_ppl_abs_diff:np.float64(0.001833038001677778) - rollout_corr/log_ppl_diff_max:np.float64(0.01848556473851204) - rollout_corr/log_ppl_diff_min:np.float64(-0.004827976226806641) - rollout_corr/ppl_ratio:np.float64(1.0009666522964835) - rollout_corr/chi2_token:np.float64(0.001090053003281355) - rollout_corr/chi2_seq:np.float64(0.2828070535324514) - actor/pg_loss:np.float64(0.0013467443641275167) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00040249306493933545) - actor/ppo_kl:np.float64(6.146384180816256e-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(0.71484375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71484375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.61328125) - self_distillation/token_reweight_w_mean:np.float64(141580.48590263491) - self_distillation/token_reweight_w_std:np.float64(2119907.649028) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000380342360586) - self_distillation/token_reweight_w_clip_frac:np.float64(0.017205427793669514) - self_distillation/empty_target_batch:np.float64(0.28515625) - actor/grad_norm:np.float64(0.27972015365958214) - perf/mfu/actor:np.float64(0.042532409561833504) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(144.1010398864746) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7014627659574468) - val-aux/sciknoweval/reward/std@16:np.float64(0.10150057696241616) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7414893617021275) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.0709393703512258) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6616702127659574) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.1000502036630689) - val-aux/sciknoweva
|
||
step:21 - global_seqlen/min:20775 - global_seqlen/max:26559 - global_seqlen/minmax_diff:5784 - global_seqlen/balanced_min:23627 - global_seqlen/balanced_max:23631 - global_seqlen/mean:23629.0 - actor/entropy:0.05921781063079834 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.8452574610710144 - training/rollout_probs_diff_mean:0.002616164507344365 - training/rollout_probs_diff_std:0.013296348042786121 - training/rollout_actor_probs_pearson_corr:0.9946975708007812 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.7421875 - 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:0.7421875 - rollout_corr/rollout_is_mean:0.9999632239341736 - rollout_corr/rollout_is_ratio_fraction_high:4.538852590485476e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00015885983884800225 - rollout_corr/rollout_is_max:2.8393192291259766 - rollout_corr/rollout_is_min:0.007855840027332306 - rollout_corr/rollout_is_std:0.03435363247990608 - rollout_corr/rollout_is_eff_sample_size:0.9988209813407378 - rollout_corr/rollout_is_seq_mean:0.9998990297317505 - rollout_corr/rollout_is_seq_std:0.0020701477769762278 - rollout_corr/rollout_is_seq_max:1.0062670707702637 - rollout_corr/rollout_is_seq_min:0.9937349557876587 - rollout_corr/rollout_is_seq_max_deviation:0.006267070770263672 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0587520101107657) - rollout_corr/training_log_ppl:np.float64(0.05634416791508556) - rollout_corr/kl:np.float64(0.0007776328535911148) - rollout_corr/k3_kl:np.float64(0.0008263086628659266) - rollout_corr/rollout_ppl:np.float64(1.0579165108501911) - rollout_corr/rollout_log_ppl:np.float64(0.055566534494573716) - rollout_corr/log_ppl_diff:np.float64(0.0007776334205118474) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018763783118629362) - rollout_corr/log_ppl_diff_max:np.float64(0.009136740118265152) - rollout_corr/log_ppl_diff_min:np.float64(-0.007562510669231415) - rollout_corr/ppl_ratio:np.float64(1.0007807007059455) - rollout_corr/chi2_token:np.float64(0.002017973456531763) - rollout_corr/chi2_seq:np.float64(0.6339192821178585) - actor/pg_loss:np.float64(0.0016546305269002914) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005765741029790661) - actor/ppo_kl:np.float64(-1.5955860524385912e-05) - actor/pg_clipfrac_lower:np.float64(1.859824169514468e-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(0.7421875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.7421875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.59375) - self_distillation/token_reweight_w_mean:np.float64(251456.31727892137) - self_distillation/token_reweight_w_std:np.float64(3611087.4808393884) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0008972657378763) - self_distillation/token_reweight_w_clip_frac:np.float64(0.023536412860266864) - self_distillation/empty_target_batch:np.float64(0.2578125) - actor/grad_norm:np.float64(0.30859530717134476) - perf/mfu/actor:np.float64(0.043484721780607125) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(144.56780242919922) - actor/lr:np.float64(1e-06) - training/global_step:21 - training/epoch:0 - critic/score/mean:0.59375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.59375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.006455099210143089 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.006455099210143089 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:3
|
||
step:22 - global_seqlen/min:18203 - global_seqlen/max:26237 - global_seqlen/minmax_diff:8034 - global_seqlen/balanced_min:22751 - global_seqlen/balanced_max:22753 - global_seqlen/mean:22751.875 - actor/entropy:0.04705189913511276 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4668269157409668 - training/rollout_probs_diff_mean:0.002335731405764818 - training/rollout_probs_diff_std:0.012947346083819866 - training/rollout_actor_probs_pearson_corr:0.993709921836853 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.78125 - 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:0.78125 - rollout_corr/rollout_is_mean:1.0000593662261963 - rollout_corr/rollout_is_ratio_fraction_high:9.395074448548257e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0002113891823682934 - rollout_corr/rollout_is_max:5.089398384094238 - rollout_corr/rollout_is_min:0.002671187976375222 - rollout_corr/rollout_is_std:0.033378586173057556 - rollout_corr/rollout_is_eff_sample_size:0.9988869909400563 - rollout_corr/rollout_is_seq_mean:1.0000618696212769 - rollout_corr/rollout_is_seq_std:0.002444080077111721 - rollout_corr/rollout_is_seq_max:1.0121523141860962 - rollout_corr/rollout_is_seq_min:0.9936273694038391 - rollout_corr/rollout_is_seq_max_deviation:0.012152314186096191 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0460349675267935) - rollout_corr/training_log_ppl:np.float64(0.04464656937489053) - rollout_corr/kl:np.float64(0.00055560777209962) - rollout_corr/k3_kl:np.float64(0.0007500760087446601) - rollout_corr/rollout_ppl:np.float64(1.0454549426212907) - rollout_corr/rollout_log_ppl:np.float64(0.04409096138442692) - rollout_corr/log_ppl_diff:np.float64(0.0005556079904636135) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019402096240810351) - rollout_corr/log_ppl_diff_max:np.float64(0.009045779705047607) - rollout_corr/log_ppl_diff_min:np.float64(-0.007189542055130005) - rollout_corr/ppl_ratio:np.float64(1.0005588978528976) - rollout_corr/chi2_token:np.float64(0.0021106130443513393) - rollout_corr/chi2_seq:np.float64(0.889191702241078) - actor/pg_loss:np.float64(0.0017684989143162966) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005356281330932688) - actor/ppo_kl:np.float64(-9.231486468763705e-05) - actor/pg_clipfrac_lower:np.float64(1.1421783710829914e-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(0.78125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.78125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.7109375) - self_distillation/token_reweight_w_mean:np.float64(110542.58722938504) - self_distillation/token_reweight_w_std:np.float64(1534410.273381344) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9997402860317379) - self_distillation/token_reweight_w_clip_frac:np.float64(0.020450474316021428) - self_distillation/empty_target_batch:np.float64(0.21875) - actor/grad_norm:np.float64(0.31053709238767624) - perf/mfu/actor:np.float64(0.040778937791466366) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(144.4330825805664) - actor/lr:np.float64(1e-06) - training/global_step:22 - training/epoch:0 - critic/score/mean:0.7109375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7109375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.004377517383545637 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.004377517383545637 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:332.62
|
||
step:23 - global_seqlen/min:16765 - global_seqlen/max:25461 - global_seqlen/minmax_diff:8696 - global_seqlen/balanced_min:22231 - global_seqlen/balanced_max:22234 - global_seqlen/mean:22233.0 - actor/entropy:0.051857516169548035 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.8924874663352966 - training/rollout_probs_diff_mean:0.002473828848451376 - training/rollout_probs_diff_std:0.014042152091860771 - training/rollout_actor_probs_pearson_corr:0.9934764504432678 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.87109375 - self_distillation/correct_sample_fraction:0.78125 - 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:0.999901294708252 - rollout_corr/rollout_is_ratio_fraction_high:8.789553248789161e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0002762430813163519 - rollout_corr/rollout_is_max:3.1934919357299805 - rollout_corr/rollout_is_min:0.05045768618583679 - rollout_corr/rollout_is_std:0.03492661938071251 - rollout_corr/rollout_is_eff_sample_size:0.9987814986875572 - rollout_corr/rollout_is_seq_mean:0.9999404549598694 - rollout_corr/rollout_is_seq_std:0.002376072108745575 - rollout_corr/rollout_is_seq_max:1.0088611841201782 - rollout_corr/rollout_is_seq_min:0.9945573806762695 - rollout_corr/rollout_is_seq_max_deviation:0.008861184120178223 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0516580799594522) - rollout_corr/training_log_ppl:np.float64(0.0500233011553064) - rollout_corr/kl:np.float64(0.0007878192633476999) - rollout_corr/k3_kl:np.float64(0.0008335432382278185) - rollout_corr/rollout_ppl:np.float64(1.0508313993923366) - rollout_corr/rollout_log_ppl:np.float64(0.04923548138322076) - rollout_corr/log_ppl_diff:np.float64(0.0007878197720856406) - rollout_corr/log_ppl_abs_diff:np.float64(0.001956198444531765) - rollout_corr/log_ppl_diff_max:np.float64(0.012837372720241547) - rollout_corr/log_ppl_diff_min:np.float64(-0.006586827337741852) - rollout_corr/ppl_ratio:np.float64(1.0007912560831755) - rollout_corr/chi2_token:np.float64(0.0018308169674128294) - rollout_corr/chi2_seq:np.float64(0.6063053519465029) - actor/pg_loss:np.float64(0.001039000111632049) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003526491568663914) - actor/ppo_kl:np.float64(-5.0933335728586826e-05) - actor/pg_clipfrac_lower:np.float64(8.510348379786592e-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(0.87109375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.87109375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.78125) - self_distillation/token_reweight_w_mean:np.float64(190123.5627077485) - self_distillation/token_reweight_w_std:np.float64(2663452.2716900078) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0003106228541583) - self_distillation/token_reweight_w_clip_frac:np.float64(0.015045241816551425) - self_distillation/empty_target_batch:np.float64(0.12890625) - actor/grad_norm:np.float64(0.28235869854688644) - perf/mfu/actor:np.float64(0.04055485686561945) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(144.30136489868164) - actor/lr:np.float64(1e-06) - training/global_step:23 - training/epoch:0 - critic/score/mean:0.78125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.78125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0057085007429122925 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0057085007429122925 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:3
|
||
step:24 - global_seqlen/min:20488 - global_seqlen/max:26007 - global_seqlen/minmax_diff:5519 - global_seqlen/balanced_min:23418 - global_seqlen/balanced_max:23421 - global_seqlen/mean:23419.625 - actor/entropy:0.048930879682302475 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.8673235774040222 - training/rollout_probs_diff_mean:0.002186768688261509 - training/rollout_probs_diff_std:0.012573663145303726 - training/rollout_actor_probs_pearson_corr:0.9943985939025879 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.9375 - self_distillation/correct_sample_fraction:0.875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.9375 - rollout_corr/rollout_is_mean:0.9998534321784973 - rollout_corr/rollout_is_ratio_fraction_high:3.3702941436786205e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0001797490258468315 - rollout_corr/rollout_is_max:2.5502538681030273 - rollout_corr/rollout_is_min:0.007685937453061342 - rollout_corr/rollout_is_std:0.030430540442466736 - rollout_corr/rollout_is_eff_sample_size:0.9990746009767545 - rollout_corr/rollout_is_seq_mean:0.9998515248298645 - rollout_corr/rollout_is_seq_std:0.0018696229672059417 - rollout_corr/rollout_is_seq_max:1.0090330839157104 - rollout_corr/rollout_is_seq_min:0.9939467906951904 - rollout_corr/rollout_is_seq_max_deviation:0.00903308391571045 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0505288592539728) - rollout_corr/training_log_ppl:np.float64(0.0487635411645897) - rollout_corr/kl:np.float64(0.0006307011686601527) - rollout_corr/k3_kl:np.float64(0.0005982627429759191) - rollout_corr/rollout_ppl:np.float64(1.049854587763548) - rollout_corr/rollout_log_ppl:np.float64(0.04813283952717029) - rollout_corr/log_ppl_diff:np.float64(0.0006307016374194063) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015282359381671995) - rollout_corr/log_ppl_diff_max:np.float64(0.009290523827075958) - rollout_corr/log_ppl_diff_min:np.float64(-0.005929280072450638) - rollout_corr/ppl_ratio:np.float64(1.000633022049442) - rollout_corr/chi2_token:np.float64(0.001259797951206565) - rollout_corr/chi2_seq:np.float64(0.3109738379716873) - actor/pg_loss:np.float64(0.0009422041475772858) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003166221840729122) - actor/ppo_kl:np.float64(-0.00011196534618562737) - 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(0.9375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.9375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.875) - self_distillation/token_reweight_w_mean:np.float64(158053.5074800374) - self_distillation/token_reweight_w_std:np.float64(2085045.1837178322) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.999543417012319) - self_distillation/token_reweight_w_clip_frac:np.float64(0.01681827876745956) - self_distillation/empty_target_batch:np.float64(0.0625) - actor/grad_norm:np.float64(0.25178469717502594) - perf/mfu/actor:np.float64(0.042570392907962495) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(144.3254165649414) - actor/lr:np.float64(1e-06) - training/global_step:24 - training/epoch:0 - critic/score/mean:0.875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0007204003632068634 - critic/advantages/max:0.5 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0007204003632068634 - critic/returns/max:0.5 - critic/returns/min:-0.875 - response_length/mean:347.70703125 - response_length/max:768.0 - re
|
||
step:25 - global_seqlen/min:17336 - global_seqlen/max:25737 - global_seqlen/minmax_diff:8401 - global_seqlen/balanced_min:22469 - global_seqlen/balanced_max:22472 - global_seqlen/mean:22471.25 - actor/entropy:0.052130524069070816 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9925975203514099 - training/rollout_probs_diff_mean:0.0023709130473434925 - training/rollout_probs_diff_std:0.013253876008093357 - training/rollout_actor_probs_pearson_corr:0.9941772222518921 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71875 - 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:0.71875 - rollout_corr/rollout_is_mean:0.9997007846832275 - rollout_corr/rollout_is_ratio_fraction_high:1.2528502338682301e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00016287053585983813 - rollout_corr/rollout_is_max:2.049886465072632 - rollout_corr/rollout_is_min:0.0007144523551687598 - rollout_corr/rollout_is_std:0.03154897689819336 - rollout_corr/rollout_is_eff_sample_size:0.9990051163742315 - rollout_corr/rollout_is_seq_mean:0.999676525592804 - rollout_corr/rollout_is_seq_std:0.002087613334879279 - rollout_corr/rollout_is_seq_max:1.0080496072769165 - rollout_corr/rollout_is_seq_min:0.9920728206634521 - rollout_corr/rollout_is_seq_max_deviation:0.008049607276916504 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0590065233409405) - rollout_corr/training_log_ppl:np.float64(0.05567418575810734) - rollout_corr/kl:np.float64(0.0014148961926281345) - rollout_corr/k3_kl:np.float64(0.0010043128627899023) - rollout_corr/rollout_ppl:np.float64(1.057496966328472) - rollout_corr/rollout_log_ppl:np.float64(0.054259289096080465) - rollout_corr/log_ppl_diff:np.float64(0.0014148966620268766) - rollout_corr/log_ppl_abs_diff:np.float64(0.002337703506782418) - rollout_corr/log_ppl_diff_max:np.float64(0.025214478373527527) - rollout_corr/log_ppl_diff_min:np.float64(-0.007557865232229233) - rollout_corr/ppl_ratio:np.float64(1.0014220480807126) - rollout_corr/chi2_token:np.float64(0.0008356994949281216) - rollout_corr/chi2_seq:np.float64(0.20744728250429034) - actor/pg_loss:np.float64(0.0009773996425792575) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005361891062420909) - actor/ppo_kl:np.float64(0.0004331388923333179) - 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(0.71875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.64453125) - self_distillation/token_reweight_w_mean:np.float64(198215.93650202313) - self_distillation/token_reweight_w_std:np.float64(2314051.2728974866) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000367683591321) - self_distillation/token_reweight_w_clip_frac:np.float64(0.017195056650962215) - self_distillation/empty_target_batch:np.float64(0.28125) - actor/grad_norm:np.float64(0.28173942118883133) - perf/mfu/actor:np.float64(0.04129532551700841) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(144.3085479736328) - actor/lr:np.float64(1e-06) - training/global_step:25 - 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:0.0005011400789953768 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.0005011400789953768 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:311.7890625 - res
|
||
step:26 - global_seqlen/min:21815 - global_seqlen/max:25977 - global_seqlen/minmax_diff:4162 - global_seqlen/balanced_min:23256 - global_seqlen/balanced_max:23259 - global_seqlen/mean:23257.375 - actor/entropy:0.05010559782385826 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9880870580673218 - training/rollout_probs_diff_mean:0.002416659612208605 - training/rollout_probs_diff_std:0.013399409130215645 - training/rollout_actor_probs_pearson_corr:0.9939301013946533 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.74609375 - 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:0.74609375 - rollout_corr/rollout_is_mean:0.9997743964195251 - rollout_corr/rollout_is_ratio_fraction_high:3.553029091563076e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00023686859640292823 - rollout_corr/rollout_is_max:2.335609197616577 - rollout_corr/rollout_is_min:0.006727981381118298 - rollout_corr/rollout_is_std:0.03265010192990303 - rollout_corr/rollout_is_eff_sample_size:0.9989349277027401 - rollout_corr/rollout_is_seq_mean:0.9997029304504395 - rollout_corr/rollout_is_seq_std:0.001995098777115345 - rollout_corr/rollout_is_seq_max:1.0078387260437012 - rollout_corr/rollout_is_seq_min:0.9932174682617188 - rollout_corr/rollout_is_seq_max_deviation:0.007838726043701172 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0507985781878233) - rollout_corr/training_log_ppl:np.float64(0.04918104125681566) - rollout_corr/kl:np.float64(0.0010938695645918983) - rollout_corr/k3_kl:np.float64(0.00082757380827303) - rollout_corr/rollout_ppl:np.float64(1.04963733907789) - rollout_corr/rollout_log_ppl:np.float64(0.04808717123887618) - rollout_corr/log_ppl_diff:np.float64(0.00109387001793948) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019342835767020006) - rollout_corr/log_ppl_diff_max:np.float64(0.013077668845653534) - rollout_corr/log_ppl_diff_min:np.float64(-0.004524499177932739) - rollout_corr/ppl_ratio:np.float64(1.0010978260543197) - rollout_corr/chi2_token:np.float64(0.0010612753685563803) - rollout_corr/chi2_seq:np.float64(0.26944553572684526) - actor/pg_loss:np.float64(0.0011050777975469828) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00036502554121398134) - actor/ppo_kl:np.float64(0.00015411552791277927) - 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(0.66796875) - self_distillation/token_reweight_w_mean:np.float64(148665.91016933904) - self_distillation/token_reweight_w_std:np.float64(2118637.270728631) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0002983210142702) - self_distillation/token_reweight_w_clip_frac:np.float64(0.018292905369889922) - self_distillation/empty_target_batch:np.float64(0.25390625) - actor/grad_norm:np.float64(0.3034274894744158) - perf/mfu/actor:np.float64(0.041829506814014934) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(144.37565231323242) - actor/lr:np.float64(1e-06) - training/global_step:26 - 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:-0.00418517179787159 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00418517179787159 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:329.82421
|
||
step:27 - global_seqlen/min:19833 - global_seqlen/max:25283 - global_seqlen/minmax_diff:5450 - global_seqlen/balanced_min:22907 - global_seqlen/balanced_max:22913 - global_seqlen/mean:22912.0 - actor/entropy:0.042662691324949265 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.8065875172615051 - training/rollout_probs_diff_mean:0.0022665185388177633 - training/rollout_probs_diff_std:0.01338112261146307 - training/rollout_actor_probs_pearson_corr:0.9929261803627014 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.875 - self_distillation/correct_sample_fraction:0.7890625 - 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:0.9998301267623901 - rollout_corr/rollout_is_ratio_fraction_high:3.683331669890322e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0002455554495099932 - rollout_corr/rollout_is_max:3.015857458114624 - rollout_corr/rollout_is_min:0.013438464142382145 - rollout_corr/rollout_is_std:0.0322062112390995 - rollout_corr/rollout_is_eff_sample_size:0.99896335887444 - rollout_corr/rollout_is_seq_mean:0.9998033046722412 - rollout_corr/rollout_is_seq_std:0.002107642823830247 - rollout_corr/rollout_is_seq_max:1.0079267024993896 - rollout_corr/rollout_is_seq_min:0.9947221279144287 - rollout_corr/rollout_is_seq_max_deviation:0.007926702499389648 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.044278597459197) - rollout_corr/training_log_ppl:np.float64(0.04298725330590969) - rollout_corr/kl:np.float64(0.0008925169439546465) - rollout_corr/k3_kl:np.float64(0.0006722692353946513) - rollout_corr/rollout_ppl:np.float64(1.0433417870663106) - rollout_corr/rollout_log_ppl:np.float64(0.04209473588707624) - rollout_corr/log_ppl_diff:np.float64(0.0008925174188334495) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019630308088380843) - rollout_corr/log_ppl_diff_max:np.float64(0.011757582426071167) - rollout_corr/log_ppl_diff_min:np.float64(-0.006372295320034027) - rollout_corr/ppl_ratio:np.float64(1.000895933015272) - rollout_corr/chi2_token:np.float64(0.0008611760567873716) - rollout_corr/chi2_seq:np.float64(0.5474486453458667) - actor/pg_loss:np.float64(0.0006962757324799895) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002548012948864198) - actor/ppo_kl:np.float64(5.430080918955582e-05) - actor/pg_clipfrac_lower:np.float64(1.073145631380612e-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(0.875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.7890625) - self_distillation/token_reweight_w_mean:np.float64(212501.34884866723) - self_distillation/token_reweight_w_std:np.float64(2760884.2314605396) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0005110597703606) - self_distillation/token_reweight_w_clip_frac:np.float64(0.012378365972836036) - self_distillation/empty_target_batch:np.float64(0.125) - actor/grad_norm:np.float64(0.23239875957369804) - perf/mfu/actor:np.float64(0.04186334349800554) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.43539905548096) - actor/lr:np.float64(1e-06) - training/global_step:27 - training/epoch:1 - critic/score/mean:0.7890625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7890625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-2.1486101104528643e-05 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-2.1486101104528643e-05 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:318.15625 - respon
|
||
step:28 - global_seqlen/min:18564 - global_seqlen/max:25860 - global_seqlen/minmax_diff:7296 - global_seqlen/balanced_min:22067 - global_seqlen/balanced_max:22074 - global_seqlen/mean:22070.75 - actor/entropy:0.050707269459962845 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6709722280502319 - training/rollout_probs_diff_mean:0.0023314820136874914 - training/rollout_probs_diff_std:0.013388410210609436 - training/rollout_actor_probs_pearson_corr:0.99385005235672 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.7109375 - 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:0.7109375 - rollout_corr/rollout_is_mean:0.9999811053276062 - rollout_corr/rollout_is_ratio_fraction_high:7.481483044102788e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00023691363458056003 - rollout_corr/rollout_is_max:3.443509578704834 - rollout_corr/rollout_is_min:0.28613391518592834 - rollout_corr/rollout_is_std:0.03239627555012703 - rollout_corr/rollout_is_eff_sample_size:0.9989513438303744 - rollout_corr/rollout_is_seq_mean:1.000024676322937 - rollout_corr/rollout_is_seq_std:0.0022317811381071806 - rollout_corr/rollout_is_seq_max:1.0114712715148926 - rollout_corr/rollout_is_seq_min:0.9920425415039062 - rollout_corr/rollout_is_seq_max_deviation:0.011471271514892578 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0535097629763186) - rollout_corr/training_log_ppl:np.float64(0.051123634291798226) - rollout_corr/kl:np.float64(0.0006726607275737706) - rollout_corr/k3_kl:np.float64(0.0008511580656858087) - rollout_corr/rollout_ppl:np.float64(1.0527968504466116) - rollout_corr/rollout_log_ppl:np.float64(0.05045097289439582) - rollout_corr/log_ppl_diff:np.float64(0.0006726613974024076) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018435009587847162) - rollout_corr/log_ppl_diff_max:np.float64(0.012318253517150879) - rollout_corr/log_ppl_diff_min:np.float64(-0.0074674468487501144) - rollout_corr/ppl_ratio:np.float64(1.0006759993266314) - rollout_corr/chi2_token:np.float64(0.0040700845420360565) - rollout_corr/chi2_seq:np.float64(0.6581041102763265) - actor/pg_loss:np.float64(0.0009383222786709666) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00029797435490763746) - actor/ppo_kl:np.float64(9.442604779663633e-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(0.63671875) - self_distillation/token_reweight_w_mean:np.float64(150508.01074047922) - self_distillation/token_reweight_w_std:np.float64(2154139.784913294) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0003592369612306) - self_distillation/token_reweight_w_clip_frac:np.float64(0.013951479126262711) - self_distillation/empty_target_batch:np.float64(0.2890625) - actor/grad_norm:np.float64(0.1988047994673252) - perf/mfu/actor:np.float64(0.03959577009416985) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.41423416137695) - actor/lr:np.float64(1e-06) - training/global_step:28 - 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:0.0020215590484440327 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0020215590484440327 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:313.27
|
||
step:29 - global_seqlen/min:18921 - global_seqlen/max:27487 - global_seqlen/minmax_diff:8566 - global_seqlen/balanced_min:22991 - global_seqlen/balanced_max:22998 - global_seqlen/mean:22995.25 - actor/entropy:0.039533860981464386 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7222251296043396 - training/rollout_probs_diff_mean:0.0019475192530080676 - training/rollout_probs_diff_std:0.011736846528947353 - training/rollout_actor_probs_pearson_corr:0.9941072463989258 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.7109375 - self_distillation/correct_sample_fraction:0.6640625 - 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:0.9999589920043945 - rollout_corr/rollout_is_ratio_fraction_high:3.7319932744139805e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00016171969764400274 - rollout_corr/rollout_is_max:2.6647143363952637 - rollout_corr/rollout_is_min:0.116624616086483 - rollout_corr/rollout_is_std:0.02993287518620491 - rollout_corr/rollout_is_eff_sample_size:0.9991047060399887 - rollout_corr/rollout_is_seq_mean:0.9999057054519653 - rollout_corr/rollout_is_seq_std:0.0018940429436042905 - rollout_corr/rollout_is_seq_max:1.0079894065856934 - rollout_corr/rollout_is_seq_min:0.9945310950279236 - rollout_corr/rollout_is_seq_max_deviation:0.00798940658569336 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.040629532188177) - rollout_corr/training_log_ppl:np.float64(0.03935588929562073) - rollout_corr/kl:np.float64(0.0008194791879141405) - rollout_corr/k3_kl:np.float64(0.0005604395872609302) - rollout_corr/rollout_ppl:np.float64(1.0397686720825732) - rollout_corr/rollout_log_ppl:np.float64(0.03853640963188809) - rollout_corr/log_ppl_diff:np.float64(0.0008194796637326363) - rollout_corr/log_ppl_abs_diff:np.float64(0.001599156311385741) - rollout_corr/log_ppl_diff_max:np.float64(0.009142909198999405) - rollout_corr/log_ppl_diff_min:np.float64(-0.004533983767032623) - rollout_corr/ppl_ratio:np.float64(1.000821965513751) - rollout_corr/chi2_token:np.float64(0.0006333421915769577) - rollout_corr/chi2_seq:np.float64(0.12789275567047298) - actor/pg_loss:np.float64(0.0002092909999191761) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(8.36119579616934e-05) - actor/ppo_kl:np.float64(0.0002650999025854972) - actor/pg_clipfrac_lower:np.float64(9.717039574752562e-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(0.7109375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.7109375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6640625) - self_distillation/token_reweight_w_mean:np.float64(155151.15066527855) - self_distillation/token_reweight_w_std:np.float64(1819004.7734375) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.00056552561) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0050811002001864836) - self_distillation/empty_target_batch:np.float64(0.2890625) - actor/grad_norm:np.float64(0.12285954970866442) - perf/mfu/actor:np.float64(0.042161963586262195) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.6112823486328) - actor/lr:np.float64(1e-06) - training/global_step:29 - training/epoch:1 - critic/score/mean:0.6640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0005940088885836303 - critic/advantages/max:0.875 - critic/advantages/min:-0.125 - critic/returns/mean:-0.0005940088885836303 - critic/returns/max:0.875 - critic/returns/min:-0.125 - response_length/mean
|
||
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
|
||
Given the following crystal structure parameters for the material RbLa9(IrO6)4 (Material ID: mp-560657), calculate the volume of the unit cell (in ų).
|
||
|
||
Lattice Parameters:
|
||
|
||
a = 7.82 Å
|
||
b = 7.82 Å
|
||
c = 17.88 Å
|
||
α = 90.00º
|
||
β = 90.00º
|
||
γ = 90.00º
|
||
|
||
|
||
|
||
A: 1025.67 ų
|
||
|
||
B: 1094.31 ų
|
||
|
||
C: 1200.45 ų
|
||
|
||
D: 1150.78 ų
|
||
|
||
Please reason step by step.
|
||
assistant
|
||
<think>
|
||
|
||
</think>
|
||
|
||
|
||
[response] <reasoning>
|
||
The volume of a unit cell in a crystal with orthogonal lattice parameters (α = β = γ = 90.00º) is calculated using the formula:
|
||
|
||
V = a × b × c
|
||
|
||
Given:
|
||
a = 7.82 Å
|
||
b = 7.82 Å
|
||
c = 17.88 Å
|
||
|
||
Substituting the values:
|
||
|
||
V = 7.82 × 7.82 × 17.88
|
||
|
||
First, calculate 7.82 × 7.82:
|
||
7.82 × 7.82 ≈ 61.1524
|
||
|
||
Then, multiply by 17.88:
|
||
61.1524 × 17.88 ≈ 1094.31
|
||
|
||
Thus, the volume of the unit cell is approximately 1094.31 ų.
|
||
</reasoning>
|
||
<answer>
|
||
B
|
||
</answer>
|
||
[ground_truth] B
|
||
[score] 1.0
|
||
[acc] 1.0
|
||
[pred] B
|
||
[incorrect_format] 1
|
||
[feedback]
|
||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-4B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_30
|
||
step:30 - global_seqlen/min:19442 - global_seqlen/max:27088 - global_seqlen/minmax_diff:7646 - global_seqlen/balanced_min:23922 - global_seqlen/balanced_max:23932 - global_seqlen/mean:23928.875 - actor/entropy:0.04492096230387688 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7854477167129517 - training/rollout_probs_diff_mean:0.0021507125347852707 - training/rollout_probs_diff_std:0.013460204005241394 - training/rollout_actor_probs_pearson_corr:0.9931148886680603 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65234375 - 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:0.65234375 - rollout_corr/rollout_is_mean:1.0000746250152588 - rollout_corr/rollout_is_ratio_fraction_high:9.3432841822505e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0002335821045562625 - rollout_corr/rollout_is_max:2.5501413345336914 - rollout_corr/rollout_is_min:0.10825343430042267 - rollout_corr/rollout_is_std:0.03233734518289566 - rollout_corr/rollout_is_eff_sample_size:0.9989556263871889 - rollout_corr/rollout_is_seq_mean:1.0000686645507812 - rollout_corr/rollout_is_seq_std:0.0021643403451889753 - rollout_corr/rollout_is_seq_max:1.0080591440200806 - rollout_corr/rollout_is_seq_min:0.9934688806533813 - rollout_corr/rollout_is_seq_max_deviation:0.008059144020080566 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0462839058600366) - rollout_corr/training_log_ppl:np.float64(0.044393875357854995) - rollout_corr/kl:np.float64(0.0005853226012781931) - rollout_corr/k3_kl:np.float64(0.00065308902737371) - rollout_corr/rollout_ppl:np.float64(1.045660697389394) - rollout_corr/rollout_log_ppl:np.float64(0.04380855248609805) - rollout_corr/log_ppl_diff:np.float64(0.0005853228717569436) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016660557507748308) - rollout_corr/log_ppl_diff_max:np.float64(0.022737577557563782) - rollout_corr/log_ppl_diff_min:np.float64(-0.0062180086970329285) - rollout_corr/ppl_ratio:np.float64(1.0005887241568416) - rollout_corr/chi2_token:np.float64(0.0014040102250874043) - rollout_corr/chi2_seq:np.float64(0.52653125859797) - actor/pg_loss:np.float64(0.0013053459115326405) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00032061494812296587) - actor/ppo_kl:np.float64(7.5638094044983e-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(0.65234375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.65234375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.55078125) - self_distillation/token_reweight_w_mean:np.float64(151288.02862975863) - self_distillation/token_reweight_w_std:np.float64(1988820.3295037607) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0011276304721832) - self_distillation/token_reweight_w_clip_frac:np.float64(0.017992489927564748) - self_distillation/empty_target_batch:np.float64(0.34765625) - actor/grad_norm:np.float64(0.24706468358635902) - perf/mfu/actor:np.float64(0.0430726144645511) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.65000915527344) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7253989361702128) - val-aux/sciknoweval/reward/std@16:np.float64(0.08655772986262227) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7613191489361701) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.05961052684599058) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6877127659574468) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.08852506830517469) - val-aux/scikn
|
||
step:31 - global_seqlen/min:21305 - global_seqlen/max:25291 - global_seqlen/minmax_diff:3986 - global_seqlen/balanced_min:23036 - global_seqlen/balanced_max:23041 - global_seqlen/mean:23039.625 - actor/entropy:0.04938299581408501 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9892370104789734 - training/rollout_probs_diff_mean:0.002258756896480918 - training/rollout_probs_diff_std:0.013091050088405609 - training/rollout_actor_probs_pearson_corr:0.9941160082817078 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71484375 - 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:0.71484375 - rollout_corr/rollout_is_mean:1.0000802278518677 - rollout_corr/rollout_is_ratio_fraction_high:1.2028917808493134e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0002165205223718658 - rollout_corr/rollout_is_max:2.550828695297241 - rollout_corr/rollout_is_min:0.0067202127538621426 - rollout_corr/rollout_is_std:0.03224320337176323 - rollout_corr/rollout_is_eff_sample_size:0.9989616934055325 - rollout_corr/rollout_is_seq_mean:1.0001165866851807 - rollout_corr/rollout_is_seq_std:0.00201058154925704 - rollout_corr/rollout_is_seq_max:1.0065182447433472 - rollout_corr/rollout_is_seq_min:0.9951715469360352 - rollout_corr/rollout_is_seq_max_deviation:0.006518244743347168 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0496189133264124) - rollout_corr/training_log_ppl:np.float64(0.047763853480319085) - rollout_corr/kl:np.float64(0.0005609981700303024) - rollout_corr/k3_kl:np.float64(0.0007823906714463646) - rollout_corr/rollout_ppl:np.float64(1.0490215150639415) - rollout_corr/rollout_log_ppl:np.float64(0.047202855099840235) - rollout_corr/log_ppl_diff:np.float64(0.0005609983804788499) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018915072992058413) - rollout_corr/log_ppl_diff_max:np.float64(0.018269948661327362) - rollout_corr/log_ppl_diff_min:np.float64(-0.009728636592626572) - rollout_corr/ppl_ratio:np.float64(1.0005653724074364) - rollout_corr/chi2_token:np.float64(0.001830502413213253) - rollout_corr/chi2_seq:np.float64(1.5425163942854851) - actor/pg_loss:np.float64(0.0014576166868209839) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00036924146979799843) - actor/ppo_kl:np.float64(6.366286959824663e-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(0.71484375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71484375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6328125) - self_distillation/token_reweight_w_mean:np.float64(209419.71391360858) - self_distillation/token_reweight_w_std:np.float64(2615016.670161984) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000283203087747) - self_distillation/token_reweight_w_clip_frac:np.float64(0.018031955318292603) - self_distillation/empty_target_batch:np.float64(0.28515625) - actor/grad_norm:np.float64(0.28416722267866135) - perf/mfu/actor:np.float64(0.04234140442013515) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.89453125) - actor/lr:np.float64(1e-06) - training/global_step:31 - training/epoch:1 - 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:0.001890043611638248 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.001890043611638248 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:324.73828125
|
||
step:32 - global_seqlen/min:18196 - global_seqlen/max:26796 - global_seqlen/minmax_diff:8600 - global_seqlen/balanced_min:22588 - global_seqlen/balanced_max:22596 - global_seqlen/mean:22594.125 - actor/entropy:0.04554832726716995 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.507658839225769 - training/rollout_probs_diff_mean:0.002093696966767311 - training/rollout_probs_diff_std:0.011700347065925598 - training/rollout_actor_probs_pearson_corr:0.9949268102645874 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71875 - 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:0.71875 - rollout_corr/rollout_is_mean:1.0000239610671997 - rollout_corr/rollout_is_ratio_fraction_high:6.305408896878362e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00012610817793756723 - rollout_corr/rollout_is_max:3.8349814414978027 - rollout_corr/rollout_is_min:0.2868506908416748 - rollout_corr/rollout_is_std:0.03249547630548477 - rollout_corr/rollout_is_eff_sample_size:0.9989452769377425 - rollout_corr/rollout_is_seq_mean:1.0000345706939697 - rollout_corr/rollout_is_seq_std:0.0022742771543562412 - rollout_corr/rollout_is_seq_max:1.0197118520736694 - rollout_corr/rollout_is_seq_min:0.9935188293457031 - rollout_corr/rollout_is_seq_max_deviation:0.019711852073669434 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0453667039982975) - rollout_corr/training_log_ppl:np.float64(0.04353243351852143) - rollout_corr/kl:np.float64(0.0006469867601166523) - rollout_corr/k3_kl:np.float64(0.0006021329918155516) - rollout_corr/rollout_ppl:np.float64(1.0446764542721212) - rollout_corr/rollout_log_ppl:np.float64(0.04288544589826415) - rollout_corr/log_ppl_diff:np.float64(0.0006469876202572777) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016619250986877887) - rollout_corr/log_ppl_diff_max:np.float64(0.00940629094839096) - rollout_corr/log_ppl_diff_min:np.float64(-0.009224757552146912) - rollout_corr/ppl_ratio:np.float64(1.0006496729329228) - rollout_corr/chi2_token:np.float64(0.0011666147038340569) - rollout_corr/chi2_seq:np.float64(0.5588384328875691) - actor/pg_loss:np.float64(0.0007467931136488914) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00025311750505352393) - actor/ppo_kl:np.float64(0.00014509259286654697) - 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(0.71875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6875) - self_distillation/token_reweight_w_mean:np.float64(25800.459359957604) - self_distillation/token_reweight_w_std:np.float64(407525.20962225855) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9998260058928281) - self_distillation/token_reweight_w_clip_frac:np.float64(0.00925165212538559) - self_distillation/empty_target_batch:np.float64(0.28125) - actor/grad_norm:np.float64(0.14854632876813412) - perf/mfu/actor:np.float64(0.041973087574399694) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.82848739624023) - actor/lr:np.float64(1e-06) - training/global_step:32 - 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:0.0028468919917941093 - critic/advantages/max:0.5 - critic/advantages/min:-0.75 - critic/returns/mean:0.0028468919917941093 - critic/returns/max:0.5 - critic/returns/min:-0.75 - response_length/mean:309.75390625 - response_length/max:73
|
||
step:33 - global_seqlen/min:18392 - global_seqlen/max:24737 - global_seqlen/minmax_diff:6345 - global_seqlen/balanced_min:22379 - global_seqlen/balanced_max:22384 - global_seqlen/mean:22382.5 - actor/entropy:0.048572197556495667 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6976189017295837 - training/rollout_probs_diff_mean:0.002187278587371111 - training/rollout_probs_diff_std:0.013085621409118176 - training/rollout_actor_probs_pearson_corr:0.9939641952514648 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.8671875 - self_distillation/correct_sample_fraction:0.74609375 - 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:1.0001271963119507 - rollout_corr/rollout_is_ratio_fraction_high:9.94480651570484e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00023618915292900056 - rollout_corr/rollout_is_max:3.2328717708587646 - rollout_corr/rollout_is_min:0.02593221887946129 - rollout_corr/rollout_is_std:0.03340000659227371 - rollout_corr/rollout_is_eff_sample_size:0.9988860393883612 - rollout_corr/rollout_is_seq_mean:1.000138759613037 - rollout_corr/rollout_is_seq_std:0.0020758023019880056 - rollout_corr/rollout_is_seq_max:1.0095303058624268 - rollout_corr/rollout_is_seq_min:0.9938346147537231 - rollout_corr/rollout_is_seq_max_deviation:0.009530305862426758 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0498127928003669) - rollout_corr/training_log_ppl:np.float64(0.04764294929736934) - rollout_corr/kl:np.float64(0.000776656330423009) - rollout_corr/k3_kl:np.float64(0.0008650090900808038) - rollout_corr/rollout_ppl:np.float64(1.048987964168191) - rollout_corr/rollout_log_ppl:np.float64(0.04686629328352865) - rollout_corr/log_ppl_diff:np.float64(0.000776656013840693) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018832581390597625) - rollout_corr/log_ppl_diff_max:np.float64(0.01899798959493637) - rollout_corr/log_ppl_diff_min:np.float64(-0.010233291424810886) - rollout_corr/ppl_ratio:np.float64(1.00078120874241) - rollout_corr/chi2_token:np.float64(0.0018827361054718494) - rollout_corr/chi2_seq:np.float64(0.43429425405338407) - actor/pg_loss:np.float64(0.0005226503126323223) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00025913522904374986) - actor/ppo_kl:np.float64(0.0002559889147786265) - 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(0.74609375) - self_distillation/token_reweight_w_mean:np.float64(285482.3281259318) - self_distillation/token_reweight_w_std:np.float64(3867517.0114729353) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0010869828984141) - self_distillation/token_reweight_w_clip_frac:np.float64(0.01669350147858495) - self_distillation/empty_target_batch:np.float64(0.1328125) - actor/grad_norm:np.float64(0.27803053706884384) - perf/mfu/actor:np.float64(0.0417407421312785) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.80722427368164) - actor/lr:np.float64(1e-06) - training/global_step:33 - training/epoch:1 - critic/score/mean:0.74609375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.74609375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0030891054775565863 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.0030891054775565863 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:314.234375 - respo
|
||
step:34 - global_seqlen/min:21779 - global_seqlen/max:25692 - global_seqlen/minmax_diff:3913 - global_seqlen/balanced_min:23445 - global_seqlen/balanced_max:23465 - global_seqlen/mean:23448.125 - actor/entropy:0.04247506335377693 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3394765853881836 - training/rollout_probs_diff_mean:0.0020800086203962564 - training/rollout_probs_diff_std:0.011497821658849716 - training/rollout_actor_probs_pearson_corr:0.9946447610855103 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.87109375 - self_distillation/correct_sample_fraction:0.8125 - 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:1.000171422958374 - rollout_corr/rollout_is_ratio_fraction_high:7.357179129030555e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0001471435825806111 - rollout_corr/rollout_is_max:9.17504596710205 - rollout_corr/rollout_is_min:0.22716903686523438 - rollout_corr/rollout_is_std:0.031074028462171555 - rollout_corr/rollout_is_eff_sample_size:0.9990353362293889 - rollout_corr/rollout_is_seq_mean:1.0002288818359375 - rollout_corr/rollout_is_seq_std:0.002648404100909829 - rollout_corr/rollout_is_seq_max:1.029280662536621 - rollout_corr/rollout_is_seq_min:0.9955718517303467 - rollout_corr/rollout_is_seq_max_deviation:0.029280662536621094 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0434476546943188) - rollout_corr/training_log_ppl:np.float64(0.04219085097838615) - rollout_corr/kl:np.float64(0.0005611525840052423) - rollout_corr/k3_kl:np.float64(0.000653065330164182) - rollout_corr/rollout_ppl:np.float64(1.0428576031699777) - rollout_corr/rollout_log_ppl:np.float64(0.041629698092947365) - rollout_corr/log_ppl_diff:np.float64(0.000561152885438787) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015396711655739637) - rollout_corr/log_ppl_diff_max:np.float64(0.009032197296619415) - rollout_corr/log_ppl_diff_min:np.float64(-0.006443668156862259) - rollout_corr/ppl_ratio:np.float64(1.000563275301829) - rollout_corr/chi2_token:np.float64(0.001628566300496459) - rollout_corr/chi2_seq:np.float64(0.45253721135668457) - actor/pg_loss:np.float64(0.0004543724935501814) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.000334290034516016) - actor/ppo_kl:np.float64(0.00019341592874866365) - 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(0.8125) - self_distillation/token_reweight_w_mean:np.float64(151244.62870255834) - self_distillation/token_reweight_w_std:np.float64(1700770.1650854703) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9998153406195343) - self_distillation/token_reweight_w_clip_frac:np.float64(0.011545672423380893) - self_distillation/empty_target_batch:np.float64(0.12890625) - actor/grad_norm:np.float64(0.24764136131852865) - perf/mfu/actor:np.float64(0.043330077101173106) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.78245162963867) - actor/lr:np.float64(1e-06) - training/global_step:34 - training/epoch:1 - critic/score/mean:0.8125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.8125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0014346498064696789 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0014346498064696789 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:318.56640625 - respons
|
||
step:35 - global_seqlen/min:20095 - global_seqlen/max:24523 - global_seqlen/minmax_diff:4428 - global_seqlen/balanced_min:21778 - global_seqlen/balanced_max:21780 - global_seqlen/mean:21779.25 - actor/entropy:0.04991421476006508 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.48201340436935425 - training/rollout_probs_diff_mean:0.002308846917003393 - training/rollout_probs_diff_std:0.012316430918872356 - training/rollout_actor_probs_pearson_corr:0.9947492480278015 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - 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:0.8125 - rollout_corr/rollout_is_mean:1.0002334117889404 - rollout_corr/rollout_is_ratio_fraction_high:1.3196442523621954e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00013196442159824073 - rollout_corr/rollout_is_max:3.959545135498047 - rollout_corr/rollout_is_min:0.29720374941825867 - rollout_corr/rollout_is_std:0.0317457914352417 - rollout_corr/rollout_is_eff_sample_size:0.9989936951727617 - rollout_corr/rollout_is_seq_mean:1.0001981258392334 - rollout_corr/rollout_is_seq_std:0.0018588469829410315 - rollout_corr/rollout_is_seq_max:1.0078638792037964 - rollout_corr/rollout_is_seq_min:0.9947773814201355 - rollout_corr/rollout_is_seq_max_deviation:0.007863879203796387 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0496695381589234) - rollout_corr/training_log_ppl:np.float64(0.048008743393438635) - rollout_corr/kl:np.float64(0.0024177409742709155) - rollout_corr/k3_kl:np.float64(0.002347025056231189) - rollout_corr/rollout_ppl:np.float64(1.0471166698262095) - rollout_corr/rollout_log_ppl:np.float64(0.045591002022774774) - rollout_corr/log_ppl_diff:np.float64(0.002417741381577798) - rollout_corr/log_ppl_abs_diff:np.float64(0.003363729747434263) - rollout_corr/log_ppl_diff_max:np.float64(0.10936760157346725) - rollout_corr/log_ppl_diff_min:np.float64(-0.0114052165299654) - rollout_corr/ppl_ratio:np.float64(1.0024650054983795) - rollout_corr/chi2_token:np.float64(0.004743414698168635) - rollout_corr/chi2_seq:np.float64(0.5235199017915875) - actor/pg_loss:np.float64(0.0007961411029100418) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005753537584496371) - actor/ppo_kl:np.float64(0.002120418961538917) - actor/pg_clipfrac_lower:np.float64(2.271075572934933e-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(0.765625) - self_distillation/token_reweight_w_mean:np.float64(210541.37167732976) - self_distillation/token_reweight_w_std:np.float64(1888247.1145525659) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0001017071772367) - self_distillation/token_reweight_w_clip_frac:np.float64(0.008670316565257963) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.49630653485655785) - perf/mfu/actor:np.float64(0.039790707574217724) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.80047988891602) - actor/lr:np.float64(1e-06) - training/global_step:35 - 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:-0.0018722452223300934 - critic/advantages/max:0.5 - critic/advantages/min:-0.75 - critic/returns/mean:-0.0018722452223300934 - critic/returns/max:0.5 - critic/returns/min:-0.75 - response_length/mean:296.0078125 - respon
|
||
step:36 - global_seqlen/min:20820 - global_seqlen/max:26212 - global_seqlen/minmax_diff:5392 - global_seqlen/balanced_min:23515 - global_seqlen/balanced_max:23517 - global_seqlen/mean:23516.25 - actor/entropy:0.05420038849115372 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4744115471839905 - training/rollout_probs_diff_mean:0.002247673226520419 - training/rollout_probs_diff_std:0.011697757057845592 - training/rollout_actor_probs_pearson_corr:0.9956129789352417 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.7421875 - 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:0.7421875 - rollout_corr/rollout_is_mean:1.0001267194747925 - rollout_corr/rollout_is_ratio_fraction_high:6.95265262038447e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00015064080071169883 - rollout_corr/rollout_is_max:3.5453145503997803 - rollout_corr/rollout_is_min:0.3200256824493408 - rollout_corr/rollout_is_std:0.031670600175857544 - rollout_corr/rollout_is_eff_sample_size:0.9989980970632323 - rollout_corr/rollout_is_seq_mean:1.0001379251480103 - rollout_corr/rollout_is_seq_std:0.0017661351012066007 - rollout_corr/rollout_is_seq_max:1.0061100721359253 - rollout_corr/rollout_is_seq_min:0.9940493106842041 - rollout_corr/rollout_is_seq_max_deviation:0.006110072135925293 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.053559670690447) - rollout_corr/training_log_ppl:np.float64(0.051443011833725905) - rollout_corr/kl:np.float64(0.0007316862326405271) - rollout_corr/k3_kl:np.float64(0.0007606058412104044) - rollout_corr/rollout_ppl:np.float64(1.052789440844208) - rollout_corr/rollout_log_ppl:np.float64(0.05071132522425614) - rollout_corr/log_ppl_diff:np.float64(0.0007316866058317828) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018277808030688902) - rollout_corr/log_ppl_diff_max:np.float64(0.021980972960591316) - rollout_corr/log_ppl_diff_min:np.float64(-0.004337247461080551) - rollout_corr/ppl_ratio:np.float64(1.0007357993163168) - rollout_corr/chi2_token:np.float64(0.0014476962387561798) - rollout_corr/chi2_seq:np.float64(0.6039434908889234) - actor/pg_loss:np.float64(0.00037011189851909876) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00023756584459988517) - actor/ppo_kl:np.float64(0.0003662115519347253) - 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(0.66796875) - self_distillation/token_reweight_w_mean:np.float64(44200.58313611476) - self_distillation/token_reweight_w_std:np.float64(806335.131862045) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0007616139482707) - self_distillation/token_reweight_w_clip_frac:np.float64(0.00920598856464494) - self_distillation/empty_target_batch:np.float64(0.2578125) - actor/grad_norm:np.float64(0.11965901032090187) - perf/mfu/actor:np.float64(0.04255113735653425) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.83895874023438) - actor/lr:np.float64(1e-06) - training/global_step:36 - training/epoch:1 - 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:-4.3454077967908233e-05 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-4.3454077967908233e-05 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:337.1015
|
||
step:37 - global_seqlen/min:18243 - global_seqlen/max:24615 - global_seqlen/minmax_diff:6372 - global_seqlen/balanced_min:21724 - global_seqlen/balanced_max:21729 - global_seqlen/mean:21726.625 - actor/entropy:0.05331616848707199 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3394768238067627 - training/rollout_probs_diff_mean:0.0020236121490597725 - training/rollout_probs_diff_std:0.010713296942412853 - training/rollout_actor_probs_pearson_corr:0.9962490200996399 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.90625 - self_distillation/correct_sample_fraction:0.8359375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.90625 - rollout_corr/rollout_is_mean:1.0001013278961182 - rollout_corr/rollout_is_ratio_fraction_high:3.556651427061297e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00015412156062666327 - rollout_corr/rollout_is_max:3.9899837970733643 - rollout_corr/rollout_is_min:0.09489698708057404 - rollout_corr/rollout_is_std:0.02917863428592682 - rollout_corr/rollout_is_eff_sample_size:0.9991495695201904 - rollout_corr/rollout_is_seq_mean:1.0000089406967163 - rollout_corr/rollout_is_seq_std:0.002048109658062458 - rollout_corr/rollout_is_seq_max:1.0107276439666748 - rollout_corr/rollout_is_seq_min:0.9943196177482605 - rollout_corr/rollout_is_seq_max_deviation:0.010727643966674805 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0528704207390547) - rollout_corr/training_log_ppl:np.float64(0.05075585774739011) - rollout_corr/kl:np.float64(0.0006263373954951135) - rollout_corr/k3_kl:np.float64(0.0006165275314771179) - rollout_corr/rollout_ppl:np.float64(1.052202373277396) - rollout_corr/rollout_log_ppl:np.float64(0.05012952038759977) - rollout_corr/log_ppl_diff:np.float64(0.000626337359790341) - rollout_corr/log_ppl_abs_diff:np.float64(0.001629049289476825) - rollout_corr/log_ppl_diff_max:np.float64(0.012944355607032776) - rollout_corr/log_ppl_diff_min:np.float64(-0.012586958706378937) - rollout_corr/ppl_ratio:np.float64(1.0006291172467172) - rollout_corr/chi2_token:np.float64(0.0011920023243874311) - rollout_corr/chi2_seq:np.float64(0.7796085118316114) - actor/pg_loss:np.float64(0.0013440322363749146) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0001347399347650935) - actor/ppo_kl:np.float64(7.59294577008518e-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(0.90625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.90625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.8359375) - self_distillation/token_reweight_w_mean:np.float64(109273.2849380807) - self_distillation/token_reweight_w_std:np.float64(1662628.403000201) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000219531590119) - self_distillation/token_reweight_w_clip_frac:np.float64(0.015951965760905296) - self_distillation/empty_target_batch:np.float64(0.09375) - actor/grad_norm:np.float64(0.24890022352337837) - perf/mfu/actor:np.float64(0.03961164662388422) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.77448272705078) - actor/lr:np.float64(1e-06) - training/global_step:37 - training/epoch:1 - critic/score/mean:0.8359375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.8359375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0019176279893144965 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0019176279893144965 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:329.48828125 - response
|
||
step:38 - global_seqlen/min:19549 - global_seqlen/max:25117 - global_seqlen/minmax_diff:5568 - global_seqlen/balanced_min:23218 - global_seqlen/balanced_max:23221 - global_seqlen/mean:23219.875 - actor/entropy:0.05836797133088112 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6118535995483398 - training/rollout_probs_diff_mean:0.00232762866653502 - training/rollout_probs_diff_std:0.011830761097371578 - training/rollout_actor_probs_pearson_corr:0.9958423972129822 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.7109375 - self_distillation/correct_sample_fraction:0.5859375 - 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:0.9998059272766113 - rollout_corr/rollout_is_ratio_fraction_high:4.672078648582101e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00016352275270037353 - rollout_corr/rollout_is_max:2.6164438724517822 - rollout_corr/rollout_is_min:0.0921701192855835 - rollout_corr/rollout_is_std:0.031520623713731766 - rollout_corr/rollout_is_eff_sample_size:0.9990072578785806 - rollout_corr/rollout_is_seq_mean:0.9997893571853638 - rollout_corr/rollout_is_seq_std:0.0019738837145268917 - rollout_corr/rollout_is_seq_max:1.0098562240600586 - rollout_corr/rollout_is_seq_min:0.9930856823921204 - rollout_corr/rollout_is_seq_max_deviation:0.009856224060058594 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0563341127708554) - rollout_corr/training_log_ppl:np.float64(0.054111441248096526) - rollout_corr/kl:np.float64(0.0006969592394456825) - rollout_corr/k3_kl:np.float64(0.0005701098419429451) - rollout_corr/rollout_ppl:np.float64(1.0555989309214056) - rollout_corr/rollout_log_ppl:np.float64(0.053414481904837885) - rollout_corr/log_ppl_diff:np.float64(0.0006969593432586407) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015633855382475303) - rollout_corr/log_ppl_diff_max:np.float64(0.008656919002532959) - rollout_corr/log_ppl_diff_min:np.float64(-0.005964074283838272) - rollout_corr/ppl_ratio:np.float64(1.000699280994013) - rollout_corr/chi2_token:np.float64(0.000926322303712368) - rollout_corr/chi2_seq:np.float64(0.32595846615731716) - actor/pg_loss:np.float64(0.0011332343565300107) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003224353213227005) - actor/ppo_kl:np.float64(-6.34805843731101e-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(0.5859375) - self_distillation/token_reweight_w_mean:np.float64(184414.60750435363) - self_distillation/token_reweight_w_std:np.float64(2728702.5535712354) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0006198440678418) - self_distillation/token_reweight_w_clip_frac:np.float64(0.01655629460583441) - self_distillation/empty_target_batch:np.float64(0.2890625) - actor/grad_norm:np.float64(0.27920282259583473) - perf/mfu/actor:np.float64(0.042506767077513594) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.75471115112305) - actor/lr:np.float64(1e-06) - training/global_step:38 - training/epoch:1 - critic/score/mean:0.5859375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5859375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0024061205331236124 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.0024061205331236124 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:334.4335
|
||
step:39 - global_seqlen/min:20942 - global_seqlen/max:27210 - global_seqlen/minmax_diff:6268 - global_seqlen/balanced_min:23015 - global_seqlen/balanced_max:23017 - global_seqlen/mean:23015.625 - actor/entropy:0.047281794250011444 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5083141922950745 - training/rollout_probs_diff_mean:0.0020674800034612417 - training/rollout_probs_diff_std:0.011524595320224762 - training/rollout_actor_probs_pearson_corr:0.9951632022857666 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.78125 - self_distillation/correct_sample_fraction:0.75390625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.78125 - rollout_corr/rollout_is_mean:1.0000718832015991 - rollout_corr/rollout_is_ratio_fraction_high:3.527876106090844e-05 - rollout_corr/rollout_is_ratio_fraction_low:8.231710671680048e-05 - rollout_corr/rollout_is_max:2.6066808700561523 - rollout_corr/rollout_is_min:0.29022958874702454 - rollout_corr/rollout_is_std:0.030541982501745224 - rollout_corr/rollout_is_eff_sample_size:0.9990680566378604 - rollout_corr/rollout_is_seq_mean:1.0000585317611694 - rollout_corr/rollout_is_seq_std:0.001705444767139852 - rollout_corr/rollout_is_seq_max:1.0067297220230103 - rollout_corr/rollout_is_seq_min:0.9937018752098083 - rollout_corr/rollout_is_seq_max_deviation:0.006729722023010254 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.04436213709414) - rollout_corr/training_log_ppl:np.float64(0.04305056084012904) - rollout_corr/kl:np.float64(0.0005853749285122944) - rollout_corr/k3_kl:np.float64(0.0005738799877832435) - rollout_corr/rollout_ppl:np.float64(1.043735504616052) - rollout_corr/rollout_log_ppl:np.float64(0.04246518543368438) - rollout_corr/log_ppl_diff:np.float64(0.0005853754064446548) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016567986331210705) - rollout_corr/log_ppl_diff_max:np.float64(0.014873620122671127) - rollout_corr/log_ppl_diff_min:np.float64(-0.0048823282122612) - rollout_corr/ppl_ratio:np.float64(1.0005881253164262) - rollout_corr/chi2_token:np.float64(0.0010363354813307524) - rollout_corr/chi2_seq:np.float64(0.46266718837432563) - actor/pg_loss:np.float64(0.000453876331448555) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00011933857240364887) - actor/ppo_kl:np.float64(0.0001878124741701015) - 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(0.78125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.78125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.75390625) - self_distillation/token_reweight_w_mean:np.float64(77515.16053502122) - self_distillation/token_reweight_w_std:np.float64(965262.9837699123) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9998419873882085) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0069773956056451425) - self_distillation/empty_target_batch:np.float64(0.21875) - actor/grad_norm:np.float64(0.11174572631716728) - perf/mfu/actor:np.float64(0.042077119502450376) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.9169692993164) - actor/lr:np.float64(1e-06) - training/global_step:39 - training/epoch:1 - critic/score/mean:0.75390625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.75390625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0008996084216050804 - critic/advantages/max:0.375 - critic/advantages/min:-0.875 - critic/returns/mean:0.0008996084216050804 - critic/returns/max:0.375 - critic/returns/min:-0.875 - response_length/mean:332.17578125 - 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': 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
|
||
Given the following crystal structure parameters for the material RbLa9(IrO6)4 (Material ID: mp-560657), calculate the volume of the unit cell (in ų).
|
||
|
||
Lattice Parameters:
|
||
|
||
a = 7.82 Å
|
||
b = 7.82 Å
|
||
c = 17.88 Å
|
||
α = 90.00º
|
||
β = 90.00º
|
||
γ = 90.00º
|
||
|
||
|
||
|
||
A: 1025.67 ų
|
||
|
||
B: 1094.31 ų
|
||
|
||
C: 1200.45 ų
|
||
|
||
D: 1150.78 ų
|
||
|
||
Please reason step by step.
|
||
assistant
|
||
<think>
|
||
|
||
</think>
|
||
|
||
|
||
[response] <reasoning>
|
||
The volume of a unit cell in a crystal structure can be calculated using the formula:
|
||
|
||
$$ V = a \times b \times c \times \sqrt{1 + 2\cos\alpha\cos\beta\cos\gamma - \cos^2\alpha - \cos^2\beta - \cos^2\gamma} $$
|
||
|
||
However, since the given angles α, β, and γ are all 90.00º, which are right angles, the crystal structure is a rectangular prism (orthorhombic). In this case, the formula simplifies to:
|
||
|
||
$$ V = a \times b \times c $$
|
||
|
||
Given:
|
||
- $ a = 7.82 $ Å
|
||
- $ b = 7.82 $ Å
|
||
- $ c = 17.88 $ Å
|
||
|
||
Substituting these values into the formula:
|
||
|
||
$$ V = 7.82 \times 7.82 \times 17.88 $$
|
||
|
||
First, calculate $ 7.82 \times 7.82 $:
|
||
|
||
$$ 7.82 \times 7.82 = 61.1524 $$
|
||
|
||
Then multiply by 17.88:
|
||
|
||
$$ 61.1524 \times 17.88 \approx 1094.31 $$
|
||
|
||
Thus, the volume of the unit cell is approximately 1094.31 ų.
|
||
</reasoning>
|
||
<answer>
|
||
B
|
||
</answer>
|
||
[ground_truth] B
|
||
[score] 1.0
|
||
[acc] 1.0
|
||
[pred] B
|
||
[incorrect_format] 1
|
||
[feedback]
|
||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-4B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_40
|
||
step:40 - global_seqlen/min:18634 - global_seqlen/max:24321 - global_seqlen/minmax_diff:5687 - global_seqlen/balanced_min:22272 - global_seqlen/balanced_max:22276 - global_seqlen/mean:22274.625 - actor/entropy:0.0525180958211422 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9980729222297668 - training/rollout_probs_diff_mean:0.0023294829297810793 - training/rollout_probs_diff_std:0.014080265536904335 - training/rollout_actor_probs_pearson_corr:0.9934572577476501 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.84375 - self_distillation/correct_sample_fraction:0.8046875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.84375 - rollout_corr/rollout_is_mean:0.99977707862854 - rollout_corr/rollout_is_ratio_fraction_high:4.945292675984092e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00021017494145780802 - rollout_corr/rollout_is_max:3.0684378147125244 - rollout_corr/rollout_is_min:0.00033825141144916415 - rollout_corr/rollout_is_std:0.03151306137442589 - rollout_corr/rollout_is_eff_sample_size:0.9990076147968647 - rollout_corr/rollout_is_seq_mean:0.9997336268424988 - rollout_corr/rollout_is_seq_std:0.002147440332919359 - rollout_corr/rollout_is_seq_max:1.0140759944915771 - rollout_corr/rollout_is_seq_min:0.9931351542472839 - rollout_corr/rollout_is_seq_max_deviation:0.014075994491577148 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0521052996627986) - rollout_corr/training_log_ppl:np.float64(0.05022077127978264) - rollout_corr/kl:np.float64(0.001029036797591143) - rollout_corr/k3_kl:np.float64(0.0009319911036609341) - rollout_corr/rollout_ppl:np.float64(1.050995112862438) - rollout_corr/rollout_log_ppl:np.float64(0.04919173403959576) - rollout_corr/log_ppl_diff:np.float64(0.0010290372401868808) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019625763334261137) - rollout_corr/log_ppl_diff_max:np.float64(0.025801215320825577) - rollout_corr/log_ppl_diff_min:np.float64(-0.005766719579696655) - rollout_corr/ppl_ratio:np.float64(1.001035234425217) - rollout_corr/chi2_token:np.float64(0.0011496609076857567) - rollout_corr/chi2_seq:np.float64(0.39592411555349827) - actor/pg_loss:np.float64(0.0008833230240270495) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00013724410791837727) - actor/ppo_kl:np.float64(-2.584329278576547e-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(0.84375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.84375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.8046875) - self_distillation/token_reweight_w_mean:np.float64(89189.93140461063) - self_distillation/token_reweight_w_std:np.float64(1213222.806374929) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0000462993048131) - self_distillation/token_reweight_w_clip_frac:np.float64(0.008561011323763523) - self_distillation/empty_target_batch:np.float64(0.15625) - actor/grad_norm:np.float64(0.2542172442190349) - perf/mfu/actor:np.float64(0.04111878608146088) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.78087615966797) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7420212765957447) - val-aux/sciknoweval/reward/std@16:np.float64(0.04708054278965243) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7608510638297873) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.03635224696233118) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.7230425531914892) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.04401129297371908) - val-aux/sciknoweval/reward
|
||
step:41 - global_seqlen/min:21642 - global_seqlen/max:25611 - global_seqlen/minmax_diff:3969 - global_seqlen/balanced_min:23700 - global_seqlen/balanced_max:23701 - global_seqlen/mean:23700.75 - actor/entropy:0.04867272451519966 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6760060787200928 - training/rollout_probs_diff_mean:0.0020650848746299744 - training/rollout_probs_diff_std:0.011649777181446552 - training/rollout_actor_probs_pearson_corr:0.9952284097671509 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.6796875 - self_distillation/correct_sample_fraction:0.5859375 - 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:1.0000721216201782 - rollout_corr/rollout_is_ratio_fraction_high:9.189698175759986e-05 - rollout_corr/rollout_is_ratio_fraction_low:9.189698175759986e-05 - rollout_corr/rollout_is_max:3.1071109771728516 - rollout_corr/rollout_is_min:0.3156760036945343 - rollout_corr/rollout_is_std:0.03104524314403534 - rollout_corr/rollout_is_eff_sample_size:0.9990371209236779 - rollout_corr/rollout_is_seq_mean:1.0000920295715332 - rollout_corr/rollout_is_seq_std:0.00191143446136266 - rollout_corr/rollout_is_seq_max:1.0089445114135742 - rollout_corr/rollout_is_seq_min:0.9925116300582886 - rollout_corr/rollout_is_seq_max_deviation:0.008944511413574219 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.048692300915718) - rollout_corr/training_log_ppl:np.float64(0.0470506060291882) - rollout_corr/kl:np.float64(0.0007030818600597399) - rollout_corr/k3_kl:np.float64(0.0006832498430933143) - rollout_corr/rollout_ppl:np.float64(1.0479420255869627) - rollout_corr/rollout_log_ppl:np.float64(0.04634752364108863) - rollout_corr/log_ppl_diff:np.float64(0.000703082388099574) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017172047000713064) - rollout_corr/log_ppl_diff_max:np.float64(0.012457795441150665) - rollout_corr/log_ppl_diff_min:np.float64(-0.007961910218000412) - rollout_corr/ppl_ratio:np.float64(1.000706325750798) - rollout_corr/chi2_token:np.float64(0.0012485028710216284) - rollout_corr/chi2_seq:np.float64(0.4721558114979416) - actor/pg_loss:np.float64(0.0008119867416098714) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003208621001249412) - actor/ppo_kl:np.float64(0.00029550659711219396) - 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(0.5859375) - self_distillation/token_reweight_w_mean:np.float64(269301.0433200388) - self_distillation/token_reweight_w_std:np.float64(3263202.0098572485) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000196203123778) - self_distillation/token_reweight_w_clip_frac:np.float64(0.012059328131726943) - self_distillation/empty_target_batch:np.float64(0.3203125) - actor/grad_norm:np.float64(0.1768742771819234) - perf/mfu/actor:np.float64(0.04364625047600226) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.9280548095703) - actor/lr:np.float64(1e-06) - training/global_step:41 - training/epoch:1 - critic/score/mean:0.5859375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5859375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00124348106328398 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00124348106328398 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:340.0546875 - response_l
|
||
step:42 - global_seqlen/min:22778 - global_seqlen/max:27169 - global_seqlen/minmax_diff:4391 - global_seqlen/balanced_min:24454 - global_seqlen/balanced_max:24460 - global_seqlen/mean:24458.25 - actor/entropy:0.05291866511106491 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6709703207015991 - training/rollout_probs_diff_mean:0.0021318153012543917 - training/rollout_probs_diff_std:0.011878360994160175 - training/rollout_actor_probs_pearson_corr:0.9954165816307068 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.75 - self_distillation/correct_sample_fraction:0.69140625 - 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:1.0000886917114258 - rollout_corr/rollout_is_ratio_fraction_high:5.514259828487411e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00011028519656974822 - rollout_corr/rollout_is_max:3.494861125946045 - rollout_corr/rollout_is_min:0.09653104096651077 - rollout_corr/rollout_is_std:0.030159026384353638 - rollout_corr/rollout_is_eff_sample_size:0.9990914976663352 - rollout_corr/rollout_is_seq_mean:1.0000706911087036 - rollout_corr/rollout_is_seq_std:0.0021617193706333637 - rollout_corr/rollout_is_seq_max:1.01577627658844 - rollout_corr/rollout_is_seq_min:0.9939882159233093 - rollout_corr/rollout_is_seq_max_deviation:0.01577627658843994 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0560827553272247) - rollout_corr/training_log_ppl:np.float64(0.053460831095435424) - rollout_corr/kl:np.float64(0.0005555153836205307) - rollout_corr/k3_kl:np.float64(0.0007181561104125933) - rollout_corr/rollout_ppl:np.float64(1.0554866567254066) - rollout_corr/rollout_log_ppl:np.float64(0.05290531515947805) - rollout_corr/log_ppl_diff:np.float64(0.0005555159359573736) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016266171451206901) - rollout_corr/log_ppl_diff_max:np.float64(0.00821683555841446) - rollout_corr/log_ppl_diff_min:np.float64(-0.008203966543078423) - rollout_corr/ppl_ratio:np.float64(1.00055792927742) - rollout_corr/chi2_token:np.float64(0.0020039721857756376) - rollout_corr/chi2_seq:np.float64(0.6317513328976929) - actor/pg_loss:np.float64(0.0017515807412564754) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00029606416092065047) - actor/ppo_kl:np.float64(9.767170322305674e-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(0.75) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.75) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.69140625) - self_distillation/token_reweight_w_mean:np.float64(114590.53960387851) - self_distillation/token_reweight_w_std:np.float64(1339969.1054521985) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9998462598305196) - self_distillation/token_reweight_w_clip_frac:np.float64(0.01849362153734546) - self_distillation/empty_target_batch:np.float64(0.25) - actor/grad_norm:np.float64(0.2964472509920597) - perf/mfu/actor:np.float64(0.04507077937465367) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.83689880371094) - actor/lr:np.float64(1e-06) - training/global_step:42 - training/epoch:1 - critic/score/mean:0.69140625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.69140625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0009236385230906308 - critic/advantages/max:0.625 - critic/advantages/min:-0.75 - critic/returns/mean:-0.0009236385230906308 - critic/returns/max:0.625 - critic/returns/min:-0.75 - response_length/mean:354.1953125 - response_length/max:643.
|
||
step:43 - global_seqlen/min:20322 - global_seqlen/max:27608 - global_seqlen/minmax_diff:7286 - global_seqlen/balanced_min:24259 - global_seqlen/balanced_max:24262 - global_seqlen/mean:24260.875 - actor/entropy:0.05833573266863823 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4241411089897156 - training/rollout_probs_diff_mean:0.002529443707317114 - training/rollout_probs_diff_std:0.012784117832779884 - training/rollout_actor_probs_pearson_corr:0.9950049519538879 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65234375 - self_distillation/correct_sample_fraction:0.546875 - 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:0.9998300075531006 - rollout_corr/rollout_is_ratio_fraction_high:4.4535991037264466e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00015587596863042563 - rollout_corr/rollout_is_max:4.229223728179932 - rollout_corr/rollout_is_min:0.31391194462776184 - rollout_corr/rollout_is_std:0.033042069524526596 - rollout_corr/rollout_is_eff_sample_size:0.9989088771294192 - rollout_corr/rollout_is_seq_mean:0.9998038411140442 - rollout_corr/rollout_is_seq_std:0.002055585151538253 - rollout_corr/rollout_is_seq_max:1.0087283849716187 - rollout_corr/rollout_is_seq_min:0.9881833791732788 - rollout_corr/rollout_is_seq_max_deviation:0.011816620826721191 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.058818947058171) - rollout_corr/training_log_ppl:np.float64(0.05630995506544423) - rollout_corr/kl:np.float64(0.0011501821353707697) - rollout_corr/k3_kl:np.float64(0.000877192553543793) - rollout_corr/rollout_ppl:np.float64(1.0575991161167622) - rollout_corr/rollout_log_ppl:np.float64(0.05515977229333657) - rollout_corr/log_ppl_diff:np.float64(0.0011501827721076552) - rollout_corr/log_ppl_abs_diff:np.float64(0.002000718541239621) - rollout_corr/log_ppl_diff_max:np.float64(0.014770753681659698) - rollout_corr/log_ppl_diff_min:np.float64(-0.004706133157014847) - rollout_corr/ppl_ratio:np.float64(1.0011541976127774) - rollout_corr/chi2_token:np.float64(0.0012625399976968765) - rollout_corr/chi2_seq:np.float64(0.3171506472863257) - actor/pg_loss:np.float64(0.0014347152318805456) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00021798857233079616) - actor/ppo_kl:np.float64(0.0003572766873529698) - 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(0.546875) - self_distillation/token_reweight_w_mean:np.float64(210587.42815934517) - self_distillation/token_reweight_w_std:np.float64(2840550.5963756745) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0007291943766177) - self_distillation/token_reweight_w_clip_frac:np.float64(0.016290025698253885) - self_distillation/empty_target_batch:np.float64(0.34765625) - actor/grad_norm:np.float64(0.28917961940169334) - perf/mfu/actor:np.float64(0.044912664362058916) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.87942504882812) - actor/lr:np.float64(1e-06) - training/global_step:43 - training/epoch:1 - critic/score/mean:0.546875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.546875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0033304570242762566 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.0033304570242762566 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:350.839843
|
||
step:44 - global_seqlen/min:21260 - global_seqlen/max:25228 - global_seqlen/minmax_diff:3968 - global_seqlen/balanced_min:22961 - global_seqlen/balanced_max:22964 - global_seqlen/mean:22962.875 - actor/entropy:0.049587756395339966 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3447294235229492 - training/rollout_probs_diff_mean:0.002002849942073226 - training/rollout_probs_diff_std:0.010673994198441505 - training/rollout_actor_probs_pearson_corr:0.996028482913971 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.75 - self_distillation/correct_sample_fraction:0.734375 - 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:1.0000413656234741 - rollout_corr/rollout_is_ratio_fraction_high:2.309708906977903e-05 - rollout_corr/rollout_is_ratio_fraction_low:8.08398108347319e-05 - rollout_corr/rollout_is_max:2.1141040325164795 - rollout_corr/rollout_is_min:0.37150999903678894 - rollout_corr/rollout_is_std:0.028536269441246986 - rollout_corr/rollout_is_eff_sample_size:0.9991863439184062 - rollout_corr/rollout_is_seq_mean:1.0000253915786743 - rollout_corr/rollout_is_seq_std:0.0017137716058641672 - rollout_corr/rollout_is_seq_max:1.0064252614974976 - rollout_corr/rollout_is_seq_min:0.9936391711235046 - rollout_corr/rollout_is_seq_max_deviation:0.006425261497497559 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0492094033397734) - rollout_corr/training_log_ppl:np.float64(0.04770922514126141) - rollout_corr/kl:np.float64(0.0005806892615183301) - rollout_corr/k3_kl:np.float64(0.0006233692477337627) - rollout_corr/rollout_ppl:np.float64(1.0486026802100241) - rollout_corr/rollout_log_ppl:np.float64(0.04712853589262522) - rollout_corr/log_ppl_diff:np.float64(0.0005806892486361903) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016339209532816312) - rollout_corr/log_ppl_diff_max:np.float64(0.011847276240587234) - rollout_corr/log_ppl_diff_min:np.float64(-0.009022675454616547) - rollout_corr/ppl_ratio:np.float64(1.0005835318006575) - rollout_corr/chi2_token:np.float64(0.001389569602906704) - rollout_corr/chi2_seq:np.float64(0.5874903546646237) - actor/pg_loss:np.float64(0.0005399615038186312) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(9.739950837683864e-05) - actor/ppo_kl:np.float64(0.00017125181667765332) - 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(0.734375) - self_distillation/token_reweight_w_mean:np.float64(4186.4649539021775) - self_distillation/token_reweight_w_std:np.float64(93057.81301470433) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9997784104198217) - self_distillation/token_reweight_w_clip_frac:np.float64(0.005308339008479379) - self_distillation/empty_target_batch:np.float64(0.25) - actor/grad_norm:np.float64(0.11222585383802652) - perf/mfu/actor:np.float64(0.04248327054451827) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.934814453125) - actor/lr:np.float64(1e-06) - training/global_step:44 - training/epoch:1 - critic/score/mean:0.734375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.734375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0022635147906839848 - critic/advantages/max:0.25 - critic/advantages/min:-0.75 - critic/returns/mean:-0.0022635147906839848 - critic/returns/max:0.25 - critic/returns/min:-0.75 - response_length/mean:338.24609375 - response_length/max:619.0 - res
|
||
step:45 - global_seqlen/min:19616 - global_seqlen/max:25545 - global_seqlen/minmax_diff:5929 - global_seqlen/balanced_min:23115 - global_seqlen/balanced_max:23117 - global_seqlen/mean:23116.125 - actor/entropy:0.055425405502319336 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.47441259026527405 - training/rollout_probs_diff_mean:0.00222170352935791 - training/rollout_probs_diff_std:0.011589748784899712 - training/rollout_actor_probs_pearson_corr:0.9956302642822266 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.90234375 - self_distillation/correct_sample_fraction:0.8125 - 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:1.0001627206802368 - rollout_corr/rollout_is_ratio_fraction_high:4.440448174136691e-05 - rollout_corr/rollout_is_ratio_fraction_low:9.991008118959144e-05 - rollout_corr/rollout_is_max:3.4121227264404297 - rollout_corr/rollout_is_min:0.2129592001438141 - rollout_corr/rollout_is_std:0.03031672164797783 - rollout_corr/rollout_is_eff_sample_size:0.9990820973248383 - rollout_corr/rollout_is_seq_mean:1.0001908540725708 - rollout_corr/rollout_is_seq_std:0.0017676244024187326 - rollout_corr/rollout_is_seq_max:1.0077521800994873 - rollout_corr/rollout_is_seq_min:0.995261549949646 - rollout_corr/rollout_is_seq_max_deviation:0.007752180099487305 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0531518626958132) - rollout_corr/training_log_ppl:np.float64(0.05147503435409817) - rollout_corr/kl:np.float64(0.0005863242159556137) - rollout_corr/k3_kl:np.float64(0.0007590526232590378) - rollout_corr/rollout_ppl:np.float64(1.0525323520414531) - rollout_corr/rollout_log_ppl:np.float64(0.05088870941017376) - rollout_corr/log_ppl_diff:np.float64(0.0005863249452886521) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017336888786303462) - rollout_corr/log_ppl_diff_max:np.float64(0.009906811639666557) - rollout_corr/log_ppl_diff_min:np.float64(-0.005315380170941353) - rollout_corr/ppl_ratio:np.float64(1.000589034985751) - rollout_corr/chi2_token:np.float64(0.0022609103471040726) - rollout_corr/chi2_seq:np.float64(0.6674370202235878) - actor/pg_loss:np.float64(0.0017804695526137948) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00047525035915896297) - actor/ppo_kl:np.float64(0.0002818986581543115) - 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(0.8125) - self_distillation/token_reweight_w_mean:np.float64(136099.57786145085) - self_distillation/token_reweight_w_std:np.float64(1974545.3167731077) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.999589323066175) - self_distillation/token_reweight_w_clip_frac:np.float64(0.022642265394097194) - self_distillation/empty_target_batch:np.float64(0.09765625) - actor/grad_norm:np.float64(0.23706506937742233) - perf/mfu/actor:np.float64(0.042947523459426026) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.75249862670898) - actor/lr:np.float64(1e-06) - training/global_step:45 - training/epoch:1 - critic/score/mean:0.8125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.8125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.004180959425866604 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.004180959425866604 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:351.87890625 - re
|
||
step:46 - global_seqlen/min:19925 - global_seqlen/max:25269 - global_seqlen/minmax_diff:5344 - global_seqlen/balanced_min:22809 - global_seqlen/balanced_max:22813 - global_seqlen/mean:22811.75 - actor/entropy:0.05739276111125946 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5554836392402649 - training/rollout_probs_diff_mean:0.002197186229750514 - training/rollout_probs_diff_std:0.011728445999324322 - training/rollout_actor_probs_pearson_corr:0.9958913922309875 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.78125 - 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:0.78125 - rollout_corr/rollout_is_mean:1.0001534223556519 - rollout_corr/rollout_is_ratio_fraction_high:1.1770522178267129e-05 - rollout_corr/rollout_is_ratio_fraction_low:8.239364979090169e-05 - rollout_corr/rollout_is_max:2.795060396194458 - rollout_corr/rollout_is_min:0.011905263178050518 - rollout_corr/rollout_is_std:0.03081785887479782 - rollout_corr/rollout_is_eff_sample_size:0.9990516366740662 - rollout_corr/rollout_is_seq_mean:1.0001554489135742 - rollout_corr/rollout_is_seq_std:0.0020248249638825655 - rollout_corr/rollout_is_seq_max:1.0076408386230469 - rollout_corr/rollout_is_seq_min:0.9923105835914612 - rollout_corr/rollout_is_seq_max_deviation:0.007689416408538818 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0574439051561058) - rollout_corr/training_log_ppl:np.float64(0.054962404239631724) - rollout_corr/kl:np.float64(0.0004922806185305717) - rollout_corr/k3_kl:np.float64(0.0006965439166108922) - rollout_corr/rollout_ppl:np.float64(1.056904122699052) - rollout_corr/rollout_log_ppl:np.float64(0.05447012373679172) - rollout_corr/log_ppl_diff:np.float64(0.0004922805028400035) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017458072843510308) - rollout_corr/log_ppl_diff_max:np.float64(0.030868887901306152) - rollout_corr/log_ppl_diff_min:np.float64(-0.0061208326369524) - rollout_corr/ppl_ratio:np.float64(1.0004963488318026) - rollout_corr/chi2_token:np.float64(0.0016424430068582296) - rollout_corr/chi2_seq:np.float64(0.6610369915142655) - actor/pg_loss:np.float64(0.00031184148974716663) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00019037215861317236) - actor/ppo_kl:np.float64(0.00011649080803977085) - 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(0.78125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.78125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.765625) - self_distillation/token_reweight_w_mean:np.float64(30034.95145637449) - self_distillation/token_reweight_w_std:np.float64(451229.26952795987) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.999818910844624) - self_distillation/token_reweight_w_clip_frac:np.float64(0.00749496826028917) - self_distillation/empty_target_batch:np.float64(0.21875) - actor/grad_norm:np.float64(0.0970448236912489) - perf/mfu/actor:np.float64(0.0420408031250471) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.8536114692688) - actor/lr:np.float64(1e-06) - training/global_step:46 - 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:0.0005267308442853391 - critic/advantages/max:0.25 - critic/advantages/min:-0.875 - critic/returns/mean:0.0005267308442853391 - critic/returns/max:0.25 - critic/returns/min:-0.875 - response_length/mean:331.8671875 - response_length/
|
||
step:47 - global_seqlen/min:20827 - global_seqlen/max:26713 - global_seqlen/minmax_diff:5886 - global_seqlen/balanced_min:24177 - global_seqlen/balanced_max:24180 - global_seqlen/mean:24179.125 - actor/entropy:0.055492036044597626 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6224586963653564 - training/rollout_probs_diff_mean:0.0022652670741081238 - training/rollout_probs_diff_std:0.011492198333144188 - training/rollout_actor_probs_pearson_corr:0.9957019686698914 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71875 - self_distillation/correct_sample_fraction:0.6953125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.71875 - rollout_corr/rollout_is_mean:0.9999520778656006 - rollout_corr/rollout_is_ratio_fraction_high:2.1471898435265757e-05 - rollout_corr/rollout_is_ratio_fraction_low:9.662354568718001e-05 - rollout_corr/rollout_is_max:2.4000132083892822 - rollout_corr/rollout_is_min:0.3577742278575897 - rollout_corr/rollout_is_std:0.030129367485642433 - rollout_corr/rollout_is_eff_sample_size:0.9990929255817853 - rollout_corr/rollout_is_seq_mean:0.9999911785125732 - rollout_corr/rollout_is_seq_std:0.0017208998324349523 - rollout_corr/rollout_is_seq_max:1.0061891078948975 - rollout_corr/rollout_is_seq_min:0.9937074184417725 - rollout_corr/rollout_is_seq_max_deviation:0.006292581558227539 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0576613186858594) - rollout_corr/training_log_ppl:np.float64(0.05537509770147153) - rollout_corr/kl:np.float64(0.0005087664376974832) - rollout_corr/k3_kl:np.float64(0.0005032766369623687) - rollout_corr/rollout_ppl:np.float64(1.057125517167151) - rollout_corr/rollout_log_ppl:np.float64(0.05486633105647343) - rollout_corr/log_ppl_diff:np.float64(0.0005087666449981043) - rollout_corr/log_ppl_abs_diff:np.float64(0.001345907576251193) - rollout_corr/log_ppl_diff_max:np.float64(0.005119312554597855) - rollout_corr/log_ppl_diff_min:np.float64(-0.004925638437271118) - rollout_corr/ppl_ratio:np.float64(1.0005102870054543) - rollout_corr/chi2_token:np.float64(0.0010242487769573927) - rollout_corr/chi2_seq:np.float64(0.25397335505113006) - actor/pg_loss:np.float64(0.0007073208689689636) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00021316877064236905) - actor/ppo_kl:np.float64(2.2292524672451464e-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(0.71875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6953125) - self_distillation/token_reweight_w_mean:np.float64(16477.566876600962) - self_distillation/token_reweight_w_std:np.float64(290898.57240845717) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9991839011199772) - self_distillation/token_reweight_w_clip_frac:np.float64(0.014518271171255037) - self_distillation/empty_target_batch:np.float64(0.28125) - actor/grad_norm:np.float64(0.1594111081212759) - perf/mfu/actor:np.float64(0.04459250728956798) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.8540496826172) - actor/lr:np.float64(1e-06) - training/global_step:47 - training/epoch:1 - critic/score/mean:0.6953125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6953125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.00286381458863616 - critic/advantages/max:0.25 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00286381458863616 - critic/returns/max:0.25 - critic/returns/min:-0.875 - response_length/mean:363.84765625 - respon
|
||
step:48 - global_seqlen/min:19556 - global_seqlen/max:26843 - global_seqlen/minmax_diff:7287 - global_seqlen/balanced_min:23908 - global_seqlen/balanced_max:23912 - global_seqlen/mean:23910.25 - actor/entropy:0.05349171906709671 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5511896014213562 - training/rollout_probs_diff_mean:0.001996841514483094 - training/rollout_probs_diff_std:0.01053113304078579 - training/rollout_actor_probs_pearson_corr:0.9963048100471497 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.8984375 - self_distillation/correct_sample_fraction:0.8203125 - 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:1.0001131296157837 - rollout_corr/rollout_is_ratio_fraction_high:3.288536026957445e-05 - rollout_corr/rollout_is_ratio_fraction_low:9.865608444670215e-05 - rollout_corr/rollout_is_max:3.7064740657806396 - rollout_corr/rollout_is_min:0.019543590024113655 - rollout_corr/rollout_is_std:0.02837498113512993 - rollout_corr/rollout_is_eff_sample_size:0.9991958652393542 - rollout_corr/rollout_is_seq_mean:1.0001505613327026 - rollout_corr/rollout_is_seq_std:0.001532649272121489 - rollout_corr/rollout_is_seq_max:1.0046343803405762 - rollout_corr/rollout_is_seq_min:0.9958474040031433 - rollout_corr/rollout_is_seq_max_deviation:0.004634380340576172 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0504649495705962) - rollout_corr/training_log_ppl:np.float64(0.04875240848195972) - rollout_corr/kl:np.float64(0.0003140236587224621) - rollout_corr/k3_kl:np.float64(0.0005160192457793755) - rollout_corr/rollout_ppl:np.float64(1.050125792156905) - rollout_corr/rollout_log_ppl:np.float64(0.0484383839811926) - rollout_corr/log_ppl_diff:np.float64(0.0003140245007671183) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014380264019564493) - rollout_corr/log_ppl_diff_max:np.float64(0.012833669781684875) - rollout_corr/log_ppl_diff_min:np.float64(-0.007530331611633301) - rollout_corr/ppl_ratio:np.float64(1.000316001707688) - rollout_corr/chi2_token:np.float64(0.0014003217220306396) - rollout_corr/chi2_seq:np.float64(0.5536817323882133) - actor/pg_loss:np.float64(0.0007914919406175613) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0001905762474052608) - actor/ppo_kl:np.float64(2.5524123401510224e-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(0.8203125) - self_distillation/token_reweight_w_mean:np.float64(199612.87102979445) - self_distillation/token_reweight_w_std:np.float64(2349294.0318124965) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0000342559069395) - self_distillation/token_reweight_w_clip_frac:np.float64(0.013467349508573534) - self_distillation/empty_target_batch:np.float64(0.1015625) - actor/grad_norm:np.float64(0.18085278756916523) - perf/mfu/actor:np.float64(0.04381698070263809) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.83398818969727) - actor/lr:np.float64(1e-06) - training/global_step:48 - training/epoch:1 - critic/score/mean:0.8203125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.8203125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0003589985426515341 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0003589985426515341 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:356.3515625 -
|
||
step:49 - global_seqlen/min:20968 - global_seqlen/max:26342 - global_seqlen/minmax_diff:5374 - global_seqlen/balanced_min:23295 - global_seqlen/balanced_max:23298 - global_seqlen/mean:23296.875 - actor/entropy:0.05485960841178894 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9957264065742493 - training/rollout_probs_diff_mean:0.002162784105166793 - training/rollout_probs_diff_std:0.012243074364960194 - training/rollout_actor_probs_pearson_corr:0.9952481389045715 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71484375 - 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:0.71484375 - rollout_corr/rollout_is_mean:1.000170350074768 - rollout_corr/rollout_is_ratio_fraction_high:4.6703327825525776e-05 - rollout_corr/rollout_is_ratio_fraction_low:9.340665565105155e-05 - rollout_corr/rollout_is_max:3.282700300216675 - rollout_corr/rollout_is_min:0.004070968832820654 - rollout_corr/rollout_is_std:0.03059268184006214 - rollout_corr/rollout_is_eff_sample_size:0.9990654389263088 - rollout_corr/rollout_is_seq_mean:1.0001893043518066 - rollout_corr/rollout_is_seq_std:0.0018957956926897168 - rollout_corr/rollout_is_seq_max:1.0093369483947754 - rollout_corr/rollout_is_seq_min:0.9951988458633423 - rollout_corr/rollout_is_seq_max_deviation:0.00933694839477539 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.055814309976995) - rollout_corr/training_log_ppl:np.float64(0.05374641695198079) - rollout_corr/kl:np.float64(0.0004828907792662207) - rollout_corr/k3_kl:np.float64(0.0005371262649731534) - rollout_corr/rollout_ppl:np.float64(1.0553037645295262) - rollout_corr/rollout_log_ppl:np.float64(0.05326352582505933) - rollout_corr/log_ppl_diff:np.float64(0.00048289112692145864) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015306145369322621) - rollout_corr/log_ppl_diff_max:np.float64(0.007901884615421295) - rollout_corr/log_ppl_diff_min:np.float64(-0.007005564868450165) - rollout_corr/ppl_ratio:np.float64(1.0004850262776017) - rollout_corr/chi2_token:np.float64(0.0012454038951545954) - rollout_corr/chi2_seq:np.float64(0.35367413074709475) - actor/pg_loss:np.float64(0.0008061726111918688) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00020583227842507767) - actor/ppo_kl:np.float64(0.00010553301872917586) - 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(0.71484375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71484375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6171875) - self_distillation/token_reweight_w_mean:np.float64(163828.87535567163) - self_distillation/token_reweight_w_std:np.float64(2273528.9927650653) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0008936680387706) - self_distillation/token_reweight_w_clip_frac:np.float64(0.014962271772674285) - self_distillation/empty_target_batch:np.float64(0.28515625) - actor/grad_norm:np.float64(0.18702556192874908) - perf/mfu/actor:np.float64(0.04264424569583532) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.88611793518066) - actor/lr:np.float64(1e-06) - training/global_step:49 - 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:0.0024738169740885496 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0024738169740885496 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:33
|
||
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
|
||
Given the following crystal structure parameters for the material RbLa9(IrO6)4 (Material ID: mp-560657), calculate the volume of the unit cell (in ų).
|
||
|
||
Lattice Parameters:
|
||
|
||
a = 7.82 Å
|
||
b = 7.82 Å
|
||
c = 17.88 Å
|
||
α = 90.00º
|
||
β = 90.00º
|
||
γ = 90.00º
|
||
|
||
|
||
|
||
A: 1025.67 ų
|
||
|
||
B: 1094.31 ų
|
||
|
||
C: 1200.45 ų
|
||
|
||
D: 1150.78 ų
|
||
|
||
Please reason step by step.
|
||
assistant
|
||
<think>
|
||
|
||
</think>
|
||
|
||
|
||
[response] <reasoning>
|
||
The volume $ V $ of a unit cell in a crystal structure can be calculated using the formula:
|
||
|
||
$$
|
||
V = a \cdot b \cdot c \cdot \sqrt{1 + 2\cos\alpha\cos\beta\cos\gamma - \cos^2\alpha - \cos^2\beta - \cos^2\gamma}
|
||
$$
|
||
|
||
Given that all angles $ \alpha, \beta, \gamma $ are 90.00º, the cosine of each angle is 0. Substituting these values into the formula:
|
||
|
||
$$
|
||
V = a \cdot b \cdot c \cdot \sqrt{1 + 2(0)(0)(0) - (0)^2 - (0)^2 - (0)^2} = a \cdot b \cdot c \cdot \sqrt{1} = a \cdot b \cdot c
|
||
$$
|
||
|
||
Now, substituting the given lattice parameters:
|
||
|
||
$$
|
||
V = 7.82 \, \text{Å} \times 7.82 \, \text{Å} \times 17.88 \, \text{Å}
|
||
$$
|
||
|
||
First, calculate $ 7.82 \times 7.82 $:
|
||
|
||
$$
|
||
7.82 \times 7.82 = 61.1524
|
||
$$
|
||
|
||
Then multiply by 17.88:
|
||
|
||
$$
|
||
61.1524 \times 17.88 \approx 1094.31
|
||
$$
|
||
|
||
Thus, the volume of the unit cell is approximately 1094.31 ų.
|
||
</reasoning>
|
||
<answer>
|
||
B
|
||
</answer>
|
||
[ground_truth] B
|
||
[score] 1.0
|
||
[acc] 1.0
|
||
[pred] B
|
||
[incorrect_format] 1
|
||
[feedback]
|
||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-4B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_50
|
||
step:50 - global_seqlen/min:18307 - global_seqlen/max:25358 - global_seqlen/minmax_diff:7051 - global_seqlen/balanced_min:22906 - global_seqlen/balanced_max:22909 - global_seqlen/mean:22907.375 - actor/entropy:0.0527101494371891 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.530918538570404 - training/rollout_probs_diff_mean:0.0021026728209108114 - training/rollout_probs_diff_std:0.01144232228398323 - training/rollout_actor_probs_pearson_corr:0.9958080053329468 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.78125 - self_distillation/correct_sample_fraction:0.75 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.78125 - rollout_corr/rollout_is_mean:0.999865710735321 - rollout_corr/rollout_is_ratio_fraction_high:2.4842560378601775e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00012421280553098768 - rollout_corr/rollout_is_max:2.086440086364746 - rollout_corr/rollout_is_min:0.33655375242233276 - rollout_corr/rollout_is_std:0.029702994972467422 - rollout_corr/rollout_is_eff_sample_size:0.9991182717602443 - rollout_corr/rollout_is_seq_mean:0.9998937845230103 - rollout_corr/rollout_is_seq_std:0.0017050968017429113 - rollout_corr/rollout_is_seq_max:1.0061534643173218 - rollout_corr/rollout_is_seq_min:0.9932185411453247 - rollout_corr/rollout_is_seq_max_deviation:0.006781458854675293 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0527593255974352) - rollout_corr/training_log_ppl:np.float64(0.05062973701751616) - rollout_corr/kl:np.float64(0.0005176823531662933) - rollout_corr/k3_kl:np.float64(0.0004894320357937687) - rollout_corr/rollout_ppl:np.float64(1.0521913496777415) - rollout_corr/rollout_log_ppl:np.float64(0.05011205441496713) - rollout_corr/log_ppl_diff:np.float64(0.00051768260254903) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014349543425851152) - rollout_corr/log_ppl_diff_max:np.float64(0.008032098412513733) - rollout_corr/log_ppl_diff_min:np.float64(-0.004613518714904785) - rollout_corr/ppl_ratio:np.float64(1.0005198165308684) - rollout_corr/chi2_token:np.float64(0.0008812225423753262) - rollout_corr/chi2_seq:np.float64(0.2764910035766661) - actor/pg_loss:np.float64(0.0006625362439081073) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00022346714558807435) - actor/ppo_kl:np.float64(-5.137884922845615e-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(0.78125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.78125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.75) - self_distillation/token_reweight_w_mean:np.float64(111777.4753512349) - self_distillation/token_reweight_w_std:np.float64(1134010.0336442993) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.999779810430482) - self_distillation/token_reweight_w_clip_frac:np.float64(0.008872004498698516) - self_distillation/empty_target_batch:np.float64(0.21875) - actor/grad_norm:np.float64(0.18878151662647724) - perf/mfu/actor:np.float64(0.04136322343812593) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.93096733093262) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7473404255319149) - val-aux/sciknoweval/reward/std@16:np.float64(0.05458479149126176) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7693617021276595) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.04424280678647665) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.7248297872340426) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.048188095845902945) - val-aux/sciknoweval/reward/maj@2/mea
|
||
step:51 - global_seqlen/min:20598 - global_seqlen/max:24762 - global_seqlen/minmax_diff:4164 - global_seqlen/balanced_min:22903 - global_seqlen/balanced_max:22909 - global_seqlen/mean:22907.125 - actor/entropy:0.05307382717728615 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.681950569152832 - training/rollout_probs_diff_mean:0.002189614810049534 - training/rollout_probs_diff_std:0.011771688237786293 - training/rollout_actor_probs_pearson_corr:0.9954410791397095 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.77734375 - 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:0.77734375 - rollout_corr/rollout_is_mean:1.0001434087753296 - rollout_corr/rollout_is_ratio_fraction_high:4.8094843805301934e-05 - rollout_corr/rollout_is_ratio_fraction_low:7.21422620699741e-05 - rollout_corr/rollout_is_max:3.7063121795654297 - rollout_corr/rollout_is_min:0.1226673275232315 - rollout_corr/rollout_is_std:0.030606316402554512 - rollout_corr/rollout_is_eff_sample_size:0.9990644870346002 - rollout_corr/rollout_is_seq_mean:1.0002003908157349 - rollout_corr/rollout_is_seq_std:0.002001448068767786 - rollout_corr/rollout_is_seq_max:1.0162816047668457 - rollout_corr/rollout_is_seq_min:0.9955185055732727 - rollout_corr/rollout_is_seq_max_deviation:0.016281604766845703 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0550247388891876) - rollout_corr/training_log_ppl:np.float64(0.05237314046280517) - rollout_corr/kl:np.float64(0.0005226194430250075) - rollout_corr/k3_kl:np.float64(0.0006975394764396903) - rollout_corr/rollout_ppl:np.float64(1.0544638223946095) - rollout_corr/rollout_log_ppl:np.float64(0.051850520834705094) - rollout_corr/log_ppl_diff:np.float64(0.0005226196281000739) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016185973963729339) - rollout_corr/log_ppl_diff_max:np.float64(0.010919332504272461) - rollout_corr/log_ppl_diff_min:np.float64(-0.005789030343294144) - rollout_corr/ppl_ratio:np.float64(1.0005250833928585) - rollout_corr/chi2_token:np.float64(0.0018498406279832125) - rollout_corr/chi2_seq:np.float64(0.5595120217185467) - actor/pg_loss:np.float64(0.0013827746734023094) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00044605470702663297) - actor/ppo_kl:np.float64(0.00020610883375127287) - 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(0.77734375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.77734375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.671875) - self_distillation/token_reweight_w_mean:np.float64(184521.5831220611) - self_distillation/token_reweight_w_std:np.float64(2546310.6744150682) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0008375360630453) - self_distillation/token_reweight_w_clip_frac:np.float64(0.01984426895796787) - self_distillation/empty_target_batch:np.float64(0.22265625) - actor/grad_norm:np.float64(0.24625525996088982) - perf/mfu/actor:np.float64(0.0423324768937178) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.52999877929688) - actor/lr:np.float64(1e-06) - training/global_step:51 - 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:-0.0062553356401622295 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.0062553356401622295 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:324.878906
|
||
step:52 - global_seqlen/min:19845 - global_seqlen/max:26627 - global_seqlen/minmax_diff:6782 - global_seqlen/balanced_min:23480 - global_seqlen/balanced_max:23481 - global_seqlen/mean:23480.75 - actor/entropy:0.04984758421778679 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7995882630348206 - training/rollout_probs_diff_mean:0.00212426227517426 - training/rollout_probs_diff_std:0.01242454070597887 - training/rollout_actor_probs_pearson_corr:0.9947465062141418 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80859375 - 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:0.80859375 - rollout_corr/rollout_is_mean:0.9999185800552368 - rollout_corr/rollout_is_ratio_fraction_high:5.9063953813165426e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00014175349497236311 - rollout_corr/rollout_is_max:5.277081489562988 - rollout_corr/rollout_is_min:0.02199946902692318 - rollout_corr/rollout_is_std:0.030726822093129158 - rollout_corr/rollout_is_eff_sample_size:0.9990565150127493 - rollout_corr/rollout_is_seq_mean:0.999808132648468 - rollout_corr/rollout_is_seq_std:0.0021094551775604486 - rollout_corr/rollout_is_seq_max:1.0133817195892334 - rollout_corr/rollout_is_seq_min:0.9917005896568298 - rollout_corr/rollout_is_seq_max_deviation:0.013381719589233398 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0494226575829089) - rollout_corr/training_log_ppl:np.float64(0.047722964010972646) - rollout_corr/kl:np.float64(0.0010447812739915463) - rollout_corr/k3_kl:np.float64(0.0007033883213365399) - rollout_corr/rollout_ppl:np.float64(1.0483222524635494) - rollout_corr/rollout_log_ppl:np.float64(0.04667818260804779) - rollout_corr/log_ppl_diff:np.float64(0.0010447814029248548) - rollout_corr/log_ppl_abs_diff:np.float64(0.001777110573129903) - rollout_corr/log_ppl_diff_max:np.float64(0.011939480900764465) - rollout_corr/log_ppl_diff_min:np.float64(-0.0066554956138134) - rollout_corr/ppl_ratio:np.float64(1.0010481965728104) - rollout_corr/chi2_token:np.float64(0.0008801114745438099) - rollout_corr/chi2_seq:np.float64(0.4005721795838326) - actor/pg_loss:np.float64(0.0008483310230076313) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0005638624024868477) - actor/ppo_kl:np.float64(0.00027875212417960427) - 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(0.7421875) - self_distillation/token_reweight_w_mean:np.float64(222058.34980464377) - self_distillation/token_reweight_w_std:np.float64(2551187.82467389) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0001434024889022) - self_distillation/token_reweight_w_clip_frac:np.float64(0.01402115351811517) - self_distillation/empty_target_batch:np.float64(0.19140625) - actor/grad_norm:np.float64(0.30175722017884254) - perf/mfu/actor:np.float64(0.04329011667396355) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.3583869934082) - actor/lr:np.float64(1e-06) - training/global_step:52 - 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:-0.0027361377142369747 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0027361377142369747 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:330.679687
|
||
step:53 - global_seqlen/min:20647 - global_seqlen/max:23970 - global_seqlen/minmax_diff:3323 - global_seqlen/balanced_min:22483 - global_seqlen/balanced_max:22486 - global_seqlen/mean:22485.375 - actor/entropy:0.04552474617958069 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.39975857734680176 - training/rollout_probs_diff_mean:0.0019856856670230627 - training/rollout_probs_diff_std:0.011346102692186832 - training/rollout_actor_probs_pearson_corr:0.995132327079773 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.6875 - self_distillation/correct_sample_fraction:0.65234375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.6875 - rollout_corr/rollout_is_mean:0.9999587535858154 - rollout_corr/rollout_is_ratio_fraction_high:5.100284397485666e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00010200568794971332 - rollout_corr/rollout_is_max:2.554445505142212 - rollout_corr/rollout_is_min:0.3733019232749939 - rollout_corr/rollout_is_std:0.029692960903048515 - rollout_corr/rollout_is_eff_sample_size:0.9991189857557254 - rollout_corr/rollout_is_seq_mean:1.0000665187835693 - rollout_corr/rollout_is_seq_std:0.0020678448490798473 - rollout_corr/rollout_is_seq_max:1.0102448463439941 - rollout_corr/rollout_is_seq_min:0.9950273036956787 - rollout_corr/rollout_is_seq_max_deviation:0.01024484634399414 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0451452140696347) - rollout_corr/training_log_ppl:np.float64(0.04381734907747159) - rollout_corr/kl:np.float64(0.00043769131724502586) - rollout_corr/k3_kl:np.float64(0.0005817715615477614) - rollout_corr/rollout_ppl:np.float64(1.0446803770028055) - rollout_corr/rollout_log_ppl:np.float64(0.04337965739887295) - rollout_corr/log_ppl_diff:np.float64(0.00043769167859863956) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015039140798762674) - rollout_corr/log_ppl_diff_max:np.float64(0.006083883345127106) - rollout_corr/log_ppl_diff_min:np.float64(-0.006761221215128899) - rollout_corr/ppl_ratio:np.float64(1.0004397267475724) - rollout_corr/chi2_token:np.float64(0.0014745418448001146) - rollout_corr/chi2_seq:np.float64(0.42398698115721345) - actor/pg_loss:np.float64(0.0009315765928477049) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00026180762870353647) - actor/ppo_kl:np.float64(4.117221907051771e-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(0.6875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.6875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.65234375) - self_distillation/token_reweight_w_mean:np.float64(49304.76177130593) - self_distillation/token_reweight_w_std:np.float64(677681.4286856137) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9997115456499159) - self_distillation/token_reweight_w_clip_frac:np.float64(0.013133775428286754) - self_distillation/empty_target_batch:np.float64(0.3125) - actor/grad_norm:np.float64(0.19336532801389694) - perf/mfu/actor:np.float64(0.041706615520811945) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.17058563232422) - actor/lr:np.float64(1e-06) - training/global_step:53 - training/epoch:2 - critic/score/mean:0.65234375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.65234375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.001973172416910529 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:-0.001973172416910529 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:306.35546875 - res
|
||
step:54 - global_seqlen/min:20588 - global_seqlen/max:26029 - global_seqlen/minmax_diff:5441 - global_seqlen/balanced_min:23719 - global_seqlen/balanced_max:23721 - global_seqlen/mean:23720.0 - actor/entropy:0.04974914714694023 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5220087766647339 - training/rollout_probs_diff_mean:0.002054911106824875 - training/rollout_probs_diff_std:0.011294582858681679 - training/rollout_actor_probs_pearson_corr:0.9953687191009521 - self_distillation/success_group_fraction:0.625 - self_distillation/success_sample_fraction:0.625 - self_distillation/correct_sample_fraction:0.5703125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.625 - rollout_corr/rollout_is_mean:1.0000876188278198 - rollout_corr/rollout_is_ratio_fraction_high:1.1769137017836329e-05 - rollout_corr/rollout_is_ratio_fraction_low:7.061482028802857e-05 - rollout_corr/rollout_is_max:2.3180410861968994 - rollout_corr/rollout_is_min:0.4453998804092407 - rollout_corr/rollout_is_std:0.028263429179787636 - rollout_corr/rollout_is_eff_sample_size:0.9992019351761722 - rollout_corr/rollout_is_seq_mean:1.0000823736190796 - rollout_corr/rollout_is_seq_std:0.0018613808788359165 - rollout_corr/rollout_is_seq_max:1.0083881616592407 - rollout_corr/rollout_is_seq_min:0.9941661357879639 - rollout_corr/rollout_is_seq_max_deviation:0.008388161659240723 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0487789153121412) - rollout_corr/training_log_ppl:np.float64(0.04697248301545187) - rollout_corr/kl:np.float64(0.0006158747329720882) - rollout_corr/k3_kl:np.float64(0.0005426218764306157) - rollout_corr/rollout_ppl:np.float64(1.0481268735602498) - rollout_corr/rollout_log_ppl:np.float64(0.04635660783787898) - rollout_corr/log_ppl_diff:np.float64(0.0006158751775728888) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014905650741638965) - rollout_corr/log_ppl_diff_max:np.float64(0.015574745833873749) - rollout_corr/log_ppl_diff_min:np.float64(-0.006432611495256424) - rollout_corr/ppl_ratio:np.float64(1.0006186675745994) - rollout_corr/chi2_token:np.float64(0.0008958489634096622) - rollout_corr/chi2_seq:np.float64(0.19869494764134288) - actor/pg_loss:np.float64(0.00033523840829730034) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00021035506097177858) - actor/ppo_kl:np.float64(0.00026634746228637596) - 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(0.625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.5703125) - self_distillation/token_reweight_w_mean:np.float64(87162.10803046962) - self_distillation/token_reweight_w_std:np.float64(1211193.186900845) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000149110564962) - self_distillation/token_reweight_w_clip_frac:np.float64(0.011021143785910681) - self_distillation/empty_target_batch:np.float64(0.375) - actor/grad_norm:np.float64(0.19004646688699722) - perf/mfu/actor:np.float64(0.04410223401166184) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.18850708007812) - actor/lr:np.float64(1e-06) - training/global_step:54 - training/epoch:2 - critic/score/mean:0.5703125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5703125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0010548088466748595 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0010548088466748595 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:331.90625 - response_length/max
|
||
step:55 - global_seqlen/min:21032 - global_seqlen/max:25317 - global_seqlen/minmax_diff:4285 - global_seqlen/balanced_min:23375 - global_seqlen/balanced_max:23376 - global_seqlen/mean:23375.25 - actor/entropy:0.0445924773812294 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3997594118118286 - training/rollout_probs_diff_mean:0.0019081803038716316 - training/rollout_probs_diff_std:0.010883488692343235 - training/rollout_actor_probs_pearson_corr:0.9953238368034363 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.75 - self_distillation/correct_sample_fraction:0.73828125 - 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:0.9999055862426758 - rollout_corr/rollout_is_ratio_fraction_high:2.459117240505293e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00011066027218475938 - rollout_corr/rollout_is_max:2.090315341949463 - rollout_corr/rollout_is_min:0.23710303008556366 - rollout_corr/rollout_is_std:0.027302302420139313 - rollout_corr/rollout_is_eff_sample_size:0.9992550205113556 - rollout_corr/rollout_is_seq_mean:0.9998701810836792 - rollout_corr/rollout_is_seq_std:0.001544875674881041 - rollout_corr/rollout_is_seq_max:1.0056719779968262 - rollout_corr/rollout_is_seq_min:0.9948750138282776 - rollout_corr/rollout_is_seq_max_deviation:0.005671977996826172 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.045014244504273) - rollout_corr/training_log_ppl:np.float64(0.043634509600451565) - rollout_corr/kl:np.float64(0.0006418770270268936) - rollout_corr/k3_kl:np.float64(0.0005075502466063142) - rollout_corr/rollout_ppl:np.float64(1.0443436088971794) - rollout_corr/rollout_log_ppl:np.float64(0.04299263210441495) - rollout_corr/log_ppl_diff:np.float64(0.0006418774960366136) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015478016607630707) - rollout_corr/log_ppl_diff_max:np.float64(0.008348822593688965) - rollout_corr/log_ppl_diff_min:np.float64(-0.004879988729953766) - rollout_corr/ppl_ratio:np.float64(1.000643964856863) - rollout_corr/chi2_token:np.float64(0.0007614318747073412) - rollout_corr/chi2_seq:np.float64(0.5452475808560848) - actor/pg_loss:np.float64(0.00022943562362343073) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(4.411085774336243e-05) - actor/ppo_kl:np.float64(0.00010750114318014425) - 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(0.73828125) - self_distillation/token_reweight_w_mean:np.float64(19822.99445243622) - self_distillation/token_reweight_w_std:np.float64(292975.02256671095) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9998247078619897) - self_distillation/token_reweight_w_clip_frac:np.float64(0.005360060549719492) - self_distillation/empty_target_batch:np.float64(0.25) - actor/grad_norm:np.float64(0.09271686151623726) - perf/mfu/actor:np.float64(0.043276245005982816) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.02014541625977) - actor/lr:np.float64(1e-06) - training/global_step:55 - training/epoch:2 - critic/score/mean:0.73828125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.73828125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.001120435306802392 - critic/advantages/max:0.125 - critic/advantages/min:-0.875 - critic/returns/mean:0.001120435306802392 - critic/returns/max:0.125 - critic/returns/min:-0.875 - response_length/mean:317.6953125 - response_length/max:6
|
||
step:56 - global_seqlen/min:19749 - global_seqlen/max:25955 - global_seqlen/minmax_diff:6206 - global_seqlen/balanced_min:23414 - global_seqlen/balanced_max:23416 - global_seqlen/mean:23415.5 - actor/entropy:0.0621294341981411 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5481129288673401 - training/rollout_probs_diff_mean:0.0021422128193080425 - training/rollout_probs_diff_std:0.011219512671232224 - training/rollout_actor_probs_pearson_corr:0.9964797496795654 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.7890625 - 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:0.9999027848243713 - rollout_corr/rollout_is_ratio_fraction_high:1.1152250408485997e-05 - rollout_corr/rollout_is_ratio_fraction_low:8.921800326788798e-05 - rollout_corr/rollout_is_max:2.068037509918213 - rollout_corr/rollout_is_min:0.025016378611326218 - rollout_corr/rollout_is_std:0.02902502380311489 - rollout_corr/rollout_is_eff_sample_size:0.9991581380701209 - rollout_corr/rollout_is_seq_mean:0.9998484253883362 - rollout_corr/rollout_is_seq_std:0.0016506003448739648 - rollout_corr/rollout_is_seq_max:1.0049660205841064 - rollout_corr/rollout_is_seq_min:0.9948515295982361 - rollout_corr/rollout_is_seq_max_deviation:0.005148470401763916 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0614002938382328) - rollout_corr/training_log_ppl:np.float64(0.058645859207899775) - rollout_corr/kl:np.float64(0.0006652666341668834) - rollout_corr/k3_kl:np.float64(0.0005047853511887801) - rollout_corr/rollout_ppl:np.float64(1.060690845362842) - rollout_corr/rollout_log_ppl:np.float64(0.057980592235253425) - rollout_corr/log_ppl_diff:np.float64(0.00066526697264635) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014072957146709086) - rollout_corr/log_ppl_diff_max:np.float64(0.01607600599527359) - rollout_corr/log_ppl_diff_min:np.float64(-0.003986669704318047) - rollout_corr/ppl_ratio:np.float64(1.0006674323230982) - rollout_corr/chi2_token:np.float64(0.000626653665676713) - rollout_corr/chi2_seq:np.float64(0.17073907726444304) - actor/pg_loss:np.float64(0.000534118851646781) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.000130811873532366) - actor/ppo_kl:np.float64(5.016306554850303e-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(0.7890625) - self_distillation/token_reweight_w_mean:np.float64(35010.49396638153) - self_distillation/token_reweight_w_std:np.float64(558753.1349300193) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9996146762277931) - self_distillation/token_reweight_w_clip_frac:np.float64(0.009658753042458557) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.12083789333701134) - perf/mfu/actor:np.float64(0.043045165586583715) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.06346893310547) - actor/lr:np.float64(1e-06) - training/global_step:56 - training/epoch:2 - critic/score/mean:0.7890625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7890625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-2.9274657208588906e-05 - critic/advantages/max:0.375 - critic/advantages/min:-0.875 - critic/returns/mean:-2.9274657208588906e-05 - critic/returns/max:0.375 - critic/returns/min:-0.875 - response_length/mean:350.265625 - response_lengt
|
||
step:57 - global_seqlen/min:20930 - global_seqlen/max:25923 - global_seqlen/minmax_diff:4993 - global_seqlen/balanced_min:23392 - global_seqlen/balanced_max:23395 - global_seqlen/mean:23393.875 - actor/entropy:0.054561372846364975 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4967532455921173 - training/rollout_probs_diff_mean:0.002252341015264392 - training/rollout_probs_diff_std:0.011561992578208447 - training/rollout_actor_probs_pearson_corr:0.9957065582275391 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.7578125 - 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:0.9999447464942932 - rollout_corr/rollout_is_ratio_fraction_high:3.6330609873402864e-05 - rollout_corr/rollout_is_ratio_fraction_low:7.266121974680573e-05 - rollout_corr/rollout_is_max:2.0692319869995117 - rollout_corr/rollout_is_min:0.24590204656124115 - rollout_corr/rollout_is_std:0.029926901683211327 - rollout_corr/rollout_is_eff_sample_size:0.9991050630279074 - rollout_corr/rollout_is_seq_mean:0.9999193549156189 - rollout_corr/rollout_is_seq_std:0.0018134000711143017 - rollout_corr/rollout_is_seq_max:1.0067952871322632 - rollout_corr/rollout_is_seq_min:0.9954656362533569 - rollout_corr/rollout_is_seq_max_deviation:0.006795287132263184 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0550112552009523) - rollout_corr/training_log_ppl:np.float64(0.05320324154399714) - rollout_corr/kl:np.float64(0.0007456150652345173) - rollout_corr/k3_kl:np.float64(0.0006398622932053399) - rollout_corr/rollout_ppl:np.float64(1.0542202196083963) - rollout_corr/rollout_log_ppl:np.float64(0.052457625864917645) - rollout_corr/log_ppl_diff:np.float64(0.0007456156790794921) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015578475331494701) - rollout_corr/log_ppl_diff_max:np.float64(0.008377820253372192) - rollout_corr/log_ppl_diff_min:np.float64(-0.003737565129995346) - rollout_corr/ppl_ratio:np.float64(1.0007477451581508) - rollout_corr/chi2_token:np.float64(0.0011580914724618196) - rollout_corr/chi2_seq:np.float64(0.10339058190584183) - actor/pg_loss:np.float64(0.0008008657023310661) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002839279932231875) - actor/ppo_kl:np.float64(0.00018597565893951895) - 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(0.7578125) - self_distillation/token_reweight_w_mean:np.float64(134570.97113014408) - self_distillation/token_reweight_w_std:np.float64(1767222.71986674) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9998309221118689) - self_distillation/token_reweight_w_clip_frac:np.float64(0.01680245746683795) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.23896951600909233) - perf/mfu/actor:np.float64(0.04274646358494116) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(145.98765563964844) - actor/lr:np.float64(1e-06) - training/global_step:57 - training/epoch:2 - critic/score/mean:0.7578125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7578125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0018664849922060966 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0018664849922060966 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:322.55859375 - respo
|
||
step:58 - global_seqlen/min:20316 - global_seqlen/max:25776 - global_seqlen/minmax_diff:5460 - global_seqlen/balanced_min:22966 - global_seqlen/balanced_max:22968 - global_seqlen/mean:22967.125 - actor/entropy:0.05042630061507225 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.48229533433914185 - training/rollout_probs_diff_mean:0.0019069842528551817 - training/rollout_probs_diff_std:0.01095327828079462 - training/rollout_actor_probs_pearson_corr:0.9958901405334473 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.90234375 - self_distillation/correct_sample_fraction:0.8359375 - 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:1.0001697540283203 - rollout_corr/rollout_is_ratio_fraction_high:3.494507836876437e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00013978031347505748 - rollout_corr/rollout_is_max:2.1670167446136475 - rollout_corr/rollout_is_min:0.20139548182487488 - rollout_corr/rollout_is_std:0.029367996379733086 - rollout_corr/rollout_is_eff_sample_size:0.999138621031368 - rollout_corr/rollout_is_seq_mean:1.00016450881958 - rollout_corr/rollout_is_seq_std:0.0020526256412267685 - rollout_corr/rollout_is_seq_max:1.0117377042770386 - rollout_corr/rollout_is_seq_min:0.9942534565925598 - rollout_corr/rollout_is_seq_max_deviation:0.011737704277038574 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0489178416319191) - rollout_corr/training_log_ppl:np.float64(0.047033522495439684) - rollout_corr/kl:np.float64(0.0005489920703753981) - rollout_corr/k3_kl:np.float64(0.0007287665723376691) - rollout_corr/rollout_ppl:np.float64(1.0483349319547415) - rollout_corr/rollout_log_ppl:np.float64(0.04648453013442122) - rollout_corr/log_ppl_diff:np.float64(0.0005489923610184633) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017202749072566803) - rollout_corr/log_ppl_diff_max:np.float64(0.008941486477851868) - rollout_corr/log_ppl_diff_min:np.float64(-0.007565155625343323) - rollout_corr/ppl_ratio:np.float64(1.000551925972104) - rollout_corr/chi2_token:np.float64(0.00182512984611094) - rollout_corr/chi2_seq:np.float64(0.32337166904471815) - actor/pg_loss:np.float64(0.00039528647903352976) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00040576129458713694) - actor/ppo_kl:np.float64(0.00020430290850015353) - 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(0.8359375) - self_distillation/token_reweight_w_mean:np.float64(142336.7662561175) - self_distillation/token_reweight_w_std:np.float64(1948578.4343058746) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.00024026283063) - self_distillation/token_reweight_w_clip_frac:np.float64(0.014259436655265745) - self_distillation/empty_target_batch:np.float64(0.09765625) - actor/grad_norm:np.float64(0.18195318803191185) - perf/mfu/actor:np.float64(0.04188602836543823) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.039306640625) - actor/lr:np.float64(1e-06) - training/global_step:58 - training/epoch:2 - critic/score/mean:0.8359375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.8359375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0026587380561977625 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0026587380561977625 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:335.347
|
||
step:59 - global_seqlen/min:20116 - global_seqlen/max:24874 - global_seqlen/minmax_diff:4758 - global_seqlen/balanced_min:22054 - global_seqlen/balanced_max:22057 - global_seqlen/mean:22055.625 - actor/entropy:0.0500376932322979 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5391990542411804 - training/rollout_probs_diff_mean:0.002085200510919094 - training/rollout_probs_diff_std:0.011515925638377666 - training/rollout_actor_probs_pearson_corr:0.9953073859214783 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.75 - self_distillation/correct_sample_fraction:0.70703125 - 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:1.000190019607544 - rollout_corr/rollout_is_ratio_fraction_high:0.00011397454363759607 - rollout_corr/rollout_is_ratio_fraction_low:0.00018995757272932678 - rollout_corr/rollout_is_max:2.691387176513672 - rollout_corr/rollout_is_min:0.25212857127189636 - rollout_corr/rollout_is_std:0.031689416617155075 - rollout_corr/rollout_is_eff_sample_size:0.9989971452998435 - rollout_corr/rollout_is_seq_mean:1.0002810955047607 - rollout_corr/rollout_is_seq_std:0.0024625116493552923 - rollout_corr/rollout_is_seq_max:1.0140364170074463 - rollout_corr/rollout_is_seq_min:0.9921025037765503 - rollout_corr/rollout_is_seq_max_deviation:0.014036417007446289 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0487651862204075) - rollout_corr/training_log_ppl:np.float64(0.04702805262786569) - rollout_corr/kl:np.float64(0.00038048617622354897) - rollout_corr/k3_kl:np.float64(0.0007050287913870079) - rollout_corr/rollout_ppl:np.float64(1.0483574215322733) - rollout_corr/rollout_log_ppl:np.float64(0.04664756611236953) - rollout_corr/log_ppl_diff:np.float64(0.00038048651549615897) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018079983346979134) - rollout_corr/log_ppl_diff_max:np.float64(0.009737156331539154) - rollout_corr/log_ppl_diff_min:np.float64(-0.014218524098396301) - rollout_corr/ppl_ratio:np.float64(1.0003841938450933) - rollout_corr/chi2_token:np.float64(0.0023028196301311255) - rollout_corr/chi2_seq:np.float64(1.128127784235403) - actor/pg_loss:np.float64(0.00022415840066969395) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00020612081243598368) - actor/ppo_kl:np.float64(0.00010932901325944044) - 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(0.70703125) - self_distillation/token_reweight_w_mean:np.float64(50009.51330480375) - self_distillation/token_reweight_w_std:np.float64(780175.9249367046) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9999425960704684) - self_distillation/token_reweight_w_clip_frac:np.float64(0.009332494970294647) - self_distillation/empty_target_batch:np.float64(0.25) - actor/grad_norm:np.float64(0.09782755840569735) - perf/mfu/actor:np.float64(0.04061451686622483) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.05368041992188) - actor/lr:np.float64(1e-06) - training/global_step:59 - training/epoch:2 - critic/score/mean:0.70703125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.70703125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0017222820315510035 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.0017222820315510035 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:308.45703125 - response_length/ma
|
||
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
|
||
Given the following crystal structure parameters for the material RbLa9(IrO6)4 (Material ID: mp-560657), calculate the volume of the unit cell (in ų).
|
||
|
||
Lattice Parameters:
|
||
|
||
a = 7.82 Å
|
||
b = 7.82 Å
|
||
c = 17.88 Å
|
||
α = 90.00º
|
||
β = 90.00º
|
||
γ = 90.00º
|
||
|
||
|
||
|
||
A: 1025.67 ų
|
||
|
||
B: 1094.31 ų
|
||
|
||
C: 1200.45 ų
|
||
|
||
D: 1150.78 ų
|
||
|
||
Please reason step by step.
|
||
assistant
|
||
<think>
|
||
|
||
</think>
|
||
|
||
|
||
[response] <reasoning>
|
||
The volume $ V $ of a unit cell in a crystal structure can be calculated using the formula:
|
||
|
||
$$
|
||
V = a \cdot b \cdot c \cdot \sqrt{1 + 2\cos\alpha\cos\beta\cos\gamma - \cos^2\alpha - \cos^2\beta - \cos^2\gamma}
|
||
$$
|
||
|
||
Given that all angles $ \alpha, \beta, \gamma $ are 90.00º, the cosine of each angle is 0. Substituting these values into the formula simplifies it to:
|
||
|
||
$$
|
||
V = a \cdot b \cdot c
|
||
$$
|
||
|
||
Using the given lattice parameters:
|
||
|
||
$$
|
||
a = 7.82 \, \text{Å}, \quad b = 7.82 \, \text{Å}, \quad c = 17.88 \, \text{Å}
|
||
$$
|
||
|
||
$$
|
||
V = 7.82 \times 7.82 \times 17.88
|
||
$$
|
||
|
||
First, calculate $ 7.82 \times 7.82 $:
|
||
|
||
$$
|
||
7.82 \times 7.82 = 61.1524
|
||
$$
|
||
|
||
Then multiply by 17.88:
|
||
|
||
$$
|
||
61.1524 \times 17.88 \approx 1094.31
|
||
$$
|
||
|
||
Thus, the volume of the unit cell is approximately 1094.31 ų.
|
||
</reasoning>
|
||
<answer>
|
||
B
|
||
</answer>
|
||
[ground_truth] B
|
||
[score] 1.0
|
||
[acc] 1.0
|
||
[pred] B
|
||
[incorrect_format] 1
|
||
[feedback]
|
||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-4B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_60
|
||
step:60 - global_seqlen/min:19789 - global_seqlen/max:26080 - global_seqlen/minmax_diff:6291 - global_seqlen/balanced_min:22973 - global_seqlen/balanced_max:22975 - global_seqlen/mean:22974.125 - actor/entropy:0.04030780494213104 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.548633337020874 - training/rollout_probs_diff_mean:0.0017152910586446524 - training/rollout_probs_diff_std:0.010967686772346497 - training/rollout_actor_probs_pearson_corr:0.9947925209999084 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65234375 - 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:0.65234375 - rollout_corr/rollout_is_mean:1.0000240802764893 - rollout_corr/rollout_is_ratio_fraction_high:3.4909295209217817e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00016291004430968314 - rollout_corr/rollout_is_max:2.5243659019470215 - rollout_corr/rollout_is_min:0.3289499878883362 - rollout_corr/rollout_is_std:0.02684214897453785 - rollout_corr/rollout_is_eff_sample_size:0.99928001780181 - rollout_corr/rollout_is_seq_mean:1.0000858306884766 - rollout_corr/rollout_is_seq_std:0.00180041731800884 - rollout_corr/rollout_is_seq_max:1.0107128620147705 - rollout_corr/rollout_is_seq_min:0.9954133033752441 - rollout_corr/rollout_is_seq_max_deviation:0.010712862014770508 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0415030517615378) - rollout_corr/training_log_ppl:np.float64(0.04021239400572085) - rollout_corr/kl:np.float64(0.0004116895854844316) - rollout_corr/k3_kl:np.float64(0.00043681999613109923) - rollout_corr/rollout_ppl:np.float64(1.0410730261355639) - rollout_corr/rollout_log_ppl:np.float64(0.039800704115805274) - rollout_corr/log_ppl_diff:np.float64(0.00041168988991557853) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012860779506809195) - rollout_corr/log_ppl_diff_max:np.float64(0.007598474621772766) - rollout_corr/log_ppl_diff_min:np.float64(-0.0053790658712387085) - rollout_corr/ppl_ratio:np.float64(1.0004133372567594) - rollout_corr/chi2_token:np.float64(0.0008693772833794355) - rollout_corr/chi2_seq:np.float64(0.27017642348073423) - actor/pg_loss:np.float64(7.360754534602165e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(4.8638998123351485e-05) - actor/ppo_kl:np.float64(8.944077894434965e-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(0.65234375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.65234375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.62890625) - self_distillation/token_reweight_w_mean:np.float64(39921.53101348528) - self_distillation/token_reweight_w_std:np.float64(599336.2581787109) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000242586946115) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0022192828328115866) - self_distillation/empty_target_batch:np.float64(0.34765625) - actor/grad_norm:np.float64(0.05733739724382758) - perf/mfu/actor:np.float64(0.04283900249480353) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.04742813110352) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7606382978723404) - val-aux/sciknoweval/reward/std@16:np.float64(0.030302917070648274) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7705531914893617) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.014673598442778796) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.7500212765957448) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.03543724585981525) - val-
|
||
step:61 - global_seqlen/min:19769 - global_seqlen/max:26095 - global_seqlen/minmax_diff:6326 - global_seqlen/balanced_min:23471 - global_seqlen/balanced_max:23472 - global_seqlen/mean:23471.5 - actor/entropy:0.04185541719198227 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4555684030056 - training/rollout_probs_diff_mean:0.0017948037711903453 - training/rollout_probs_diff_std:0.010674769058823586 - training/rollout_actor_probs_pearson_corr:0.9954621195793152 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.875 - self_distillation/correct_sample_fraction:0.83984375 - 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:1.0000584125518799 - rollout_corr/rollout_is_ratio_fraction_high:1.218739362229826e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00013406132347881794 - rollout_corr/rollout_is_max:3.2528841495513916 - rollout_corr/rollout_is_min:0.2887685298919678 - rollout_corr/rollout_is_std:0.028913918882608414 - rollout_corr/rollout_is_eff_sample_size:0.9991648025994506 - rollout_corr/rollout_is_seq_mean:1.0000717639923096 - rollout_corr/rollout_is_seq_std:0.0017382517689839005 - rollout_corr/rollout_is_seq_max:1.0059905052185059 - rollout_corr/rollout_is_seq_min:0.9953637719154358 - rollout_corr/rollout_is_seq_max_deviation:0.005990505218505859 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0430718776769936) - rollout_corr/training_log_ppl:np.float64(0.041932695094146766) - rollout_corr/kl:np.float64(0.00048433673313930115) - rollout_corr/k3_kl:np.float64(0.0005176498646184413) - rollout_corr/rollout_ppl:np.float64(1.0425654179416597) - rollout_corr/rollout_log_ppl:np.float64(0.04144835810348013) - rollout_corr/log_ppl_diff:np.float64(0.00048433699066663394) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014789276692681597) - rollout_corr/log_ppl_diff_max:np.float64(0.00832807645201683) - rollout_corr/log_ppl_diff_min:np.float64(-0.005034495145082474) - rollout_corr/ppl_ratio:np.float64(1.0004862751811743) - rollout_corr/chi2_token:np.float64(0.0011503582354635) - rollout_corr/chi2_seq:np.float64(0.4335774367209524) - actor/pg_loss:np.float64(0.0006099274614825845) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(9.566592052578926e-05) - actor/ppo_kl:np.float64(0.0001280484544565752) - 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(0.83984375) - self_distillation/token_reweight_w_mean:np.float64(108078.03998975107) - self_distillation/token_reweight_w_std:np.float64(1306800.5061330106) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0000541384797543) - self_distillation/token_reweight_w_clip_frac:np.float64(0.006611851000343449) - self_distillation/empty_target_batch:np.float64(0.125) - actor/grad_norm:np.float64(0.14281610306352377) - perf/mfu/actor:np.float64(0.04358161947414355) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.56989288330078) - actor/lr:np.float64(1e-06) - training/global_step:61 - training/epoch:2 - critic/score/mean:0.83984375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.83984375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.001662055728957057 - critic/advantages/max:0.5 - critic/advantages/min:-0.875 - critic/returns/mean:0.001662055728957057 - critic/returns/max:0.5 - critic/returns/min:-0.875 - response_length/mean:320.515625 - response_length/max:624.
|
||
step:62 - global_seqlen/min:18082 - global_seqlen/max:26343 - global_seqlen/minmax_diff:8261 - global_seqlen/balanced_min:22891 - global_seqlen/balanced_max:22894 - global_seqlen/mean:22892.875 - actor/entropy:0.04360160604119301 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9823201894760132 - training/rollout_probs_diff_mean:0.001706660958006978 - training/rollout_probs_diff_std:0.011156455613672733 - training/rollout_actor_probs_pearson_corr:0.9951454997062683 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.9375 - self_distillation/correct_sample_fraction:0.82421875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.9375 - rollout_corr/rollout_is_mean:1.000069260597229 - rollout_corr/rollout_is_ratio_fraction_high:4.773326509166509e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00011933316272916272 - rollout_corr/rollout_is_max:3.4115493297576904 - rollout_corr/rollout_is_min:0.0067672124132514 - rollout_corr/rollout_is_std:0.02733721025288105 - rollout_corr/rollout_is_eff_sample_size:0.999253473101148 - rollout_corr/rollout_is_seq_mean:1.000159740447998 - rollout_corr/rollout_is_seq_std:0.0017442916287109256 - rollout_corr/rollout_is_seq_max:1.0067665576934814 - rollout_corr/rollout_is_seq_min:0.9937759637832642 - rollout_corr/rollout_is_seq_max_deviation:0.006766557693481445 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0414578090421855) - rollout_corr/training_log_ppl:np.float64(0.039926511135945475) - rollout_corr/kl:np.float64(0.00042428443190600795) - rollout_corr/k3_kl:np.float64(0.0007075274741481508) - rollout_corr/rollout_ppl:np.float64(1.041011284571141) - rollout_corr/rollout_log_ppl:np.float64(0.03950222592038699) - rollout_corr/log_ppl_diff:np.float64(0.0004242852155584842) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015441864725289633) - rollout_corr/log_ppl_diff_max:np.float64(0.024814952164888382) - rollout_corr/log_ppl_diff_min:np.float64(-0.005558319389820099) - rollout_corr/ppl_ratio:np.float64(1.000428590690717) - rollout_corr/chi2_token:np.float64(0.001744407694786787) - rollout_corr/chi2_seq:np.float64(0.31756631121970713) - actor/pg_loss:np.float64(0.0008215298876166344) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003732443005901587) - actor/ppo_kl:np.float64(7.225550693901184e-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(0.9375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.9375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.82421875) - self_distillation/token_reweight_w_mean:np.float64(257022.4825459295) - self_distillation/token_reweight_w_std:np.float64(3306871.5474428586) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000569719588384) - self_distillation/token_reweight_w_clip_frac:np.float64(0.016503850085427985) - self_distillation/empty_target_batch:np.float64(0.0625) - actor/grad_norm:np.float64(0.21665046736598015) - perf/mfu/actor:np.float64(0.04265161260571607) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.3099021911621) - actor/lr:np.float64(1e-06) - training/global_step:62 - training/epoch:2 - critic/score/mean:0.82421875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.82421875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0036590532399713993 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0036590532399713993 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:327.33984375 - response_len
|
||
step:63 - global_seqlen/min:20795 - global_seqlen/max:25263 - global_seqlen/minmax_diff:4468 - global_seqlen/balanced_min:23732 - global_seqlen/balanced_max:23734 - global_seqlen/mean:23732.875 - actor/entropy:0.04523954167962074 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6751085519790649 - training/rollout_probs_diff_mean:0.0018330732127651572 - training/rollout_probs_diff_std:0.010520632378757 - training/rollout_actor_probs_pearson_corr:0.9957168698310852 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.74609375 - 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:0.74609375 - rollout_corr/rollout_is_mean:1.000042200088501 - rollout_corr/rollout_is_ratio_fraction_high:1.1687840924423654e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0001285662583541125 - rollout_corr/rollout_is_max:2.0695643424987793 - rollout_corr/rollout_is_min:0.005992737133055925 - rollout_corr/rollout_is_std:0.028119659051299095 - rollout_corr/rollout_is_eff_sample_size:0.9992100285399822 - rollout_corr/rollout_is_seq_mean:1.000035285949707 - rollout_corr/rollout_is_seq_std:0.001678181579336524 - rollout_corr/rollout_is_seq_max:1.0064517259597778 - rollout_corr/rollout_is_seq_min:0.992402970790863 - rollout_corr/rollout_is_seq_max_deviation:0.007597029209136963 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0453694420866668) - rollout_corr/training_log_ppl:np.float64(0.043938727694694535) - rollout_corr/kl:np.float64(0.0006067258537747477) - rollout_corr/k3_kl:np.float64(0.0005313017011587817) - rollout_corr/rollout_ppl:np.float64(1.0447322241961956) - rollout_corr/rollout_log_ppl:np.float64(0.04333200147084426) - rollout_corr/log_ppl_diff:np.float64(0.0006067262238502735) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013585495998995611) - rollout_corr/log_ppl_diff_max:np.float64(0.016803178936243057) - rollout_corr/log_ppl_diff_min:np.float64(-0.006356984376907349) - rollout_corr/ppl_ratio:np.float64(1.0006088686641306) - rollout_corr/chi2_token:np.float64(0.0007685464806854725) - rollout_corr/chi2_seq:np.float64(0.4505869532003999) - actor/pg_loss:np.float64(0.0004227783065289259) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(8.712959288459388e-05) - actor/ppo_kl:np.float64(0.00015468594890499787) - 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(0.68359375) - self_distillation/token_reweight_w_mean:np.float64(139172.26291169436) - self_distillation/token_reweight_w_std:np.float64(1867242.3280806076) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0001536742784083) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0052675219340017065) - self_distillation/empty_target_batch:np.float64(0.25390625) - actor/grad_norm:np.float64(0.1391439251601696) - perf/mfu/actor:np.float64(0.043777487128157024) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.31776428222656) - actor/lr:np.float64(1e-06) - training/global_step:63 - training/epoch:2 - 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:0.001742949360050261 - critic/advantages/max:0.875 - critic/advantages/min:-0.5 - critic/returns/mean:0.001742949360050261 - critic/returns/max:0.875 - critic/returns/min:-0.5 - response_length/mean:334.21484
|
||
step:64 - global_seqlen/min:20088 - global_seqlen/max:25585 - global_seqlen/minmax_diff:5497 - global_seqlen/balanced_min:23119 - global_seqlen/balanced_max:23121 - global_seqlen/mean:23120.25 - actor/entropy:0.042551927268505096 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5083586573600769 - training/rollout_probs_diff_mean:0.0017879114020615816 - training/rollout_probs_diff_std:0.010721568018198013 - training/rollout_actor_probs_pearson_corr:0.9953984022140503 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.7890625 - 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:1.0001569986343384 - rollout_corr/rollout_is_ratio_fraction_high:7.074637687765062e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00011791062570409849 - rollout_corr/rollout_is_max:10.94495964050293 - rollout_corr/rollout_is_min:0.27272939682006836 - rollout_corr/rollout_is_std:0.028316102921962738 - rollout_corr/rollout_is_eff_sample_size:0.9991989597077329 - rollout_corr/rollout_is_seq_mean:1.000223994255066 - rollout_corr/rollout_is_seq_std:0.0025728975888341665 - rollout_corr/rollout_is_seq_max:1.0256704092025757 - rollout_corr/rollout_is_seq_min:0.9949301481246948 - rollout_corr/rollout_is_seq_max_deviation:0.025670409202575684 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0416441741399467) - rollout_corr/training_log_ppl:np.float64(0.04043069246199593) - rollout_corr/kl:np.float64(0.0003189195853421367) - rollout_corr/k3_kl:np.float64(0.0004776339879981606) - rollout_corr/rollout_ppl:np.float64(1.0413051219657063) - rollout_corr/rollout_log_ppl:np.float64(0.040111772246746114) - rollout_corr/log_ppl_diff:np.float64(0.0003189202152498183) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013647890837091836) - rollout_corr/log_ppl_diff_max:np.float64(0.006027184426784515) - rollout_corr/log_ppl_diff_min:np.float64(-0.007712792605161667) - rollout_corr/ppl_ratio:np.float64(1.000320767518133) - rollout_corr/chi2_token:np.float64(0.0015714333858340979) - rollout_corr/chi2_seq:np.float64(0.45822112914174795) - actor/pg_loss:np.float64(0.0006036241538822651) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(9.704760395834455e-05) - actor/ppo_kl:np.float64(3.3609087846109276e-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(0.7890625) - self_distillation/token_reweight_w_mean:np.float64(25306.706190291326) - self_distillation/token_reweight_w_std:np.float64(409929.1284506519) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9997767452150583) - self_distillation/token_reweight_w_clip_frac:np.float64(0.006463439349317923) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.1090348232537508) - perf/mfu/actor:np.float64(0.04248081094347521) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.26595306396484) - actor/lr:np.float64(1e-06) - training/global_step:64 - training/epoch:2 - critic/score/mean:0.7890625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7890625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:3.537318843882531e-05 - critic/advantages/max:0.25 - critic/advantages/min:-0.75 - critic/returns/mean:3.537318843882531e-05 - critic/returns/max:0.25 - critic/returns/min:-0.75 - response_length/mean:331.2890625 - response_lengt
|
||
step:65 - global_seqlen/min:20020 - global_seqlen/max:25600 - global_seqlen/minmax_diff:5580 - global_seqlen/balanced_min:23287 - global_seqlen/balanced_max:23289 - global_seqlen/mean:23288.125 - actor/entropy:0.04803186282515526 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4967540204524994 - training/rollout_probs_diff_mean:0.001957700354978442 - training/rollout_probs_diff_std:0.01209981832653284 - training/rollout_actor_probs_pearson_corr:0.9949580430984497 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71875 - 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:0.71875 - rollout_corr/rollout_is_mean:1.0000300407409668 - rollout_corr/rollout_is_ratio_fraction_high:3.559521064744331e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0001898411283036694 - rollout_corr/rollout_is_max:2.54838228225708 - rollout_corr/rollout_is_min:0.11307276040315628 - rollout_corr/rollout_is_std:0.02980916015803814 - rollout_corr/rollout_is_eff_sample_size:0.9991120838421531 - rollout_corr/rollout_is_seq_mean:1.0000916719436646 - rollout_corr/rollout_is_seq_std:0.001912038424052298 - rollout_corr/rollout_is_seq_max:1.0090073347091675 - rollout_corr/rollout_is_seq_min:0.9943963289260864 - rollout_corr/rollout_is_seq_max_deviation:0.00900733470916748 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.047498204279691) - rollout_corr/training_log_ppl:np.float64(0.0458258127400768) - rollout_corr/kl:np.float64(0.0006745162798118542) - rollout_corr/k3_kl:np.float64(0.0006516177231024045) - rollout_corr/rollout_ppl:np.float64(1.0467801187187433) - rollout_corr/rollout_log_ppl:np.float64(0.045151295971663785) - rollout_corr/log_ppl_diff:np.float64(0.0006745167684130138) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016444117263745284) - rollout_corr/log_ppl_diff_max:np.float64(0.010444030165672302) - rollout_corr/log_ppl_diff_min:np.float64(-0.007610205560922623) - rollout_corr/ppl_ratio:np.float64(1.0006771273911) - rollout_corr/chi2_token:np.float64(0.0012553422711789608) - rollout_corr/chi2_seq:np.float64(0.21363996737636626) - actor/pg_loss:np.float64(0.0007566583808511496) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003426368589316553) - actor/ppo_kl:np.float64(0.0002434091673113592) - 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(0.71875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.62890625) - self_distillation/token_reweight_w_mean:np.float64(132425.8512902113) - self_distillation/token_reweight_w_std:np.float64(1851148.4486397996) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9997992021963) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0151770984812174) - self_distillation/empty_target_batch:np.float64(0.28125) - actor/grad_norm:np.float64(0.18636032566428185) - perf/mfu/actor:np.float64(0.0430304463908559) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.26295471191406) - actor/lr:np.float64(1e-06) - training/global_step:65 - training/epoch:2 - 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:0.0038264852482825518 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.0038264852482825518 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:329.22265625 - response_length/m
|
||
step:66 - global_seqlen/min:17558 - global_seqlen/max:25411 - global_seqlen/minmax_diff:7853 - global_seqlen/balanced_min:22498 - global_seqlen/balanced_max:22502 - global_seqlen/mean:22500.875 - actor/entropy:0.039825305342674255 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4400331974029541 - training/rollout_probs_diff_mean:0.0018288729479536414 - training/rollout_probs_diff_std:0.011651387438178062 - training/rollout_actor_probs_pearson_corr:0.994309663772583 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.72265625 - 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:1.0000613927841187 - rollout_corr/rollout_is_ratio_fraction_high:1.240217807207955e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00016122830857057124 - rollout_corr/rollout_is_max:2.01051664352417 - rollout_corr/rollout_is_min:0.2722620368003845 - rollout_corr/rollout_is_std:0.02900242619216442 - rollout_corr/rollout_is_eff_sample_size:0.9991596851850771 - rollout_corr/rollout_is_seq_mean:1.000057339668274 - rollout_corr/rollout_is_seq_std:0.0020476924255490303 - rollout_corr/rollout_is_seq_max:1.007577657699585 - rollout_corr/rollout_is_seq_min:0.9939850568771362 - rollout_corr/rollout_is_seq_max_deviation:0.007577657699584961 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0409869737923145) - rollout_corr/training_log_ppl:np.float64(0.03976573963245755) - rollout_corr/kl:np.float64(0.0005508014146560214) - rollout_corr/k3_kl:np.float64(0.0007626194736367609) - rollout_corr/rollout_ppl:np.float64(1.040397803299129) - rollout_corr/rollout_log_ppl:np.float64(0.03921493800089593) - rollout_corr/log_ppl_diff:np.float64(0.000550801631561626) - rollout_corr/log_ppl_abs_diff:np.float64(0.001862790147697524) - rollout_corr/log_ppl_diff_max:np.float64(0.012821070849895477) - rollout_corr/log_ppl_diff_min:np.float64(-0.006489671766757965) - rollout_corr/ppl_ratio:np.float64(1.0005543171428144) - rollout_corr/chi2_token:np.float64(0.002054796786978841) - rollout_corr/chi2_seq:np.float64(0.5148234020452946) - actor/pg_loss:np.float64(0.0013034224975854158) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003754316767299315) - actor/ppo_kl:np.float64(0.00012462263222268177) - 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(0.72265625) - self_distillation/token_reweight_w_mean:np.float64(184329.6079326775) - self_distillation/token_reweight_w_std:np.float64(2585825.313373127) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0004278225824237) - self_distillation/token_reweight_w_clip_frac:np.float64(0.015970634303812403) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.3207257315516472) - perf/mfu/actor:np.float64(0.041364441977581275) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.18589401245117) - actor/lr:np.float64(1e-06) - training/global_step:66 - training/epoch:2 - critic/score/mean:0.72265625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.72265625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0010541850933805108 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0010541850933805108 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:314.96484375 - response_lengt
|
||
step:67 - global_seqlen/min:19699 - global_seqlen/max:23633 - global_seqlen/minmax_diff:3934 - global_seqlen/balanced_min:21599 - global_seqlen/balanced_max:21601 - global_seqlen/mean:21600.25 - actor/entropy:0.03609660267829895 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3185198903083801 - training/rollout_probs_diff_mean:0.0016948558622971177 - training/rollout_probs_diff_std:0.01066279411315918 - training/rollout_actor_probs_pearson_corr:0.9946100115776062 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.80078125 - 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:1.0001091957092285 - rollout_corr/rollout_is_ratio_fraction_high:4.045634705107659e-05 - rollout_corr/rollout_is_ratio_fraction_low:5.394179606810212e-05 - rollout_corr/rollout_is_max:3.5274367332458496 - rollout_corr/rollout_is_min:0.22336265444755554 - rollout_corr/rollout_is_std:0.027806321159005165 - rollout_corr/rollout_is_eff_sample_size:0.9992275249362932 - rollout_corr/rollout_is_seq_mean:1.0002102851867676 - rollout_corr/rollout_is_seq_std:0.00230899965390563 - rollout_corr/rollout_is_seq_max:1.0178509950637817 - rollout_corr/rollout_is_seq_min:0.995329737663269 - rollout_corr/rollout_is_seq_max_deviation:0.01785099506378174 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0353782731108367) - rollout_corr/training_log_ppl:np.float64(0.034475101392672514) - rollout_corr/kl:np.float64(0.0006411960302656894) - rollout_corr/k3_kl:np.float64(0.0007077014818066552) - rollout_corr/rollout_ppl:np.float64(1.0347134619951248) - rollout_corr/rollout_log_ppl:np.float64(0.03383390524686547) - rollout_corr/log_ppl_diff:np.float64(0.0006411961458070436) - rollout_corr/log_ppl_abs_diff:np.float64(0.001701803797914181) - rollout_corr/log_ppl_diff_max:np.float64(0.011103812605142593) - rollout_corr/log_ppl_diff_min:np.float64(-0.006912939250469208) - rollout_corr/ppl_ratio:np.float64(1.000644029583782) - rollout_corr/chi2_token:np.float64(0.0017004713881760836) - rollout_corr/chi2_seq:np.float64(0.27700750855728984) - actor/pg_loss:np.float64(0.00027153268456459045) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00010332653255318291) - actor/ppo_kl:np.float64(0.00038016634875148725) - 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(0.80078125) - self_distillation/token_reweight_w_mean:np.float64(33479.28833753336) - self_distillation/token_reweight_w_std:np.float64(415466.60549042985) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0001371470279992) - self_distillation/token_reweight_w_clip_frac:np.float64(0.002036116988165304) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.11304023303091526) - perf/mfu/actor:np.float64(0.03992824987806824) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.28586959838867) - actor/lr:np.float64(1e-06) - training/global_step:67 - training/epoch:2 - critic/score/mean:0.80078125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.80078125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.000815869658254087 - critic/advantages/max:0.375 - critic/advantages/min:-0.625 - critic/returns/mean:0.000815869658254087 - critic/returns/max:0.375 - critic/returns/min:-0.625 - response_length/mean:289.6640625 - response_le
|
||
step:68 - global_seqlen/min:16526 - global_seqlen/max:23491 - global_seqlen/minmax_diff:6965 - global_seqlen/balanced_min:21012 - global_seqlen/balanced_max:21016 - global_seqlen/mean:21015.25 - actor/entropy:0.05100872367620468 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2822123169898987 - training/rollout_probs_diff_mean:0.001955451210960746 - training/rollout_probs_diff_std:0.010564039461314678 - training/rollout_actor_probs_pearson_corr:0.9961839318275452 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.87109375 - self_distillation/correct_sample_fraction:0.8203125 - 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:1.000083565711975 - rollout_corr/rollout_is_ratio_fraction_high:5.4052594350650907e-05 - rollout_corr/rollout_is_ratio_fraction_low:9.459204011363909e-05 - rollout_corr/rollout_is_max:4.378950119018555 - rollout_corr/rollout_is_min:0.22583216428756714 - rollout_corr/rollout_is_std:0.028775470331311226 - rollout_corr/rollout_is_eff_sample_size:0.9991726573374431 - rollout_corr/rollout_is_seq_mean:1.0001046657562256 - rollout_corr/rollout_is_seq_std:0.0021570385433733463 - rollout_corr/rollout_is_seq_max:1.0077972412109375 - rollout_corr/rollout_is_seq_min:0.9944034814834595 - rollout_corr/rollout_is_seq_max_deviation:0.0077972412109375 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0534572070464492) - rollout_corr/training_log_ppl:np.float64(0.05067192515161878) - rollout_corr/kl:np.float64(0.0007594044693028934) - rollout_corr/k3_kl:np.float64(0.0008045174472961492) - rollout_corr/rollout_ppl:np.float64(1.0526364957913756) - rollout_corr/rollout_log_ppl:np.float64(0.04991252027957671) - rollout_corr/log_ppl_diff:np.float64(0.0007594048720420687) - rollout_corr/log_ppl_abs_diff:np.float64(0.002040182679593272) - rollout_corr/log_ppl_diff_max:np.float64(0.01110206451267004) - rollout_corr/log_ppl_diff_min:np.float64(-0.00639449805021286) - rollout_corr/ppl_ratio:np.float64(1.000763491494581) - rollout_corr/chi2_token:np.float64(0.0016917763277888298) - rollout_corr/chi2_seq:np.float64(0.38484108657576144) - actor/pg_loss:np.float64(0.0016131881857290864) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00037061344210087555) - actor/ppo_kl:np.float64(0.00039652320700866994) - 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(0.8203125) - self_distillation/token_reweight_w_mean:np.float64(208727.49827101128) - self_distillation/token_reweight_w_std:np.float64(2129898.3024121746) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9995387236122042) - self_distillation/token_reweight_w_clip_frac:np.float64(0.022508326030219905) - self_distillation/empty_target_batch:np.float64(0.12890625) - actor/grad_norm:np.float64(0.2767256274819374) - perf/mfu/actor:np.float64(0.03884775522311526) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.36953353881836) - actor/lr:np.float64(1e-06) - training/global_step:68 - training/epoch:2 - critic/score/mean:0.8203125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.8203125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0013580714585259557 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0013580714585259557 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:289.07031
|
||
step:69 - global_seqlen/min:17617 - global_seqlen/max:23611 - global_seqlen/minmax_diff:5994 - global_seqlen/balanced_min:20846 - global_seqlen/balanced_max:20850 - global_seqlen/mean:20847.125 - actor/entropy:0.04170255362987518 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.430754691362381 - training/rollout_probs_diff_mean:0.0017275827703997493 - training/rollout_probs_diff_std:0.010673193261027336 - training/rollout_actor_probs_pearson_corr:0.9951395392417908 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.875 - self_distillation/correct_sample_fraction:0.828125 - 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:1.0000512599945068 - rollout_corr/rollout_is_ratio_fraction_high:4.057508340338245e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00020287542429286987 - rollout_corr/rollout_is_max:2.476072311401367 - rollout_corr/rollout_is_min:0.26981011033058167 - rollout_corr/rollout_is_std:0.028634270653128624 - rollout_corr/rollout_is_eff_sample_size:0.9991807502269672 - rollout_corr/rollout_is_seq_mean:0.9999962449073792 - rollout_corr/rollout_is_seq_std:0.0020436632912606 - rollout_corr/rollout_is_seq_max:1.011947512626648 - rollout_corr/rollout_is_seq_min:0.9935392141342163 - rollout_corr/rollout_is_seq_max_deviation:0.01194751262664795 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0424060276709497) - rollout_corr/training_log_ppl:np.float64(0.040941476891930506) - rollout_corr/kl:np.float64(0.0003941324239971067) - rollout_corr/k3_kl:np.float64(0.0005813903811582588) - rollout_corr/rollout_ppl:np.float64(1.0419914699159563) - rollout_corr/rollout_log_ppl:np.float64(0.04054734420378736) - rollout_corr/log_ppl_diff:np.float64(0.00039413268814314506) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016268106724055542) - rollout_corr/log_ppl_diff_max:np.float64(0.010919928550720215) - rollout_corr/log_ppl_diff_min:np.float64(-0.007021263241767883) - rollout_corr/ppl_ratio:np.float64(1.0003970407415181) - rollout_corr/chi2_token:np.float64(0.001484232721850276) - rollout_corr/chi2_seq:np.float64(0.34094764827750623) - actor/pg_loss:np.float64(0.00032893416937440634) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00021659644880855922) - actor/ppo_kl:np.float64(-7.26069886276548e-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(0.875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.828125) - self_distillation/token_reweight_w_mean:np.float64(98340.4069654087) - self_distillation/token_reweight_w_std:np.float64(1267416.7012998087) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0006331857293844) - self_distillation/token_reweight_w_clip_frac:np.float64(0.009909781474561896) - self_distillation/empty_target_batch:np.float64(0.125) - actor/grad_norm:np.float64(0.21244935132563114) - perf/mfu/actor:np.float64(0.03856798348514151) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.3545265197754) - actor/lr:np.float64(1e-06) - training/global_step:69 - training/epoch:2 - critic/score/mean:0.828125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.828125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.00031783815938979387 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00031783815938979387 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:288.81640625 - response_length/max:6
|
||
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
|
||
Given the following crystal structure parameters for the material RbLa9(IrO6)4 (Material ID: mp-560657), calculate the volume of the unit cell (in ų).
|
||
|
||
Lattice Parameters:
|
||
|
||
a = 7.82 Å
|
||
b = 7.82 Å
|
||
c = 17.88 Å
|
||
α = 90.00º
|
||
β = 90.00º
|
||
γ = 90.00º
|
||
|
||
|
||
|
||
A: 1025.67 ų
|
||
|
||
B: 1094.31 ų
|
||
|
||
C: 1200.45 ų
|
||
|
||
D: 1150.78 ų
|
||
|
||
Please reason step by step.
|
||
assistant
|
||
<think>
|
||
|
||
</think>
|
||
|
||
|
||
[response] <reasoning>
|
||
The volume $ V $ of a unit cell in a crystal structure can be calculated using the formula:
|
||
|
||
$$
|
||
V = a \cdot b \cdot c \cdot \sqrt{1 + 2\cos\alpha\cos\beta\cos\gamma - \cos^2\alpha - \cos^2\beta - \cos^2\gamma}
|
||
$$
|
||
|
||
Given that all angles $ \alpha, \beta, \gamma $ are 90.00º, their cosines are 0. Substituting these values into the formula simplifies it to:
|
||
|
||
$$
|
||
V = a \cdot b \cdot c
|
||
$$
|
||
|
||
Substituting the given lattice parameters:
|
||
|
||
$$
|
||
V = 7.82 \, \text{Å} \times 7.82 \, \text{Å} \times 17.88 \, \text{Å}
|
||
$$
|
||
|
||
First, calculate $ 7.82 \times 7.82 $:
|
||
|
||
$$
|
||
7.82 \times 7.82 = 61.1524
|
||
$$
|
||
|
||
Then multiply by 17.88:
|
||
|
||
$$
|
||
61.1524 \times 17.88 \approx 1094.31
|
||
$$
|
||
|
||
Thus, the volume of the unit cell is approximately 1094.31 ų.
|
||
</reasoning>
|
||
<answer>
|
||
B
|
||
</answer>
|
||
[ground_truth] B
|
||
[score] 1.0
|
||
[acc] 1.0
|
||
[pred] B
|
||
[incorrect_format] 1
|
||
[feedback]
|
||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-4B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_70
|
||
step:70 - global_seqlen/min:18920 - global_seqlen/max:23896 - global_seqlen/minmax_diff:4976 - global_seqlen/balanced_min:21167 - global_seqlen/balanced_max:21171 - global_seqlen/mean:21168.875 - actor/entropy:0.044530898332595825 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5623742341995239 - training/rollout_probs_diff_mean:0.0020385412499308586 - training/rollout_probs_diff_std:0.012092887423932552 - training/rollout_actor_probs_pearson_corr:0.9944683313369751 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.77734375 - 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:0.77734375 - rollout_corr/rollout_is_mean:0.9999082684516907 - rollout_corr/rollout_is_ratio_fraction_high:1.431864711776143e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0002577356353867799 - rollout_corr/rollout_is_max:2.489572763442993 - rollout_corr/rollout_is_min:1.5778536180732772e-05 - rollout_corr/rollout_is_std:0.0301333237439394 - rollout_corr/rollout_is_eff_sample_size:0.9990926875955936 - rollout_corr/rollout_is_seq_mean:0.9998764991760254 - rollout_corr/rollout_is_seq_std:0.0022292432840913534 - rollout_corr/rollout_is_seq_max:1.0073078870773315 - rollout_corr/rollout_is_seq_min:0.9925515055656433 - rollout_corr/rollout_is_seq_max_deviation:0.0074484944343566895 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0505306674167514) - rollout_corr/training_log_ppl:np.float64(0.04811194552348752) - rollout_corr/kl:np.float64(0.0014961613618638125) - rollout_corr/k3_kl:np.float64(0.0010580281840617545) - rollout_corr/rollout_ppl:np.float64(1.0489219408482313) - rollout_corr/rollout_log_ppl:np.float64(0.04661578337800165) - rollout_corr/log_ppl_diff:np.float64(0.0014961621454858687) - rollout_corr/log_ppl_abs_diff:np.float64(0.00244221169850789) - rollout_corr/log_ppl_diff_max:np.float64(0.03330881893634796) - rollout_corr/log_ppl_diff_min:np.float64(-0.006231758743524551) - rollout_corr/ppl_ratio:np.float64(1.0015048009809107) - rollout_corr/chi2_token:np.float64(0.0012201927602291107) - rollout_corr/chi2_seq:np.float64(0.11082092486321926) - actor/pg_loss:np.float64(0.000838939449749887) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002512547634978546) - actor/ppo_kl:np.float64(0.0007138061801215123) - 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(0.77734375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.77734375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.7109375) - self_distillation/token_reweight_w_mean:np.float64(323953.5671025575) - self_distillation/token_reweight_w_std:np.float64(3053798.2800295064) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0005679957102984) - self_distillation/token_reweight_w_clip_frac:np.float64(0.014940109875169583) - self_distillation/empty_target_batch:np.float64(0.22265625) - actor/grad_norm:np.float64(0.32167357206344604) - perf/mfu/actor:np.float64(0.03943325036909458) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.27843856811523) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7666223404255319) - val-aux/sciknoweval/reward/std@16:np.float64(0.032785963866969546) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.778436170212766) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.02531669454454758) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.7547553191489362) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.02932435814267925) - val-aux/s
|
||
step:71 - global_seqlen/min:19003 - global_seqlen/max:24898 - global_seqlen/minmax_diff:5895 - global_seqlen/balanced_min:22877 - global_seqlen/balanced_max:22879 - global_seqlen/mean:22878.5 - actor/entropy:0.041066378355026245 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7417160868644714 - training/rollout_probs_diff_mean:0.0017004754627123475 - training/rollout_probs_diff_std:0.0106857530772686 - training/rollout_actor_probs_pearson_corr:0.9951813220977783 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.78515625 - 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:0.9999306201934814 - rollout_corr/rollout_is_ratio_fraction_high:6.025548282195814e-05 - rollout_corr/rollout_is_ratio_fraction_low:9.640877397032455e-05 - rollout_corr/rollout_is_max:4.695746421813965 - rollout_corr/rollout_is_min:0.13378670811653137 - rollout_corr/rollout_is_std:0.027188541367650032 - rollout_corr/rollout_is_eff_sample_size:0.9992613292333393 - rollout_corr/rollout_is_seq_mean:0.9999163746833801 - rollout_corr/rollout_is_seq_std:0.00201141694560647 - rollout_corr/rollout_is_seq_max:1.0119117498397827 - rollout_corr/rollout_is_seq_min:0.9936652779579163 - rollout_corr/rollout_is_seq_max_deviation:0.011911749839782715 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0394931584596634) - rollout_corr/training_log_ppl:np.float64(0.03842429356336652) - rollout_corr/kl:np.float64(0.00044337915408476647) - rollout_corr/k3_kl:np.float64(0.0005245370683890638) - rollout_corr/rollout_ppl:np.float64(1.0390328136272728) - rollout_corr/rollout_log_ppl:np.float64(0.03798091409771587) - rollout_corr/log_ppl_diff:np.float64(0.00044337946565065067) - rollout_corr/log_ppl_abs_diff:np.float64(0.001516479414931382) - rollout_corr/log_ppl_diff_max:np.float64(0.00839998945593834) - rollout_corr/log_ppl_diff_min:np.float64(-0.007669523358345032) - rollout_corr/ppl_ratio:np.float64(1.000445791753009) - rollout_corr/chi2_token:np.float64(0.0013469457626342773) - rollout_corr/chi2_seq:np.float64(0.7061824975535274) - actor/pg_loss:np.float64(0.00031901232432574034) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(5.6846769439289346e-05) - actor/ppo_kl:np.float64(-8.79328427032533e-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(0.78515625) - self_distillation/token_reweight_w_mean:np.float64(78056.61540558911) - self_distillation/token_reweight_w_std:np.float64(985978.0488043064) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9999284080695361) - self_distillation/token_reweight_w_clip_frac:np.float64(0.003921596973668784) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.06964961811900139) - perf/mfu/actor:np.float64(0.04247672842872548) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.3838882446289) - actor/lr:np.float64(1e-06) - training/global_step:71 - training/epoch:2 - critic/score/mean:0.78515625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.78515625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0023409256245940924 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.0023409256245940924 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:324.140625 - response_length/
|
||
step:72 - global_seqlen/min:21927 - global_seqlen/max:24576 - global_seqlen/minmax_diff:2649 - global_seqlen/balanced_min:22883 - global_seqlen/balanced_max:22884 - global_seqlen/mean:22883.375 - actor/entropy:0.03342093899846077 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5599771738052368 - training/rollout_probs_diff_mean:0.0015238827327266335 - training/rollout_probs_diff_std:0.010387895628809929 - training/rollout_actor_probs_pearson_corr:0.9947546124458313 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83984375 - self_distillation/correct_sample_fraction:0.7890625 - 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:1.0000593662261963 - rollout_corr/rollout_is_ratio_fraction_high:6.152558489702642e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00019688188331201673 - rollout_corr/rollout_is_max:2.7454488277435303 - rollout_corr/rollout_is_min:0.009296300821006298 - rollout_corr/rollout_is_std:0.027067698538303375 - rollout_corr/rollout_is_eff_sample_size:0.9992679951393448 - rollout_corr/rollout_is_seq_mean:1.0000627040863037 - rollout_corr/rollout_is_seq_std:0.0017418585484847426 - rollout_corr/rollout_is_seq_max:1.0082443952560425 - rollout_corr/rollout_is_seq_min:0.9941825270652771 - rollout_corr/rollout_is_seq_max_deviation:0.00824439525604248 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0355365513823926) - rollout_corr/training_log_ppl:np.float64(0.03459935712726292) - rollout_corr/kl:np.float64(0.0005253947009382642) - rollout_corr/k3_kl:np.float64(0.0011788816827351667) - rollout_corr/rollout_ppl:np.float64(1.0349831562489271) - rollout_corr/rollout_log_ppl:np.float64(0.03407396220063674) - rollout_corr/log_ppl_diff:np.float64(0.000525394926626177) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018047186922558467) - rollout_corr/log_ppl_diff_max:np.float64(0.01845739781856537) - rollout_corr/log_ppl_diff_min:np.float64(-0.006578452885150909) - rollout_corr/ppl_ratio:np.float64(1.000529364682734) - rollout_corr/chi2_token:np.float64(0.008092012256383896) - rollout_corr/chi2_seq:np.float64(0.6922405788209289) - actor/pg_loss:np.float64(0.0009021894074976444) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0004105189245819929) - actor/ppo_kl:np.float64(-7.494122602480502e-06) - actor/pg_clipfrac_lower:np.float64(1.333191084995633e-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(0.83984375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.83984375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.7890625) - self_distillation/token_reweight_w_mean:np.float64(197008.8856558064) - self_distillation/token_reweight_w_std:np.float64(2194593.5106205577) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.999858915572986) - self_distillation/token_reweight_w_clip_frac:np.float64(0.009888861561194062) - self_distillation/empty_target_batch:np.float64(0.16015625) - actor/grad_norm:np.float64(0.37002852745354176) - perf/mfu/actor:np.float64(0.04237542423474614) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.3786392211914) - actor/lr:np.float64(1e-06) - training/global_step:72 - training/epoch:2 - critic/score/mean:0.7890625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7890625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0017073351191356778 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0017073351191356778 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_l
|
||
step:73 - global_seqlen/min:18163 - global_seqlen/max:25690 - global_seqlen/minmax_diff:7527 - global_seqlen/balanced_min:21913 - global_seqlen/balanced_max:21917 - global_seqlen/mean:21916.0 - actor/entropy:0.034132421016693115 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.35351723432540894 - training/rollout_probs_diff_mean:0.0017482901457697153 - training/rollout_probs_diff_std:0.011693055741488934 - training/rollout_actor_probs_pearson_corr:0.9933797121047974 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71484375 - 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:0.71484375 - rollout_corr/rollout_is_mean:1.000124454498291 - rollout_corr/rollout_is_ratio_fraction_high:9.308510925620794e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.3297872101247776e-05 - rollout_corr/rollout_is_max:4.1709747314453125 - rollout_corr/rollout_is_min:0.43125781416893005 - rollout_corr/rollout_is_std:0.029188847169280052 - rollout_corr/rollout_is_eff_sample_size:0.9991488554809967 - rollout_corr/rollout_is_seq_mean:1.0000839233398438 - rollout_corr/rollout_is_seq_std:0.0020293211564421654 - rollout_corr/rollout_is_seq_max:1.0085396766662598 - rollout_corr/rollout_is_seq_min:0.9942524433135986 - rollout_corr/rollout_is_seq_max_deviation:0.008539676666259766 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0363794229924679) - rollout_corr/training_log_ppl:np.float64(0.03516461659455672) - rollout_corr/kl:np.float64(0.0003281725150534598) - rollout_corr/k3_kl:np.float64(0.000549684357956437) - rollout_corr/rollout_ppl:np.float64(1.0360313290730119) - rollout_corr/rollout_log_ppl:np.float64(0.03483644372136041) - rollout_corr/log_ppl_diff:np.float64(0.0003281728731963085) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015313210878957761) - rollout_corr/log_ppl_diff_max:np.float64(0.010771263390779495) - rollout_corr/log_ppl_diff_min:np.float64(-0.006783545948565006) - rollout_corr/ppl_ratio:np.float64(1.0003303056582808) - rollout_corr/chi2_token:np.float64(0.0016802684403955936) - rollout_corr/chi2_seq:np.float64(0.5110847533214837) - actor/pg_loss:np.float64(0.00020943302661180496) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(7.329807704081759e-05) - actor/ppo_kl:np.float64(-7.272707647376642e-05) - actor/pg_clipfrac_lower:np.float64(1.268262985831825e-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(0.71484375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71484375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6875) - self_distillation/token_reweight_w_mean:np.float64(30931.851405258058) - self_distillation/token_reweight_w_std:np.float64(314383.1071442036) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0003154934383929) - self_distillation/token_reweight_w_clip_frac:np.float64(0.002947135504655307) - self_distillation/empty_target_batch:np.float64(0.28515625) - actor/grad_norm:np.float64(0.05511173512786627) - perf/mfu/actor:np.float64(0.04059150667566496) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.59348678588867) - actor/lr:np.float64(1e-06) - training/global_step:73 - training/epoch:2 - 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:0.00014461435785051435 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00014461435785051435 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length
|
||
step:74 - global_seqlen/min:18699 - global_seqlen/max:24638 - global_seqlen/minmax_diff:5939 - global_seqlen/balanced_min:22666 - global_seqlen/balanced_max:22668 - global_seqlen/mean:22667.375 - actor/entropy:0.04269804805517197 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4211313724517822 - training/rollout_probs_diff_mean:0.0018847070168703794 - training/rollout_probs_diff_std:0.010922428220510483 - training/rollout_actor_probs_pearson_corr:0.9953285455703735 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.84375 - self_distillation/correct_sample_fraction:0.84375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.84375 - rollout_corr/rollout_is_mean:1.0002391338348389 - rollout_corr/rollout_is_ratio_fraction_high:6.0671511164400727e-05 - rollout_corr/rollout_is_ratio_fraction_low:9.707442222861573e-05 - rollout_corr/rollout_is_max:5.670766830444336 - rollout_corr/rollout_is_min:0.28868961334228516 - rollout_corr/rollout_is_std:0.029143888503313065 - rollout_corr/rollout_is_eff_sample_size:0.9991514736296966 - rollout_corr/rollout_is_seq_mean:1.0002909898757935 - rollout_corr/rollout_is_seq_std:0.0020770577248185873 - rollout_corr/rollout_is_seq_max:1.013790249824524 - rollout_corr/rollout_is_seq_min:0.9944962859153748 - rollout_corr/rollout_is_seq_max_deviation:0.013790249824523926 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0447109127417207) - rollout_corr/training_log_ppl:np.float64(0.04343737147610227) - rollout_corr/kl:np.float64(0.000363680085879281) - rollout_corr/k3_kl:np.float64(0.0005548008483327038) - rollout_corr/rollout_ppl:np.float64(1.044323279056698) - rollout_corr/rollout_log_ppl:np.float64(0.043073691116660484) - rollout_corr/log_ppl_diff:np.float64(0.000363680359441787) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015776339314470533) - rollout_corr/log_ppl_diff_max:np.float64(0.0066324397921562195) - rollout_corr/log_ppl_diff_min:np.float64(-0.008656047284603119) - rollout_corr/ppl_ratio:np.float64(1.0003659820649773) - rollout_corr/chi2_token:np.float64(0.0016228372696787119) - rollout_corr/chi2_seq:np.float64(0.7596687572076917) - actor/pg_loss:np.float64(0.0) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0) - actor/ppo_kl:np.float64(0.00016098882441184514) - 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(0.84375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.84375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.84375) - self_distillation/token_reweight_w_mean:np.float64(0.999999993480742) - self_distillation/token_reweight_w_std:np.float64(0.0) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.999999993480742) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.15625) - actor/grad_norm:np.float64(0.0) - perf/mfu/actor:np.float64(0.04193188049080382) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.48727798461914) - actor/lr:np.float64(1e-06) - training/global_step:74 - training/epoch:2 - critic/score/mean:0.84375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.84375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0 - critic/advantages/max:0.0 - critic/advantages/min:0.0 - critic/returns/mean:0.0 - critic/returns/max:0.0 - critic/returns/min:0.0 - response_length/mean:321.91796875 - response_length/max:589.0 - response_length/min:142.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:321.91796875 - respon
|
||
step:75 - global_seqlen/min:20181 - global_seqlen/max:24422 - global_seqlen/minmax_diff:4241 - global_seqlen/balanced_min:22420 - global_seqlen/balanced_max:22422 - global_seqlen/mean:22421.0 - actor/entropy:0.0377618744969368 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7632365226745605 - training/rollout_probs_diff_mean:0.0016583539545536041 - training/rollout_probs_diff_std:0.011271195486187935 - training/rollout_actor_probs_pearson_corr:0.9941364526748657 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.68359375 - 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:0.68359375 - rollout_corr/rollout_is_mean:0.9999065399169922 - rollout_corr/rollout_is_ratio_fraction_high:5.097359462524764e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00015292078023776412 - rollout_corr/rollout_is_max:3.3628957271575928 - rollout_corr/rollout_is_min:0.10317613184452057 - rollout_corr/rollout_is_std:0.026788802817463875 - rollout_corr/rollout_is_eff_sample_size:0.9992826366379459 - rollout_corr/rollout_is_seq_mean:0.9999075531959534 - rollout_corr/rollout_is_seq_std:0.0017399507341906428 - rollout_corr/rollout_is_seq_max:1.0067952871322632 - rollout_corr/rollout_is_seq_min:0.9949383735656738 - rollout_corr/rollout_is_seq_max_deviation:0.006795287132263184 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0384952994063497) - rollout_corr/training_log_ppl:np.float64(0.03707945561973247) - rollout_corr/kl:np.float64(0.0004538998949357165) - rollout_corr/k3_kl:np.float64(0.00041609512296765516) - rollout_corr/rollout_ppl:np.float64(1.038021398242563) - rollout_corr/rollout_log_ppl:np.float64(0.03662555549408353) - rollout_corr/log_ppl_diff:np.float64(0.0004539001256489428) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012520181262516417) - rollout_corr/log_ppl_diff_max:np.float64(0.006798019632697105) - rollout_corr/log_ppl_diff_min:np.float64(-0.00393272191286087) - rollout_corr/ppl_ratio:np.float64(1.0004553578328341) - rollout_corr/chi2_token:np.float64(0.0007796948775649071) - rollout_corr/chi2_seq:np.float64(0.202988384058699) - actor/pg_loss:np.float64(8.659297600388527e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(3.3403119232389145e-05) - actor/ppo_kl:np.float64(-7.427042106478154e-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(0.68359375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.68359375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.66015625) - self_distillation/token_reweight_w_mean:np.float64(65656.72851561918) - self_distillation/token_reweight_w_std:np.float64(913989.140625) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0001852267887443) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0024617357703391463) - self_distillation/empty_target_batch:np.float64(0.31640625) - actor/grad_norm:np.float64(0.07150154281407595) - perf/mfu/actor:np.float64(0.04083979931261415) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.35407638549805) - actor/lr:np.float64(1e-06) - training/global_step:75 - training/epoch:2 - 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:-0.00022141655790619552 - critic/advantages/max:0.875 - critic/advantages/min:-0.125 - critic/returns/mean:-0.00022141655790619552 - critic/returns/max:0.875 - critic/returns/min:-0.125 - response_length/mean:30
|
||
step:76 - global_seqlen/min:17637 - global_seqlen/max:23218 - global_seqlen/minmax_diff:5581 - global_seqlen/balanced_min:20738 - global_seqlen/balanced_max:20744 - global_seqlen/mean:20742.5 - actor/entropy:0.04465673118829727 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.29321733117103577 - training/rollout_probs_diff_mean:0.0018123796908184886 - training/rollout_probs_diff_std:0.010481405071914196 - training/rollout_actor_probs_pearson_corr:0.995635449886322 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71484375 - self_distillation/correct_sample_fraction:0.69140625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.71484375 - rollout_corr/rollout_is_mean:1.0001299381256104 - rollout_corr/rollout_is_ratio_fraction_high:1.4342263966682367e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.4342263966682367e-05 - rollout_corr/rollout_is_max:2.523247241973877 - rollout_corr/rollout_is_min:0.47835788130760193 - rollout_corr/rollout_is_std:0.027491578832268715 - rollout_corr/rollout_is_eff_sample_size:0.9992450219453466 - rollout_corr/rollout_is_seq_mean:1.0001345872879028 - rollout_corr/rollout_is_seq_std:0.0018495450494810939 - rollout_corr/rollout_is_seq_max:1.0072492361068726 - rollout_corr/rollout_is_seq_min:0.9933908581733704 - rollout_corr/rollout_is_seq_max_deviation:0.007249236106872559 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.043047591112554) - rollout_corr/training_log_ppl:np.float64(0.04124056530963571) - rollout_corr/kl:np.float64(0.0005134855972919183) - rollout_corr/k3_kl:np.float64(0.000566033174557834) - rollout_corr/rollout_ppl:np.float64(1.0424995855428278) - rollout_corr/rollout_log_ppl:np.float64(0.04072707948353127) - rollout_corr/log_ppl_diff:np.float64(0.0005134858261044428) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018579333004709042) - rollout_corr/log_ppl_diff_max:np.float64(0.0113980732858181) - rollout_corr/log_ppl_diff_min:np.float64(-0.010368887335062027) - rollout_corr/ppl_ratio:np.float64(1.0005171394441277) - rollout_corr/chi2_token:np.float64(0.0013373454567044973) - rollout_corr/chi2_seq:np.float64(0.5531120426021516) - actor/pg_loss:np.float64(4.2089493945240974e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00014904459294484695) - actor/ppo_kl:np.float64(0.00027603819385291217) - actor/pg_clipfrac_lower:np.float64(1.641281596675981e-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(0.71484375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71484375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.69140625) - self_distillation/token_reweight_w_mean:np.float64(104931.70410155761) - self_distillation/token_reweight_w_std:np.float64(1154588.890625) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.000077618053183) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0010802401811815798) - self_distillation/empty_target_batch:np.float64(0.28515625) - actor/grad_norm:np.float64(0.06183192692697048) - perf/mfu/actor:np.float64(0.038350335680134795) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.40346145629883) - actor/lr:np.float64(1e-06) - training/global_step:76 - training/epoch:2 - critic/score/mean:0.69140625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.69140625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.000932247145101428 - critic/advantages/max:0.875 - critic/advantages/min:-0.125 - critic/returns/mean:0.000932247145101428 - critic/returns/max:0.875 - critic/returns/min:-0.125 - response
|
||
step:77 - global_seqlen/min:18473 - global_seqlen/max:24783 - global_seqlen/minmax_diff:6310 - global_seqlen/balanced_min:22922 - global_seqlen/balanced_max:22924 - global_seqlen/mean:22923.0 - actor/entropy:0.029721224680542946 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7918662428855896 - training/rollout_probs_diff_mean:0.001373450388200581 - training/rollout_probs_diff_std:0.01168912835419178 - training/rollout_actor_probs_pearson_corr:0.9923051595687866 - self_distillation/success_group_fraction:0.5 - self_distillation/success_sample_fraction:0.5 - 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:0.5 - rollout_corr/rollout_is_mean:0.9999530911445618 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.00025326714967377484 - rollout_corr/rollout_is_max:1.9884089231491089 - rollout_corr/rollout_is_min:0.15751142799854279 - rollout_corr/rollout_is_std:0.02383342757821083 - rollout_corr/rollout_is_eff_sample_size:0.9994322902127667 - rollout_corr/rollout_is_seq_mean:0.9998948574066162 - rollout_corr/rollout_is_seq_std:0.001541188801638782 - rollout_corr/rollout_is_seq_max:1.0081614255905151 - rollout_corr/rollout_is_seq_min:0.9954944849014282 - rollout_corr/rollout_is_seq_max_deviation:0.008161425590515137 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.028552311938256) - rollout_corr/training_log_ppl:np.float64(0.02791491772450172) - rollout_corr/kl:np.float64(0.0004828657112110335) - rollout_corr/k3_kl:np.float64(0.0004112039243029386) - rollout_corr/rollout_ppl:np.float64(1.028052968904376) - rollout_corr/rollout_log_ppl:np.float64(0.027432051831056015) - rollout_corr/log_ppl_diff:np.float64(0.0004828658934457053) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012247920726622397) - rollout_corr/log_ppl_diff_max:np.float64(0.006324823945760727) - rollout_corr/log_ppl_diff_min:np.float64(-0.0035044103860855103) - rollout_corr/ppl_ratio:np.float64(1.000484426273033) - rollout_corr/chi2_token:np.float64(0.0005913209170103073) - rollout_corr/chi2_seq:np.float64(0.18623502389527857) - actor/pg_loss:np.float64(0.0) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0) - actor/ppo_kl:np.float64(-2.6408372267638924e-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(0.5) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.5) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.5) - self_distillation/token_reweight_w_mean:np.float64(0.9999999969732016) - self_distillation/token_reweight_w_std:np.float64(0.0) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9999999969732016) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.5) - actor/grad_norm:np.float64(0.0) - perf/mfu/actor:np.float64(0.042214026092317475) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.41495895385742) - actor/lr:np.float64(1e-06) - training/global_step:77 - training/epoch:2 - 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:0.0 - critic/advantages/max:0.0 - critic/advantages/min:0.0 - critic/returns/mean:0.0 - critic/returns/max:0.0 - critic/returns/min:0.0 - response_length/mean:308.46875 - response_length/max:617.0 - response_length/min:144.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:308.46875 - response_length_non_aborted/max:617.0 - response_length_non_aborted/
|
||
step:78 - global_seqlen/min:21618 - global_seqlen/max:23393 - global_seqlen/minmax_diff:1775 - global_seqlen/balanced_min:22640 - global_seqlen/balanced_max:22641 - global_seqlen/mean:22640.375 - actor/entropy:0.035092517733573914 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.35351723432540894 - training/rollout_probs_diff_mean:0.0017425608821213245 - training/rollout_probs_diff_std:0.011111678555607796 - training/rollout_actor_probs_pearson_corr:0.994017481803894 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65625 - 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:0.65625 - rollout_corr/rollout_is_mean:0.9998819828033447 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:3.965578798670322e-05 - rollout_corr/rollout_is_max:1.9659780263900757 - rollout_corr/rollout_is_min:0.4614439904689789 - rollout_corr/rollout_is_std:0.028242332860827446 - rollout_corr/rollout_is_eff_sample_size:0.9992027683105108 - rollout_corr/rollout_is_seq_mean:0.9999055862426758 - rollout_corr/rollout_is_seq_std:0.0017442753305658698 - rollout_corr/rollout_is_seq_max:1.0089272260665894 - rollout_corr/rollout_is_seq_min:0.994486927986145 - rollout_corr/rollout_is_seq_max_deviation:0.008927226066589355 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0356996422633529) - rollout_corr/training_log_ppl:np.float64(0.0347957349367789) - rollout_corr/kl:np.float64(0.0006287374900068726) - rollout_corr/k3_kl:np.float64(0.00041870254129960927) - rollout_corr/rollout_ppl:np.float64(1.0350423878990114) - rollout_corr/rollout_log_ppl:np.float64(0.03416699716126459) - rollout_corr/log_ppl_diff:np.float64(0.0006287377755143098) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013772529555353685) - rollout_corr/log_ppl_diff_max:np.float64(0.009740196168422699) - rollout_corr/log_ppl_diff_min:np.float64(-0.006851349025964737) - rollout_corr/ppl_ratio:np.float64(1.0006305843126029) - rollout_corr/chi2_token:np.float64(0.0004125772975385189) - rollout_corr/chi2_seq:np.float64(0.027380227576941252) - actor/pg_loss:np.float64(0.0002134254900738597) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0001747056612657616) - actor/ppo_kl:np.float64(0.00013020554026610398) - 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(0.65625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.65625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6328125) - self_distillation/token_reweight_w_mean:np.float64(73278.90783314756) - self_distillation/token_reweight_w_std:np.float64(810226.5895272555) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9998990218155086) - self_distillation/token_reweight_w_clip_frac:np.float64(0.003775067198148463) - self_distillation/empty_target_batch:np.float64(0.34375) - actor/grad_norm:np.float64(0.09012467134743929) - perf/mfu/actor:np.float64(0.04218577321591646) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.53057479858398) - actor/lr:np.float64(1e-06) - training/global_step:78 - training/epoch:2 - 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:0.0010013086721301079 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:0.0010013086721301079 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:295.51171875 - response_length/max:
|
||
step:79 - global_seqlen/min:19064 - global_seqlen/max:24752 - global_seqlen/minmax_diff:5688 - global_seqlen/balanced_min:22609 - global_seqlen/balanced_max:22611 - global_seqlen/mean:22609.875 - actor/entropy:0.03175250440835953 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7543225288391113 - training/rollout_probs_diff_mean:0.0016824816120788455 - training/rollout_probs_diff_std:0.01265895739197731 - training/rollout_actor_probs_pearson_corr:0.9913443922996521 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.80078125 - 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:1.0001208782196045 - rollout_corr/rollout_is_ratio_fraction_high:7.674302469240502e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00015348604938481003 - rollout_corr/rollout_is_max:4.387160778045654 - rollout_corr/rollout_is_min:0.029560457915067673 - rollout_corr/rollout_is_std:0.02866755798459053 - rollout_corr/rollout_is_eff_sample_size:0.9991788460058731 - rollout_corr/rollout_is_seq_mean:1.0002002716064453 - rollout_corr/rollout_is_seq_std:0.00217239186167717 - rollout_corr/rollout_is_seq_max:1.0119655132293701 - rollout_corr/rollout_is_seq_min:0.9956554174423218 - rollout_corr/rollout_is_seq_max_deviation:0.011965513229370117 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.033602093346417) - rollout_corr/training_log_ppl:np.float64(0.03276922370059765) - rollout_corr/kl:np.float64(0.0006594466274378874) - rollout_corr/k3_kl:np.float64(0.000879152259249949) - rollout_corr/rollout_ppl:np.float64(1.0329237426631153) - rollout_corr/rollout_log_ppl:np.float64(0.03210977676053517) - rollout_corr/log_ppl_diff:np.float64(0.0006594469382434909) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018515425176701683) - rollout_corr/log_ppl_diff_max:np.float64(0.02449878305196762) - rollout_corr/log_ppl_diff_min:np.float64(-0.008788060396909714) - rollout_corr/ppl_ratio:np.float64(1.0006653296295553) - rollout_corr/chi2_token:np.float64(0.001893435139209032) - rollout_corr/chi2_seq:np.float64(0.5223376241046935) - actor/pg_loss:np.float64(0.0001657054526731372) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(7.394158183160471e-05) - actor/ppo_kl:np.float64(0.00019161261689504272) - 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(0.80078125) - self_distillation/token_reweight_w_mean:np.float64(32477.99980539549) - self_distillation/token_reweight_w_std:np.float64(403046.4446096127) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0000050456728786) - self_distillation/token_reweight_w_clip_frac:np.float64(0.002672293430805439) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.09365954529494047) - perf/mfu/actor:np.float64(0.04209851712862714) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.6111183166504) - actor/lr:np.float64(1e-06) - training/global_step:79 - training/epoch:3 - critic/score/mean:0.80078125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.80078125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0006427228217944503 - critic/advantages/max:0.25 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0006427228217944503 - critic/returns/max:0.25 - critic/returns/min:-0.875 - response_length/mean:305.40234375 - response_leng
|
||
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
|
||
Given the following crystal structure parameters for the material RbLa9(IrO6)4 (Material ID: mp-560657), calculate the volume of the unit cell (in ų).
|
||
|
||
Lattice Parameters:
|
||
|
||
a = 7.82 Å
|
||
b = 7.82 Å
|
||
c = 17.88 Å
|
||
α = 90.00º
|
||
β = 90.00º
|
||
γ = 90.00º
|
||
|
||
|
||
|
||
A: 1025.67 ų
|
||
|
||
B: 1094.31 ų
|
||
|
||
C: 1200.45 ų
|
||
|
||
D: 1150.78 ų
|
||
|
||
Please reason step by step.
|
||
assistant
|
||
<think>
|
||
|
||
</think>
|
||
|
||
|
||
[response] <reasoning>
|
||
The volume $ V $ of a unit cell in a crystal structure can be calculated using the formula:
|
||
|
||
$$
|
||
V = a \cdot b \cdot c \cdot \sqrt{1 + 2\cos\alpha\cos\beta\cos\gamma - \cos^2\alpha - \cos^2\beta - \cos^2\gamma}
|
||
$$
|
||
|
||
Given that all angles $ \alpha, \beta, \gamma $ are 90.00º, the cosine of each angle is 0. Substituting these values into the formula:
|
||
|
||
$$
|
||
V = a \cdot b \cdot c \cdot \sqrt{1 + 2(0)(0)(0) - (0)^2 - (0)^2 - (0)^2} = a \cdot b \cdot c \cdot \sqrt{1} = a \cdot b \cdot c
|
||
$$
|
||
|
||
Now, substituting the given lattice parameters:
|
||
|
||
$$
|
||
V = 7.82 \cdot 7.82 \cdot 17.88
|
||
$$
|
||
|
||
First, calculate $ 7.82 \cdot 7.82 $:
|
||
|
||
$$
|
||
7.82 \cdot 7.82 = 61.1524
|
||
$$
|
||
|
||
Then multiply by 17.88:
|
||
|
||
$$
|
||
61.1524 \cdot 17.88 \approx 1094.31
|
||
$$
|
||
|
||
Thus, the volume of the unit cell is approximately 1094.31 ų.
|
||
</reasoning>
|
||
<answer>
|
||
B
|
||
</answer>
|
||
[ground_truth] B
|
||
[score] 1.0
|
||
[acc] 1.0
|
||
[pred] B
|
||
[incorrect_format] 1
|
||
[feedback]
|
||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-4B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_80
|
||
step:80 - global_seqlen/min:19660 - global_seqlen/max:25315 - global_seqlen/minmax_diff:5655 - global_seqlen/balanced_min:22802 - global_seqlen/balanced_max:22805 - global_seqlen/mean:22803.375 - actor/entropy:0.050714973360300064 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6335345506668091 - training/rollout_probs_diff_mean:0.001896037021651864 - training/rollout_probs_diff_std:0.011774555779993534 - training/rollout_actor_probs_pearson_corr:0.9953871369361877 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.74609375 - self_distillation/correct_sample_fraction:0.69140625 - 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:1.000020146369934 - rollout_corr/rollout_is_ratio_fraction_high:9.482830500928685e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0002607778296805918 - rollout_corr/rollout_is_max:5.18784761428833 - rollout_corr/rollout_is_min:0.19069860875606537 - rollout_corr/rollout_is_std:0.03135092556476593 - rollout_corr/rollout_is_eff_sample_size:0.9990177275875637 - rollout_corr/rollout_is_seq_mean:1.0000741481781006 - rollout_corr/rollout_is_seq_std:0.0026087467558681965 - rollout_corr/rollout_is_seq_max:1.0172052383422852 - rollout_corr/rollout_is_seq_min:0.9944260120391846 - rollout_corr/rollout_is_seq_max_deviation:0.017205238342285156 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.048643831629306) - rollout_corr/training_log_ppl:np.float64(0.046666585320053855) - rollout_corr/kl:np.float64(0.0012581646745264408) - rollout_corr/k3_kl:np.float64(0.0014205996778304808) - rollout_corr/rollout_ppl:np.float64(1.0472968486137688) - rollout_corr/rollout_log_ppl:np.float64(0.04540842036749382) - rollout_corr/log_ppl_diff:np.float64(0.001258164952560037) - rollout_corr/log_ppl_abs_diff:np.float64(0.002346423177186807) - rollout_corr/log_ppl_diff_max:np.float64(0.030772797763347626) - rollout_corr/log_ppl_diff_min:np.float64(-0.010465197265148163) - rollout_corr/ppl_ratio:np.float64(1.0012677058111876) - rollout_corr/chi2_token:np.float64(0.0037672733888030052) - rollout_corr/chi2_seq:np.float64(0.9763107842300087) - actor/pg_loss:np.float64(0.000655685318633914) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002488995287421858) - actor/ppo_kl:np.float64(0.0006420044184398677) - 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(0.69140625) - self_distillation/token_reweight_w_mean:np.float64(110585.75917376974) - self_distillation/token_reweight_w_std:np.float64(1607578.4135247644) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0001894689630717) - self_distillation/token_reweight_w_clip_frac:np.float64(0.011867681285366416) - self_distillation/empty_target_batch:np.float64(0.25390625) - actor/grad_norm:np.float64(0.17190873017534614) - perf/mfu/actor:np.float64(0.041936679075923314) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(146.61267471313477) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7579787234042553) - val-aux/sciknoweval/reward/std@16:np.float64(0.020546155153965868) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7650425531914894) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.011835424325934357) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.7508829787234043) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.02186188398559465) - val-aux/scik
|
||
step:81 - global_seqlen/min:18437 - global_seqlen/max:24732 - global_seqlen/minmax_diff:6295 - global_seqlen/balanced_min:20958 - global_seqlen/balanced_max:20960 - global_seqlen/mean:20959.375 - actor/entropy:0.03692462667822838 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.669526994228363 - training/rollout_probs_diff_mean:0.0017158862901851535 - training/rollout_probs_diff_std:0.011773472651839256 - training/rollout_actor_probs_pearson_corr:0.9934026002883911 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.75 - 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:0.75 - rollout_corr/rollout_is_mean:1.0001301765441895 - rollout_corr/rollout_is_ratio_fraction_high:0.00014152478252071887 - rollout_corr/rollout_is_ratio_fraction_low:0.0001981347013497725 - rollout_corr/rollout_is_max:3.121652126312256 - rollout_corr/rollout_is_min:0.06086057052016258 - rollout_corr/rollout_is_std:0.02928262948989868 - rollout_corr/rollout_is_eff_sample_size:0.9991435002195741 - rollout_corr/rollout_is_seq_mean:1.0000874996185303 - rollout_corr/rollout_is_seq_std:0.0021697808988392353 - rollout_corr/rollout_is_seq_max:1.0103951692581177 - rollout_corr/rollout_is_seq_min:0.9942268133163452 - rollout_corr/rollout_is_seq_max_deviation:0.010395169258117676 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.035950763616711) - rollout_corr/training_log_ppl:np.float64(0.03461706963435063) - rollout_corr/kl:np.float64(0.00124807827964446) - rollout_corr/k3_kl:np.float64(0.0011595660929395457) - rollout_corr/rollout_ppl:np.float64(1.0346353389322758) - rollout_corr/rollout_log_ppl:np.float64(0.03336899111445746) - rollout_corr/log_ppl_diff:np.float64(0.0012480785198931699) - rollout_corr/log_ppl_abs_diff:np.float64(0.002425113902972953) - rollout_corr/log_ppl_diff_max:np.float64(0.0498611256480217) - rollout_corr/log_ppl_diff_min:np.float64(-0.013390138745307922) - rollout_corr/ppl_ratio:np.float64(1.001260134857148) - rollout_corr/chi2_token:np.float64(0.0016280324198305607) - rollout_corr/chi2_seq:np.float64(0.5298847754020244) - actor/pg_loss:np.float64(0.0006970755057409406) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00016883460511962767) - actor/ppo_kl:np.float64(0.0007997096730001374) - 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(0.69921875) - self_distillation/token_reweight_w_mean:np.float64(92006.85765184229) - self_distillation/token_reweight_w_std:np.float64(980502.2056374671) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0004657644312829) - self_distillation/token_reweight_w_clip_frac:np.float64(0.011001636754372157) - self_distillation/empty_target_batch:np.float64(0.25) - actor/grad_norm:np.float64(0.259071440435946) - perf/mfu/actor:np.float64(0.0382410973678092) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.5239028930664) - actor/lr:np.float64(1e-06) - training/global_step:81 - training/epoch:3 - 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:0.0011233530240133405 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.0011233530240133405 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:276.01171875 - response_length/max:610.0 - resp
|
||
step:82 - global_seqlen/min:18511 - global_seqlen/max:24253 - global_seqlen/minmax_diff:5742 - global_seqlen/balanced_min:21238 - global_seqlen/balanced_max:21240 - global_seqlen/mean:21239.375 - actor/entropy:0.03823944181203842 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3175753355026245 - training/rollout_probs_diff_mean:0.0017360865604132414 - training/rollout_probs_diff_std:0.01078617200255394 - training/rollout_actor_probs_pearson_corr:0.9948719143867493 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71875 - 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:0.71875 - rollout_corr/rollout_is_mean:1.0000228881835938 - rollout_corr/rollout_is_ratio_fraction_high:5.631661042571068e-05 - rollout_corr/rollout_is_ratio_fraction_low:8.447491563856602e-05 - rollout_corr/rollout_is_max:3.3646881580352783 - rollout_corr/rollout_is_min:0.23710303008556366 - rollout_corr/rollout_is_std:0.02898392453789711 - rollout_corr/rollout_is_eff_sample_size:0.9991605182489611 - rollout_corr/rollout_is_seq_mean:1.000075340270996 - rollout_corr/rollout_is_seq_std:0.0025008083321154118 - rollout_corr/rollout_is_seq_max:1.0169435739517212 - rollout_corr/rollout_is_seq_min:0.991999626159668 - rollout_corr/rollout_is_seq_max_deviation:0.01694357395172119 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0359556558541954) - rollout_corr/training_log_ppl:np.float64(0.034949058814163436) - rollout_corr/kl:np.float64(0.0005831834411189618) - rollout_corr/k3_kl:np.float64(0.0009131975491611932) - rollout_corr/rollout_ppl:np.float64(1.0353414616547525) - rollout_corr/rollout_log_ppl:np.float64(0.03436587487749421) - rollout_corr/log_ppl_diff:np.float64(0.0005831839366692293) - rollout_corr/log_ppl_abs_diff:np.float64(0.001916763685130718) - rollout_corr/log_ppl_diff_max:np.float64(0.013963993638753891) - rollout_corr/log_ppl_diff_min:np.float64(-0.015550876036286354) - rollout_corr/ppl_ratio:np.float64(1.0005876692011952) - rollout_corr/chi2_token:np.float64(0.004429039778187871) - rollout_corr/chi2_seq:np.float64(0.7082133176736534) - actor/pg_loss:np.float64(0.00022834050469100475) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00011931026392630883) - actor/ppo_kl:np.float64(0.00015909244585188276) - 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(0.71875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.7109375) - self_distillation/token_reweight_w_mean:np.float64(30193.831576111028) - self_distillation/token_reweight_w_std:np.float64(321280.0764630872) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9998594501521438) - self_distillation/token_reweight_w_clip_frac:np.float64(0.003927729223505594) - self_distillation/empty_target_batch:np.float64(0.28125) - actor/grad_norm:np.float64(0.08594363182783127) - perf/mfu/actor:np.float64(0.03934670631693053) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.38236618041992) - actor/lr:np.float64(1e-06) - training/global_step:82 - training/epoch:3 - critic/score/mean:0.7109375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7109375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0003238205099478364 - critic/advantages/max:0.125 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0003238205099478364 - critic/returns/max:0.125 - critic/returns/min:-0.875 - response_length/mean:277.44921875 - resp
|
||
step:83 - global_seqlen/min:20094 - global_seqlen/max:24098 - global_seqlen/minmax_diff:4004 - global_seqlen/balanced_min:21362 - global_seqlen/balanced_max:21364 - global_seqlen/mean:21363.375 - actor/entropy:0.028036868199706078 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.938510537147522 - training/rollout_probs_diff_mean:0.0014980696141719818 - training/rollout_probs_diff_std:0.013240817934274673 - training/rollout_actor_probs_pearson_corr:0.9897127151489258 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.74609375 - self_distillation/correct_sample_fraction:0.6953125 - 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:1.0001375675201416 - rollout_corr/rollout_is_ratio_fraction_high:0.000113104579213541 - rollout_corr/rollout_is_ratio_fraction_low:0.0003393137303646654 - rollout_corr/rollout_is_max:3.193481922149658 - rollout_corr/rollout_is_min:0.04810234531760216 - rollout_corr/rollout_is_std:0.02907426841557026 - rollout_corr/rollout_is_eff_sample_size:0.9991558769107263 - rollout_corr/rollout_is_seq_mean:1.0002412796020508 - rollout_corr/rollout_is_seq_std:0.0025487029924988747 - rollout_corr/rollout_is_seq_max:1.015842080116272 - rollout_corr/rollout_is_seq_min:0.9927569627761841 - rollout_corr/rollout_is_seq_max_deviation:0.015842080116271973 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0276337102986872) - rollout_corr/training_log_ppl:np.float64(0.027054207998844504) - rollout_corr/kl:np.float64(0.0004606549058898679) - rollout_corr/k3_kl:np.float64(0.0007622376264420083) - rollout_corr/rollout_ppl:np.float64(1.0271579953841865) - rollout_corr/rollout_log_ppl:np.float64(0.026593552976464707) - rollout_corr/log_ppl_diff:np.float64(0.00046065502192504937) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017162600461233524) - rollout_corr/log_ppl_diff_max:np.float64(0.008437082171440125) - rollout_corr/log_ppl_diff_min:np.float64(-0.013399794697761536) - rollout_corr/ppl_ratio:np.float64(1.0004641304258257) - rollout_corr/chi2_token:np.float64(0.0028343098238110542) - rollout_corr/chi2_seq:np.float64(0.47074131085537374) - actor/pg_loss:np.float64(0.00014830310828983784) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00010468151140230475) - actor/ppo_kl:np.float64(3.112560744966686e-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(0.6953125) - self_distillation/token_reweight_w_mean:np.float64(141561.96050062706) - self_distillation/token_reweight_w_std:np.float64(1861106.9816170656) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0002270576078445) - self_distillation/token_reweight_w_clip_frac:np.float64(0.005041797267040238) - self_distillation/empty_target_batch:np.float64(0.25390625) - actor/grad_norm:np.float64(0.059840649366378784) - perf/mfu/actor:np.float64(0.03939250038014562) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.21709823608398) - actor/lr:np.float64(1e-06) - training/global_step:83 - training/epoch:3 - critic/score/mean:0.6953125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6953125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0016824306221678853 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0016824306221678853 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:
|
||
step:84 - global_seqlen/min:19137 - global_seqlen/max:26002 - global_seqlen/minmax_diff:6865 - global_seqlen/balanced_min:22053 - global_seqlen/balanced_max:22057 - global_seqlen/mean:22055.875 - actor/entropy:0.030365219339728355 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3535175025463104 - training/rollout_probs_diff_mean:0.0013786866329610348 - training/rollout_probs_diff_std:0.009515224024653435 - training/rollout_actor_probs_pearson_corr:0.9947026371955872 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.84375 - self_distillation/correct_sample_fraction:0.84375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.84375 - rollout_corr/rollout_is_mean:1.000044822692871 - rollout_corr/rollout_is_ratio_fraction_high:2.6288464141543955e-05 - rollout_corr/rollout_is_ratio_fraction_low:2.6288464141543955e-05 - rollout_corr/rollout_is_max:5.18784761428833 - rollout_corr/rollout_is_min:0.2994607388973236 - rollout_corr/rollout_is_std:0.024178551509976387 - rollout_corr/rollout_is_eff_sample_size:0.9994156201342632 - rollout_corr/rollout_is_seq_mean:0.9999868273735046 - rollout_corr/rollout_is_seq_std:0.0017935303039848804 - rollout_corr/rollout_is_seq_max:1.0134811401367188 - rollout_corr/rollout_is_seq_min:0.9930511713027954 - rollout_corr/rollout_is_seq_max_deviation:0.01348114013671875 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0276782847940922) - rollout_corr/training_log_ppl:np.float64(0.027085085138423892) - rollout_corr/kl:np.float64(0.00033988594790629323) - rollout_corr/k3_kl:np.float64(0.0003952844845345749) - rollout_corr/rollout_ppl:np.float64(1.027325198519975) - rollout_corr/rollout_log_ppl:np.float64(0.026745198925254954) - rollout_corr/log_ppl_diff:np.float64(0.0003398862131689384) - rollout_corr/log_ppl_abs_diff:np.float64(0.001232237534622982) - rollout_corr/log_ppl_diff_max:np.float64(0.012646578252315521) - rollout_corr/log_ppl_diff_min:np.float64(-0.004848688840866089) - rollout_corr/ppl_ratio:np.float64(1.0003415157552809) - rollout_corr/chi2_token:np.float64(0.0012291001621633768) - rollout_corr/chi2_seq:np.float64(0.22799045592546463) - actor/pg_loss:np.float64(0.0) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0) - actor/ppo_kl:np.float64(2.268064918098389e-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(0.84375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.84375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.84375) - self_distillation/token_reweight_w_mean:np.float64(0.9999999923165888) - self_distillation/token_reweight_w_std:np.float64(0.0) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9999999923165888) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.15625) - actor/grad_norm:np.float64(0.0) - perf/mfu/actor:np.float64(0.040526068192393905) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.13608169555664) - actor/lr:np.float64(1e-06) - training/global_step:84 - training/epoch:3 - critic/score/mean:0.84375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.84375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0 - critic/advantages/max:0.0 - critic/advantages/min:0.0 - critic/returns/mean:0.0 - critic/returns/max:0.0 - critic/returns/min:0.0 - response_length/mean:297.18359375 - response_length/max:750.0 - response_length/min:133.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:297.18359375 - re
|
||
step:85 - global_seqlen/min:17600 - global_seqlen/max:24927 - global_seqlen/minmax_diff:7327 - global_seqlen/balanced_min:21543 - global_seqlen/balanced_max:21545 - global_seqlen/mean:21544.25 - actor/entropy:0.03369623050093651 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4004964232444763 - training/rollout_probs_diff_mean:0.0016652214108034968 - training/rollout_probs_diff_std:0.010939822532236576 - training/rollout_actor_probs_pearson_corr:0.9941464066505432 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.77734375 - self_distillation/correct_sample_fraction:0.71875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.77734375 - rollout_corr/rollout_is_mean:1.00010085105896 - rollout_corr/rollout_is_ratio_fraction_high:9.723032417241484e-05 - rollout_corr/rollout_is_ratio_fraction_low:8.334028098033741e-05 - rollout_corr/rollout_is_max:3.8197498321533203 - rollout_corr/rollout_is_min:0.18857713043689728 - rollout_corr/rollout_is_std:0.02875475026667118 - rollout_corr/rollout_is_eff_sample_size:0.9991739664724485 - rollout_corr/rollout_is_seq_mean:1.0000773668289185 - rollout_corr/rollout_is_seq_std:0.0023834542371332645 - rollout_corr/rollout_is_seq_max:1.023589015007019 - rollout_corr/rollout_is_seq_min:0.9939945936203003 - rollout_corr/rollout_is_seq_max_deviation:0.023589015007019043 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0349912350066006) - rollout_corr/training_log_ppl:np.float64(0.034070851845172) - rollout_corr/kl:np.float64(0.000908018300947333) - rollout_corr/k3_kl:np.float64(0.0008562989545923827) - rollout_corr/rollout_ppl:np.float64(1.03404114022851) - rollout_corr/rollout_log_ppl:np.float64(0.03316283329104408) - rollout_corr/log_ppl_diff:np.float64(0.0009080185541279207) - rollout_corr/log_ppl_abs_diff:np.float64(0.001848379795319488) - rollout_corr/log_ppl_diff_max:np.float64(0.026849649846553802) - rollout_corr/log_ppl_diff_min:np.float64(-0.007245257496833801) - rollout_corr/ppl_ratio:np.float64(1.0009134255815297) - rollout_corr/chi2_token:np.float64(0.0018449586350470781) - rollout_corr/chi2_seq:np.float64(0.9608175957109779) - actor/pg_loss:np.float64(0.0002799328649416566) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002830132093549764) - actor/ppo_kl:np.float64(0.00048529359718330056) - 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(0.77734375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.77734375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.71875) - self_distillation/token_reweight_w_mean:np.float64(166367.9715071246) - self_distillation/token_reweight_w_std:np.float64(1815967.0959693075) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0000693320762366) - self_distillation/token_reweight_w_clip_frac:np.float64(0.007904627491370775) - self_distillation/empty_target_batch:np.float64(0.22265625) - actor/grad_norm:np.float64(0.19982873648405075) - perf/mfu/actor:np.float64(0.039475850572369256) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.0826416015625) - actor/lr:np.float64(1e-06) - training/global_step:85 - training/epoch:3 - critic/score/mean:0.71875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.71875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.002161638578400016 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.002161638578400016 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:281.2265625 - respons
|
||
step:86 - global_seqlen/min:18423 - global_seqlen/max:23773 - global_seqlen/minmax_diff:5350 - global_seqlen/balanced_min:21398 - global_seqlen/balanced_max:21401 - global_seqlen/mean:21400.0 - actor/entropy:0.039314959198236465 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.41023650765419006 - training/rollout_probs_diff_mean:0.0018119228770956397 - training/rollout_probs_diff_std:0.011676772497594357 - training/rollout_actor_probs_pearson_corr:0.9939371347427368 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.84375 - self_distillation/correct_sample_fraction:0.81640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.84375 - rollout_corr/rollout_is_mean:1.0000507831573486 - rollout_corr/rollout_is_ratio_fraction_high:9.458437125431374e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0001756566925905645 - rollout_corr/rollout_is_max:2.5501339435577393 - rollout_corr/rollout_is_min:0.18535448610782623 - rollout_corr/rollout_is_std:0.030159026384353638 - rollout_corr/rollout_is_eff_sample_size:0.9990913786735653 - rollout_corr/rollout_is_seq_mean:1.0000463724136353 - rollout_corr/rollout_is_seq_std:0.002288077026605606 - rollout_corr/rollout_is_seq_max:1.0129072666168213 - rollout_corr/rollout_is_seq_min:0.9928169250488281 - rollout_corr/rollout_is_seq_max_deviation:0.012907266616821289 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0370242814533412) - rollout_corr/training_log_ppl:np.float64(0.036050161899311206) - rollout_corr/kl:np.float64(0.0012159123268471461) - rollout_corr/k3_kl:np.float64(0.001352685779481999) - rollout_corr/rollout_ppl:np.float64(1.0357652008533478) - rollout_corr/rollout_log_ppl:np.float64(0.03483424940327495) - rollout_corr/log_ppl_diff:np.float64(0.0012159124960362533) - rollout_corr/log_ppl_abs_diff:np.float64(0.0024147777801317716) - rollout_corr/log_ppl_diff_max:np.float64(0.016132771968841553) - rollout_corr/log_ppl_diff_min:np.float64(-0.022180378437042236) - rollout_corr/ppl_ratio:np.float64(1.0012233622837812) - rollout_corr/chi2_token:np.float64(0.007148264441639185) - rollout_corr/chi2_seq:np.float64(2.3162417996209115) - actor/pg_loss:np.float64(0.000582026899792254) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00015828989535293658) - actor/ppo_kl:np.float64(0.0007632725505146709) - 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(0.84375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.84375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.81640625) - self_distillation/token_reweight_w_mean:np.float64(79115.98013575724) - self_distillation/token_reweight_w_std:np.float64(957300.9500577603) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9999438552185893) - self_distillation/token_reweight_w_clip_frac:np.float64(0.005835453615873121) - self_distillation/empty_target_batch:np.float64(0.15625) - actor/grad_norm:np.float64(0.1951440516859293) - perf/mfu/actor:np.float64(0.039398913577472425) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.11903762817383) - actor/lr:np.float64(1e-06) - training/global_step:86 - training/epoch:3 - critic/score/mean:0.81640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.81640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0009458436979912221 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:0.0009458436979912221 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:289.09375 - respo
|
||
step:87 - global_seqlen/min:19044 - global_seqlen/max:24169 - global_seqlen/minmax_diff:5125 - global_seqlen/balanced_min:21777 - global_seqlen/balanced_max:21778 - global_seqlen/mean:21777.125 - actor/entropy:0.027197115123271942 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.875576913356781 - training/rollout_probs_diff_mean:0.0015121392207220197 - training/rollout_probs_diff_std:0.011220099404454231 - training/rollout_actor_probs_pearson_corr:0.9922543168067932 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65625 - 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:0.65625 - rollout_corr/rollout_is_mean:0.9999395608901978 - rollout_corr/rollout_is_ratio_fraction_high:5.836263517267071e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00016049724945332855 - rollout_corr/rollout_is_max:3.3198060989379883 - rollout_corr/rollout_is_min:0.0059265922755002975 - rollout_corr/rollout_is_std:0.027668792754411697 - rollout_corr/rollout_is_eff_sample_size:0.9992347855255346 - rollout_corr/rollout_is_seq_mean:0.99995356798172 - rollout_corr/rollout_is_seq_std:0.0020202931482344866 - rollout_corr/rollout_is_seq_max:1.010077953338623 - rollout_corr/rollout_is_seq_min:0.9906949400901794 - rollout_corr/rollout_is_seq_max_deviation:0.010077953338623047 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.030908034183085) - rollout_corr/training_log_ppl:np.float64(0.030158942176967685) - rollout_corr/kl:np.float64(0.0022416504295428297) - rollout_corr/k3_kl:np.float64(0.0021501615189396077) - rollout_corr/rollout_ppl:np.float64(1.0285401679575443) - rollout_corr/rollout_log_ppl:np.float64(0.02791729165619472) - rollout_corr/log_ppl_diff:np.float64(0.0022416505207729642) - rollout_corr/log_ppl_abs_diff:np.float64(0.0032178409592233947) - rollout_corr/log_ppl_diff_max:np.float64(0.07320429384708405) - rollout_corr/log_ppl_diff_min:np.float64(-0.006806857883930206) - rollout_corr/ppl_ratio:np.float64(1.002269170479849) - rollout_corr/chi2_token:np.float64(0.004207434365525842) - rollout_corr/chi2_seq:np.float64(0.16284064133651555) - actor/pg_loss:np.float64(0.0011388499988242984) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00020194145145069342) - actor/ppo_kl:np.float64(0.0016807883800389334) - 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(0.65625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.65625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.62109375) - self_distillation/token_reweight_w_mean:np.float64(184774.13582514809) - self_distillation/token_reweight_w_std:np.float64(1593571.5521162562) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0002801124937832) - self_distillation/token_reweight_w_clip_frac:np.float64(0.006642151573032606) - self_distillation/empty_target_batch:np.float64(0.34375) - actor/grad_norm:np.float64(0.26061127707362175) - perf/mfu/actor:np.float64(0.040144429518951516) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.14898681640625) - actor/lr:np.float64(1e-06) - training/global_step:87 - training/epoch:3 - 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:0.0004906108952127397 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:0.0004906108952127397 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:267.72265625
|
||
step:88 - global_seqlen/min:18952 - global_seqlen/max:23483 - global_seqlen/minmax_diff:4531 - global_seqlen/balanced_min:21164 - global_seqlen/balanced_max:21170 - global_seqlen/mean:21167.875 - actor/entropy:0.03469604626297951 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7770388722419739 - training/rollout_probs_diff_mean:0.0016765209147706628 - training/rollout_probs_diff_std:0.012051005847752094 - training/rollout_actor_probs_pearson_corr:0.9929908514022827 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71484375 - self_distillation/correct_sample_fraction:0.6640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.71484375 - rollout_corr/rollout_is_mean:1.00003981590271 - rollout_corr/rollout_is_ratio_fraction_high:0.00011329037806717679 - rollout_corr/rollout_is_ratio_fraction_low:0.00011329037806717679 - rollout_corr/rollout_is_max:3.9595537185668945 - rollout_corr/rollout_is_min:0.06086057052016258 - rollout_corr/rollout_is_std:0.02961859479546547 - rollout_corr/rollout_is_eff_sample_size:0.9991232697500432 - rollout_corr/rollout_is_seq_mean:1.0000507831573486 - rollout_corr/rollout_is_seq_std:0.0024844331201165915 - rollout_corr/rollout_is_seq_max:1.0169881582260132 - rollout_corr/rollout_is_seq_min:0.9927829504013062 - rollout_corr/rollout_is_seq_max_deviation:0.016988158226013184 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0361049468629062) - rollout_corr/training_log_ppl:np.float64(0.034795299737993446) - rollout_corr/kl:np.float64(0.00043864597136611394) - rollout_corr/k3_kl:np.float64(0.0006331470921853155) - rollout_corr/rollout_ppl:np.float64(1.0356523073278368) - rollout_corr/rollout_log_ppl:np.float64(0.03435665349843475) - rollout_corr/log_ppl_diff:np.float64(0.0004386462377397038) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017799362379946615) - rollout_corr/log_ppl_diff_max:np.float64(0.018246829509735107) - rollout_corr/log_ppl_diff_min:np.float64(-0.006508391350507736) - rollout_corr/ppl_ratio:np.float64(1.0004424944054335) - rollout_corr/chi2_token:np.float64(0.0017431846354156733) - rollout_corr/chi2_seq:np.float64(0.6097231933381408) - actor/pg_loss:np.float64(0.00028333289083093405) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00020099821449548472) - actor/ppo_kl:np.float64(-5.0402789252501634e-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(0.71484375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71484375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6640625) - self_distillation/token_reweight_w_mean:np.float64(131093.8798913369) - self_distillation/token_reweight_w_std:np.float64(1555027.8888158603) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0006680756341666) - self_distillation/token_reweight_w_clip_frac:np.float64(0.010408774585812353) - self_distillation/empty_target_batch:np.float64(0.28515625) - actor/grad_norm:np.float64(0.1889299713075161) - perf/mfu/actor:np.float64(0.038953613110281254) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.14186477661133) - actor/lr:np.float64(1e-06) - training/global_step:88 - training/epoch:3 - critic/score/mean:0.6640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00011329037806717679 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00011329037806717679 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/m
|
||
step:89 - global_seqlen/min:20201 - global_seqlen/max:24059 - global_seqlen/minmax_diff:3858 - global_seqlen/balanced_min:22120 - global_seqlen/balanced_max:22122 - global_seqlen/mean:22121.375 - actor/entropy:0.0294000506401062 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9222151041030884 - training/rollout_probs_diff_mean:0.0016106810653582215 - training/rollout_probs_diff_std:0.01242443360388279 - training/rollout_actor_probs_pearson_corr:0.9914741516113281 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.78125 - self_distillation/correct_sample_fraction:0.73828125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.78125 - rollout_corr/rollout_is_mean:1.0001593828201294 - rollout_corr/rollout_is_ratio_fraction_high:0.00010028222459368408 - rollout_corr/rollout_is_ratio_fraction_low:0.0002292165154358372 - rollout_corr/rollout_is_max:2.4047913551330566 - rollout_corr/rollout_is_min:0.0020848875865340233 - rollout_corr/rollout_is_std:0.02992490865290165 - rollout_corr/rollout_is_eff_sample_size:0.9991055390121959 - rollout_corr/rollout_is_seq_mean:1.0000648498535156 - rollout_corr/rollout_is_seq_std:0.0025368130300194025 - rollout_corr/rollout_is_seq_max:1.0112179517745972 - rollout_corr/rollout_is_seq_min:0.9908243417739868 - rollout_corr/rollout_is_seq_max_deviation:0.011217951774597168 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0337297716178) - rollout_corr/training_log_ppl:np.float64(0.03286470607508818) - rollout_corr/kl:np.float64(0.0006779706385913187) - rollout_corr/k3_kl:np.float64(0.001490014383888183) - rollout_corr/rollout_ppl:np.float64(1.0330208893865347) - rollout_corr/rollout_log_ppl:np.float64(0.03218673507581116) - rollout_corr/log_ppl_diff:np.float64(0.0006779709992770222) - rollout_corr/log_ppl_abs_diff:np.float64(0.002508861419300956) - rollout_corr/log_ppl_diff_max:np.float64(0.03041791543364525) - rollout_corr/log_ppl_diff_min:np.float64(-0.01392044872045517) - rollout_corr/ppl_ratio:np.float64(1.0006865614559501) - rollout_corr/chi2_token:np.float64(0.0062223493587225676) - rollout_corr/chi2_seq:np.float64(0.9317113894503564) - actor/pg_loss:np.float64(0.0008178113494068384) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00023295565551961772) - actor/ppo_kl:np.float64(0.0001360127029670366) - actor/pg_clipfrac_lower:np.float64(2.396472336840816e-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(0.78125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.78125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.73828125) - self_distillation/token_reweight_w_mean:np.float64(144165.8516279969) - self_distillation/token_reweight_w_std:np.float64(1638643.5006887442) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0003369075711817) - self_distillation/token_reweight_w_clip_frac:np.float64(0.006198407252668403) - self_distillation/empty_target_batch:np.float64(0.21875) - actor/grad_norm:np.float64(0.32256839238107204) - perf/mfu/actor:np.float64(0.04058150884673786) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.04215621948242) - actor/lr:np.float64(1e-06) - training/global_step:89 - training/epoch:3 - critic/score/mean:0.73828125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.73828125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0013556007761508226 - critic/advantages/max:0.75 - critic/advantages/min:-0.75 - critic/returns/mean:-0.0013556007761508226 - critic/returns/max:0.75 - critic/returns/min:-0.75 - response_length/mean:272.
|
||
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
|
||
Given the following crystal structure parameters for the material RbLa9(IrO6)4 (Material ID: mp-560657), calculate the volume of the unit cell (in ų).
|
||
|
||
Lattice Parameters:
|
||
|
||
a = 7.82 Å
|
||
b = 7.82 Å
|
||
c = 17.88 Å
|
||
α = 90.00º
|
||
β = 90.00º
|
||
γ = 90.00º
|
||
|
||
|
||
|
||
A: 1025.67 ų
|
||
|
||
B: 1094.31 ų
|
||
|
||
C: 1200.45 ų
|
||
|
||
D: 1150.78 ų
|
||
|
||
Please reason step by step.
|
||
assistant
|
||
<think>
|
||
|
||
</think>
|
||
|
||
|
||
[response] <reasoning>
|
||
The volume $ V $ of a unit cell in a crystal structure can be calculated using the formula:
|
||
|
||
$$
|
||
V = a \cdot b \cdot c \cdot \sqrt{1 + 2\cos\alpha\cos\beta\cos\gamma - \cos^2\alpha - \cos^2\beta - \cos^2\gamma}
|
||
$$
|
||
|
||
Given that all angles $ \alpha, \beta, \gamma $ are 90.00º, the cosine of each angle is 0. Substituting these values into the formula:
|
||
|
||
$$
|
||
V = a \cdot b \cdot c \cdot \sqrt{1 + 2(0)(0)(0) - (0)^2 - (0)^2 - (0)^2} = a \cdot b \cdot c \cdot \sqrt{1} = a \cdot b \cdot c
|
||
$$
|
||
|
||
Now, substitute the given lattice parameters:
|
||
|
||
$$
|
||
V = 7.82 \, \text{Å} \times 7.82 \, \text{Å} \times 17.88 \, \text{Å}
|
||
$$
|
||
|
||
First, calculate $ 7.82 \times 7.82 $:
|
||
|
||
$$
|
||
7.82 \times 7.82 = 61.1524
|
||
$$
|
||
|
||
Then multiply by 17.88:
|
||
|
||
$$
|
||
61.1524 \times 17.88 \approx 1094.31
|
||
$$
|
||
|
||
Thus, the volume of the unit cell is approximately 1094.31 ų.
|
||
</reasoning>
|
||
<answer>
|
||
B
|
||
</answer>
|
||
[ground_truth] B
|
||
[score] 1.0
|
||
[acc] 1.0
|
||
[pred] B
|
||
[incorrect_format] 1
|
||
[feedback]
|
||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-4B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_90
|
||
step:90 - global_seqlen/min:18160 - global_seqlen/max:22613 - global_seqlen/minmax_diff:4453 - global_seqlen/balanced_min:21124 - global_seqlen/balanced_max:21127 - global_seqlen/mean:21126.0 - actor/entropy:0.03373492509126663 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5032570362091064 - training/rollout_probs_diff_mean:0.0016449851682409644 - training/rollout_probs_diff_std:0.010777240619063377 - training/rollout_actor_probs_pearson_corr:0.9940168261528015 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.78125 - self_distillation/correct_sample_fraction:0.78125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.78125 - rollout_corr/rollout_is_mean:0.9999426603317261 - rollout_corr/rollout_is_ratio_fraction_high:4.441023338586092e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00013323070015758276 - rollout_corr/rollout_is_max:3.03580379486084 - rollout_corr/rollout_is_min:0.14749553799629211 - rollout_corr/rollout_is_std:0.027225784957408905 - rollout_corr/rollout_is_eff_sample_size:0.9992594247051011 - rollout_corr/rollout_is_seq_mean:0.9998467564582825 - rollout_corr/rollout_is_seq_std:0.0020057959482073784 - rollout_corr/rollout_is_seq_max:1.0077792406082153 - rollout_corr/rollout_is_seq_min:0.9924387335777283 - rollout_corr/rollout_is_seq_max_deviation:0.007779240608215332 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0404213974252343) - rollout_corr/training_log_ppl:np.float64(0.03911794312261918) - rollout_corr/kl:np.float64(0.006334157641134652) - rollout_corr/k3_kl:np.float64(0.005529801496451547) - rollout_corr/rollout_ppl:np.float64(1.0335820065811276) - rollout_corr/rollout_log_ppl:np.float64(0.03278378526101733) - rollout_corr/log_ppl_diff:np.float64(0.006334157883429725) - rollout_corr/log_ppl_abs_diff:np.float64(0.007246610600304848) - rollout_corr/log_ppl_diff_max:np.float64(0.0966530591249466) - rollout_corr/log_ppl_diff_min:np.float64(-0.011298129335045815) - rollout_corr/ppl_ratio:np.float64(1.0064935444388539) - rollout_corr/chi2_token:np.float64(0.006344154244288802) - rollout_corr/chi2_seq:np.float64(1.2312942533753812) - actor/pg_loss:np.float64(0.0) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0) - actor/ppo_kl:np.float64(0.005713566887692423) - 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(0.78125) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.78125) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.78125) - self_distillation/token_reweight_w_mean:np.float64(0.9999999932479113) - self_distillation/token_reweight_w_std:np.float64(0.0) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9999999932479113) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.21875) - actor/grad_norm:np.float64(0.0) - perf/mfu/actor:np.float64(0.03905662064435432) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.02539443969727) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7639627659574468) - val-aux/sciknoweval/reward/std@16:np.float64(0.03739248197220592) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7783723404255319) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.025295053531890987) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.7499042553191487) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.03754459470577036) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7638936170212767) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0
|
||
step:91 - global_seqlen/min:18107 - global_seqlen/max:23168 - global_seqlen/minmax_diff:5061 - global_seqlen/balanced_min:21250 - global_seqlen/balanced_max:21254 - global_seqlen/mean:21251.875 - actor/entropy:0.04534173756837845 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4046512842178345 - training/rollout_probs_diff_mean:0.0016173782059922814 - training/rollout_probs_diff_std:0.010041984729468822 - training/rollout_actor_probs_pearson_corr:0.9960680603981018 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80859375 - self_distillation/correct_sample_fraction:0.75 - 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:0.9999968409538269 - rollout_corr/rollout_is_ratio_fraction_high:2.5360435756738298e-05 - rollout_corr/rollout_is_ratio_fraction_low:2.5360435756738298e-05 - rollout_corr/rollout_is_max:5.243876934051514 - rollout_corr/rollout_is_min:0.37942102551460266 - rollout_corr/rollout_is_std:0.025519412010908127 - rollout_corr/rollout_is_eff_sample_size:0.9993493025433431 - rollout_corr/rollout_is_seq_mean:0.9999523162841797 - rollout_corr/rollout_is_seq_std:0.002020125510171056 - rollout_corr/rollout_is_seq_max:1.012386441230774 - rollout_corr/rollout_is_seq_min:0.9923055171966553 - rollout_corr/rollout_is_seq_max_deviation:0.012386441230773926 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0488830306567252) - rollout_corr/training_log_ppl:np.float64(0.0465941856673453) - rollout_corr/kl:np.float64(0.0016078722916241794) - rollout_corr/k3_kl:np.float64(0.0012860047578371336) - rollout_corr/rollout_ppl:np.float64(1.047186788637191) - rollout_corr/rollout_log_ppl:np.float64(0.04498631308433687) - rollout_corr/log_ppl_diff:np.float64(0.0016078725830084295) - rollout_corr/log_ppl_abs_diff:np.float64(0.0026501509328227257) - rollout_corr/log_ppl_diff_max:np.float64(0.04505942016839981) - rollout_corr/log_ppl_diff_min:np.float64(-0.015204224735498428) - rollout_corr/ppl_ratio:np.float64(1.0016248126048595) - rollout_corr/chi2_token:np.float64(0.0017210114747285843) - rollout_corr/chi2_seq:np.float64(2.4243051193188876) - actor/pg_loss:np.float64(0.0012130095856264234) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00044657633543465636) - actor/ppo_kl:np.float64(0.0011831951068810298) - 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(0.75) - self_distillation/token_reweight_w_mean:np.float64(150807.6977907042) - self_distillation/token_reweight_w_std:np.float64(1827940.7117348753) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0000478033907712) - self_distillation/token_reweight_w_clip_frac:np.float64(0.012809829524485394) - self_distillation/empty_target_batch:np.float64(0.19140625) - actor/grad_norm:np.float64(0.2270626574754715) - perf/mfu/actor:np.float64(0.038449014366350065) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.40208053588867) - actor/lr:np.float64(1e-06) - training/global_step:91 - training/epoch:3 - critic/score/mean:0.75 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.75 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.00025835944688878953 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00025835944688878953 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:308.05859375 - response
|
||
step:92 - global_seqlen/min:21040 - global_seqlen/max:26095 - global_seqlen/minmax_diff:5055 - global_seqlen/balanced_min:23108 - global_seqlen/balanced_max:23110 - global_seqlen/mean:23109.375 - actor/entropy:0.049469124525785446 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.38041767477989197 - training/rollout_probs_diff_mean:0.001835858216509223 - training/rollout_probs_diff_std:0.010576015338301659 - training/rollout_actor_probs_pearson_corr:0.9960457682609558 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71484375 - self_distillation/correct_sample_fraction:0.69140625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.71484375 - rollout_corr/rollout_is_mean:0.999874472618103 - rollout_corr/rollout_is_ratio_fraction_high:3.575727896532044e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00011919093230972067 - rollout_corr/rollout_is_max:2.2295782566070557 - rollout_corr/rollout_is_min:0.2960941791534424 - rollout_corr/rollout_is_std:0.02729356847703457 - rollout_corr/rollout_is_eff_sample_size:0.9992554966385375 - rollout_corr/rollout_is_seq_mean:0.9999048709869385 - rollout_corr/rollout_is_seq_std:0.0016976064071059227 - rollout_corr/rollout_is_seq_max:1.0060241222381592 - rollout_corr/rollout_is_seq_min:0.9939712285995483 - rollout_corr/rollout_is_seq_max_deviation:0.00602877140045166 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0476838406175375) - rollout_corr/training_log_ppl:np.float64(0.045997846729278535) - rollout_corr/kl:np.float64(0.000593296980607505) - rollout_corr/k3_kl:np.float64(0.0005309369002016728) - rollout_corr/rollout_ppl:np.float64(1.0470488839782774) - rollout_corr/rollout_log_ppl:np.float64(0.04540454928155668) - rollout_corr/log_ppl_diff:np.float64(0.0005932974477218522) - rollout_corr/log_ppl_abs_diff:np.float64(0.001585291376159148) - rollout_corr/log_ppl_diff_max:np.float64(0.011447396129369736) - rollout_corr/log_ppl_diff_min:np.float64(-0.007249504327774048) - rollout_corr/ppl_ratio:np.float64(1.0005958129186183) - rollout_corr/chi2_token:np.float64(0.001005234895274043) - rollout_corr/chi2_seq:np.float64(0.339298699516803) - actor/pg_loss:np.float64(0.00011454732157289982) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(6.591906640096568e-05) - actor/ppo_kl:np.float64(0.00011144039859800614) - 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(0.71484375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71484375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.69140625) - self_distillation/token_reweight_w_mean:np.float64(27679.604563228786) - self_distillation/token_reweight_w_std:np.float64(400284.27851104736) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0002795667387545) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0033208947570528835) - self_distillation/empty_target_batch:np.float64(0.28515625) - actor/grad_norm:np.float64(0.04137406311929226) - perf/mfu/actor:np.float64(0.042280402409404685) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.34305953979492) - actor/lr:np.float64(1e-06) - training/global_step:92 - training/epoch:3 - critic/score/mean:0.69140625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.69140625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0007419635658152401 - critic/advantages/max:0.875 - critic/advantages/min:-0.125 - critic/returns/mean:0.0007419635658152401 - critic/returns/max:0.875 - critic/returns/min:-0.125 - response_length/mean
|
||
step:93 - global_seqlen/min:16732 - global_seqlen/max:24478 - global_seqlen/minmax_diff:7746 - global_seqlen/balanced_min:21236 - global_seqlen/balanced_max:21239 - global_seqlen/mean:21237.875 - actor/entropy:0.037446752190589905 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7017467021942139 - training/rollout_probs_diff_mean:0.0014789922861382365 - training/rollout_probs_diff_std:0.010445463471114635 - training/rollout_actor_probs_pearson_corr:0.9950976371765137 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.7734375 - 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:1.000028133392334 - rollout_corr/rollout_is_ratio_fraction_high:6.884113099658862e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0001101458037737757 - rollout_corr/rollout_is_max:2.2795581817626953 - rollout_corr/rollout_is_min:0.27706241607666016 - rollout_corr/rollout_is_std:0.02562429942190647 - rollout_corr/rollout_is_eff_sample_size:0.999343826080078 - rollout_corr/rollout_is_seq_mean:0.9999866485595703 - rollout_corr/rollout_is_seq_std:0.0016226578736677766 - rollout_corr/rollout_is_seq_max:1.0060209035873413 - rollout_corr/rollout_is_seq_min:0.9960787892341614 - rollout_corr/rollout_is_seq_max_deviation:0.006020903587341309 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0422259448096156) - rollout_corr/training_log_ppl:np.float64(0.040301359121713176) - rollout_corr/kl:np.float64(0.0005277691595435563) - rollout_corr/k3_kl:np.float64(0.000485558227204419) - rollout_corr/rollout_ppl:np.float64(1.0416553649120033) - rollout_corr/rollout_log_ppl:np.float64(0.03977358995780378) - rollout_corr/log_ppl_diff:np.float64(0.0005277691602714185) - rollout_corr/log_ppl_abs_diff:np.float64(0.001454660314720968) - rollout_corr/log_ppl_diff_max:np.float64(0.008309893310070038) - rollout_corr/log_ppl_diff_min:np.float64(-0.005868792533874512) - rollout_corr/ppl_ratio:np.float64(1.0005299146287143) - rollout_corr/chi2_token:np.float64(0.0008362585213035345) - rollout_corr/chi2_seq:np.float64(0.17302076471969485) - actor/pg_loss:np.float64(0.0005822493694722652) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00018108219865098363) - actor/ppo_kl:np.float64(0.0001428027055716541) - 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(0.7734375) - self_distillation/token_reweight_w_mean:np.float64(143667.05467315274) - self_distillation/token_reweight_w_std:np.float64(1510607.7448990536) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0000457826536149) - self_distillation/token_reweight_w_clip_frac:np.float64(0.011807282658992335) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.18011473584920168) - perf/mfu/actor:np.float64(0.0392293668835817) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.27584838867188) - actor/lr:np.float64(1e-06) - training/global_step:93 - training/epoch:3 - critic/score/mean:0.7734375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7734375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0028964905068278313 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.0028964905068278313 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:283.71484375 - response_len
|
||
step:94 - global_seqlen/min:18330 - global_seqlen/max:25428 - global_seqlen/minmax_diff:7098 - global_seqlen/balanced_min:22781 - global_seqlen/balanced_max:22786 - global_seqlen/mean:22783.75 - actor/entropy:0.041148312389850616 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4031469225883484 - training/rollout_probs_diff_mean:0.0015067117055878043 - training/rollout_probs_diff_std:0.00942541379481554 - training/rollout_actor_probs_pearson_corr:0.9961856603622437 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.77734375 - self_distillation/correct_sample_fraction:0.75 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.77734375 - rollout_corr/rollout_is_mean:0.999973714351654 - rollout_corr/rollout_is_ratio_fraction_high:2.427950494165998e-05 - rollout_corr/rollout_is_ratio_fraction_low:6.069876326364465e-05 - rollout_corr/rollout_is_max:3.678283214569092 - rollout_corr/rollout_is_min:0.35399362444877625 - rollout_corr/rollout_is_std:0.024225344881415367 - rollout_corr/rollout_is_eff_sample_size:0.9994133578093227 - rollout_corr/rollout_is_seq_mean:1.0000309944152832 - rollout_corr/rollout_is_seq_std:0.0017812533769756556 - rollout_corr/rollout_is_seq_max:1.0156874656677246 - rollout_corr/rollout_is_seq_min:0.9939485788345337 - rollout_corr/rollout_is_seq_max_deviation:0.01568746566772461 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0413265423849225) - rollout_corr/training_log_ppl:np.float64(0.039565609684359515) - rollout_corr/kl:np.float64(0.00037305394429409944) - rollout_corr/k3_kl:np.float64(0.00047105288938809053) - rollout_corr/rollout_ppl:np.float64(1.040933893993497) - rollout_corr/rollout_log_ppl:np.float64(0.03919255502250962) - rollout_corr/log_ppl_diff:np.float64(0.000373054661849892) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014055275255486777) - rollout_corr/log_ppl_diff_max:np.float64(0.009166814386844635) - rollout_corr/log_ppl_diff_min:np.float64(-0.007318452000617981) - rollout_corr/ppl_ratio:np.float64(1.00037496490404) - rollout_corr/chi2_token:np.float64(0.0012893092352896929) - rollout_corr/chi2_seq:np.float64(0.37059705168940127) - actor/pg_loss:np.float64(0.0002516477834433317) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(4.616753176378552e-05) - actor/ppo_kl:np.float64(7.975473046561987e-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(0.77734375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.77734375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.75) - self_distillation/token_reweight_w_mean:np.float64(70703.44456977467) - self_distillation/token_reweight_w_std:np.float64(886742.9536180737) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0004530518781394) - self_distillation/token_reweight_w_clip_frac:np.float64(0.008093894488411024) - self_distillation/empty_target_batch:np.float64(0.22265625) - actor/grad_norm:np.float64(0.08848729822784662) - perf/mfu/actor:np.float64(0.042086406035394576) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.20873260498047) - actor/lr:np.float64(1e-06) - training/global_step:94 - training/epoch:3 - critic/score/mean:0.75 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.75 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.001916563487611711 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.001916563487611711 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:321.7734375 - response_length
|
||
step:95 - global_seqlen/min:19988 - global_seqlen/max:25783 - global_seqlen/minmax_diff:5795 - global_seqlen/balanced_min:22801 - global_seqlen/balanced_max:22805 - global_seqlen/mean:22803.375 - actor/entropy:0.03662053495645523 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3295740783214569 - training/rollout_probs_diff_mean:0.0016202006954699755 - training/rollout_probs_diff_std:0.01027343887835741 - training/rollout_actor_probs_pearson_corr:0.9951478242874146 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71875 - self_distillation/correct_sample_fraction:0.71875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.71875 - rollout_corr/rollout_is_mean:1.0001161098480225 - rollout_corr/rollout_is_ratio_fraction_high:8.639948646305129e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0001604561839485541 - rollout_corr/rollout_is_max:3.8464818000793457 - rollout_corr/rollout_is_min:0.3074188232421875 - rollout_corr/rollout_is_std:0.027437321841716766 - rollout_corr/rollout_is_eff_sample_size:0.999247997688126 - rollout_corr/rollout_is_seq_mean:1.0001269578933716 - rollout_corr/rollout_is_seq_std:0.0018558091251179576 - rollout_corr/rollout_is_seq_max:1.0104142427444458 - rollout_corr/rollout_is_seq_min:0.9959591627120972 - rollout_corr/rollout_is_seq_max_deviation:0.0104142427444458 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.037611531559378) - rollout_corr/training_log_ppl:np.float64(0.0367084229574175) - rollout_corr/kl:np.float64(0.00037698546704056923) - rollout_corr/k3_kl:np.float64(0.00048532903274178807) - rollout_corr/rollout_ppl:np.float64(1.0372191835194826) - rollout_corr/rollout_log_ppl:np.float64(0.036331437019725854) - rollout_corr/log_ppl_diff:np.float64(0.0003769859376916429) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013970075160614215) - rollout_corr/log_ppl_diff_max:np.float64(0.007364674936980009) - rollout_corr/log_ppl_diff_min:np.float64(-0.006315544247627258) - rollout_corr/ppl_ratio:np.float64(1.0003787786699831) - rollout_corr/chi2_token:np.float64(0.001227242173627019) - rollout_corr/chi2_seq:np.float64(0.30870523513294756) - actor/pg_loss:np.float64(0.0) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0) - actor/ppo_kl:np.float64(6.90859128940069e-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(0.71875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.71875) - self_distillation/token_reweight_w_mean:np.float64(0.9999999955762178) - self_distillation/token_reweight_w_std:np.float64(0.0) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9999999955762178) - self_distillation/token_reweight_w_clip_frac:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.28125) - actor/grad_norm:np.float64(0.0) - perf/mfu/actor:np.float64(0.04232899000757017) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.19301223754883) - actor/lr:np.float64(1e-06) - training/global_step:95 - training/epoch:3 - critic/score/mean:0.71875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.71875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0 - critic/advantages/max:0.0 - critic/advantages/min:0.0 - critic/returns/mean:0.0 - critic/returns/max:0.0 - critic/returns/min:0.0 - response_length/mean:316.48046875 - response_length/max:757.0 - response_length/min:145.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:316.48046875 - response
|
||
step:96 - global_seqlen/min:19812 - global_seqlen/max:25392 - global_seqlen/minmax_diff:5580 - global_seqlen/balanced_min:22738 - global_seqlen/balanced_max:22741 - global_seqlen/mean:22739.75 - actor/entropy:0.0352025143802166 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.722224771976471 - training/rollout_probs_diff_mean:0.001679497305303812 - training/rollout_probs_diff_std:0.011236628517508507 - training/rollout_actor_probs_pearson_corr:0.993776261806488 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65625 - 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:0.65625 - rollout_corr/rollout_is_mean:1.0001745223999023 - rollout_corr/rollout_is_ratio_fraction_high:6.445458711823449e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00011601825099205598 - rollout_corr/rollout_is_max:4.976936340332031 - rollout_corr/rollout_is_min:0.20165325701236725 - rollout_corr/rollout_is_std:0.02686212584376335 - rollout_corr/rollout_is_eff_sample_size:0.9992793035761552 - rollout_corr/rollout_is_seq_mean:1.0002118349075317 - rollout_corr/rollout_is_seq_std:0.0018694534664973617 - rollout_corr/rollout_is_seq_max:1.0106958150863647 - rollout_corr/rollout_is_seq_min:0.9955504536628723 - rollout_corr/rollout_is_seq_max_deviation:0.010695815086364746 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0348785421811044) - rollout_corr/training_log_ppl:np.float64(0.03406564036595228) - rollout_corr/kl:np.float64(0.00037139787588214013) - rollout_corr/k3_kl:np.float64(0.0005362462779121557) - rollout_corr/rollout_ppl:np.float64(1.03448999626562) - rollout_corr/rollout_log_ppl:np.float64(0.03369424184438685) - rollout_corr/log_ppl_diff:np.float64(0.00037139852156542474) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014863652131680283) - rollout_corr/log_ppl_diff_max:np.float64(0.00833505392074585) - rollout_corr/log_ppl_diff_min:np.float64(-0.007462680339813232) - rollout_corr/ppl_ratio:np.float64(1.0003736605867743) - rollout_corr/chi2_token:np.float64(0.002083134837448597) - rollout_corr/chi2_seq:np.float64(1.332067912677303) - actor/pg_loss:np.float64(-0.00010274117812514305) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(7.768613795633428e-05) - actor/ppo_kl:np.float64(0.00016692980348764763) - 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(0.65625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.65625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.6484375) - self_distillation/token_reweight_w_mean:np.float64(36289.31735826656) - self_distillation/token_reweight_w_std:np.float64(368548.66153580655) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0000207382254303) - self_distillation/token_reweight_w_clip_frac:np.float64(0.001245500759978313) - self_distillation/empty_target_batch:np.float64(0.34375) - actor/grad_norm:np.float64(0.05526710581034422) - perf/mfu/actor:np.float64(0.041736199174072765) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.15214920043945) - actor/lr:np.float64(1e-06) - training/global_step:96 - training/epoch:3 - 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:-0.0001611364568816498 - critic/advantages/max:0.25 - critic/advantages/min:-0.75 - critic/returns/mean:-0.0001611364568816498 - critic/returns/max:0.25 - critic/returns/min:-0.75 - response_length/mean:303.0234375 - response_le
|
||
step:97 - global_seqlen/min:18675 - global_seqlen/max:24360 - global_seqlen/minmax_diff:5685 - global_seqlen/balanced_min:21800 - global_seqlen/balanced_max:21804 - global_seqlen/mean:21803.5 - actor/entropy:0.041224244982004166 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2449004352092743 - training/rollout_probs_diff_mean:0.0014201895100995898 - training/rollout_probs_diff_std:0.0084342360496521 - training/rollout_actor_probs_pearson_corr:0.9970186352729797 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.90625 - self_distillation/correct_sample_fraction:0.87890625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.90625 - rollout_corr/rollout_is_mean:1.000179409980774 - rollout_corr/rollout_is_ratio_fraction_high:3.7255973438732326e-05 - rollout_corr/rollout_is_ratio_fraction_low:4.9674632464302704e-05 - rollout_corr/rollout_is_max:2.874107599258423 - rollout_corr/rollout_is_min:0.41924646496772766 - rollout_corr/rollout_is_std:0.02389836125075817 - rollout_corr/rollout_is_eff_sample_size:0.999429551518837 - rollout_corr/rollout_is_seq_mean:1.0001500844955444 - rollout_corr/rollout_is_seq_std:0.0014063200214877725 - rollout_corr/rollout_is_seq_max:1.004724144935608 - rollout_corr/rollout_is_seq_min:0.9953712821006775 - rollout_corr/rollout_is_seq_max_deviation:0.00472414493560791 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0406718435697258) - rollout_corr/training_log_ppl:np.float64(0.039368912926875055) - rollout_corr/kl:np.float64(0.00030021515657274733) - rollout_corr/k3_kl:np.float64(0.0004091687973215574) - rollout_corr/rollout_ppl:np.float64(1.040352996904403) - rollout_corr/rollout_log_ppl:np.float64(0.03906869714410277) - rollout_corr/log_ppl_diff:np.float64(0.000300215782772284) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011868714937008917) - rollout_corr/log_ppl_diff_max:np.float64(0.010238464921712875) - rollout_corr/log_ppl_diff_min:np.float64(-0.00530620664358139) - rollout_corr/ppl_ratio:np.float64(1.0003018721472472) - rollout_corr/chi2_token:np.float64(0.0010263379663228989) - rollout_corr/chi2_seq:np.float64(0.36499369889497757) - actor/pg_loss:np.float64(8.685840293765068e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00010191214551014127) - actor/ppo_kl:np.float64(0.0001593927016365626) - 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(0.90625) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.90625) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.87890625) - self_distillation/token_reweight_w_mean:np.float64(37507.985115043586) - self_distillation/token_reweight_w_std:np.float64(574614.4153616531) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9999924194999039) - self_distillation/token_reweight_w_clip_frac:np.float64(0.005771750904386863) - self_distillation/empty_target_batch:np.float64(0.09375) - actor/grad_norm:np.float64(0.10139033989980817) - perf/mfu/actor:np.float64(0.040042574424636254) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.15067863464355) - actor/lr:np.float64(1e-06) - training/global_step:97 - training/epoch:3 - critic/score/mean:0.87890625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.87890625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00021422184363473207 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.00021422184363473207 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:314.546875 - respon
|
||
step:98 - global_seqlen/min:20744 - global_seqlen/max:24464 - global_seqlen/minmax_diff:3720 - global_seqlen/balanced_min:22195 - global_seqlen/balanced_max:22198 - global_seqlen/mean:22197.25 - actor/entropy:0.044722869992256165 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.37975114583969116 - training/rollout_probs_diff_mean:0.0017690645763650537 - training/rollout_probs_diff_std:0.010033019818365574 - training/rollout_actor_probs_pearson_corr:0.9960561990737915 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.9375 - self_distillation/correct_sample_fraction:0.921875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.9375 - rollout_corr/rollout_is_mean:1.0002034902572632 - rollout_corr/rollout_is_ratio_fraction_high:5.368695201468654e-05 - rollout_corr/rollout_is_ratio_fraction_low:5.368695201468654e-05 - rollout_corr/rollout_is_max:4.066667556762695 - rollout_corr/rollout_is_min:0.3545277714729309 - rollout_corr/rollout_is_std:0.025945613160729408 - rollout_corr/rollout_is_eff_sample_size:0.9993276351484325 - rollout_corr/rollout_is_seq_mean:1.0001791715621948 - rollout_corr/rollout_is_seq_std:0.001851679291576147 - rollout_corr/rollout_is_seq_max:1.0125370025634766 - rollout_corr/rollout_is_seq_min:0.9958828091621399 - rollout_corr/rollout_is_seq_max_deviation:0.012537002563476562 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0432171327993274) - rollout_corr/training_log_ppl:np.float64(0.04202124249331973) - rollout_corr/kl:np.float64(0.00027291100581017247) - rollout_corr/k3_kl:np.float64(0.000395780640820409) - rollout_corr/rollout_ppl:np.float64(1.0429281839169562) - rollout_corr/rollout_log_ppl:np.float64(0.04174833083379781) - rollout_corr/log_ppl_diff:np.float64(0.00027291165952192387) - rollout_corr/log_ppl_abs_diff:np.float64(0.001402678571139404) - rollout_corr/log_ppl_diff_max:np.float64(0.005851060152053833) - rollout_corr/log_ppl_diff_min:np.float64(-0.0076254382729530334) - rollout_corr/ppl_ratio:np.float64(1.0002746109385043) - rollout_corr/chi2_token:np.float64(0.0011082068085670471) - rollout_corr/chi2_seq:np.float64(0.39986567525193095) - actor/pg_loss:np.float64(0.00015399395488202572) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(6.233245085240924e-05) - actor/ppo_kl:np.float64(8.391451634537361e-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(0.9375) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.9375) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.921875) - self_distillation/token_reweight_w_mean:np.float64(28126.951660413528) - self_distillation/token_reweight_w_std:np.float64(436973.87273890217) - self_distillation/token_reweight_w_clipped_mean:np.float64(0.9999532324727625) - self_distillation/token_reweight_w_clip_frac:np.float64(0.004359587779617868) - self_distillation/empty_target_batch:np.float64(0.0625) - actor/grad_norm:np.float64(0.07238347781822085) - perf/mfu/actor:np.float64(0.04097528863884095) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.1676902770996) - actor/lr:np.float64(1e-06) - training/global_step:98 - training/epoch:3 - critic/score/mean:0.921875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.921875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0017314042197540402 - critic/advantages/max:0.375 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0017314042197540402 - critic/returns/max:0.375 - critic/returns/min:-0.875 - response_length/mean:291.0390625 - response_le
|
||
step:99 - global_seqlen/min:18553 - global_seqlen/max:24227 - global_seqlen/minmax_diff:5674 - global_seqlen/balanced_min:21762 - global_seqlen/balanced_max:21765 - global_seqlen/mean:21764.375 - actor/entropy:0.05101349577307701 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7172134518623352 - training/rollout_probs_diff_mean:0.002091911854222417 - training/rollout_probs_diff_std:0.011637764982879162 - training/rollout_actor_probs_pearson_corr:0.9952604174613953 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80859375 - self_distillation/correct_sample_fraction:0.7578125 - 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:0.9999338984489441 - rollout_corr/rollout_is_ratio_fraction_high:4.100433216081001e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00012301300012040883 - rollout_corr/rollout_is_max:2.5501413345336914 - rollout_corr/rollout_is_min:0.23693661391735077 - rollout_corr/rollout_is_std:0.02900037169456482 - rollout_corr/rollout_is_eff_sample_size:0.9991595661760642 - rollout_corr/rollout_is_seq_mean:0.9999666213989258 - rollout_corr/rollout_is_seq_std:0.0020682818721979856 - rollout_corr/rollout_is_seq_max:1.0070600509643555 - rollout_corr/rollout_is_seq_min:0.9947633743286133 - rollout_corr/rollout_is_seq_max_deviation:0.007060050964355469 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0553245856426656) - rollout_corr/training_log_ppl:np.float64(0.0528352315795928) - rollout_corr/kl:np.float64(0.0008055590208186914) - rollout_corr/k3_kl:np.float64(0.0006251534139227743) - rollout_corr/rollout_ppl:np.float64(1.054449059534818) - rollout_corr/rollout_log_ppl:np.float64(0.052029672435310204) - rollout_corr/log_ppl_diff:np.float64(0.0008055591442825971) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018310855939489556) - rollout_corr/log_ppl_diff_max:np.float64(0.018006175756454468) - rollout_corr/log_ppl_diff_min:np.float64(-0.006862897425889969) - rollout_corr/ppl_ratio:np.float64(1.000809220597148) - rollout_corr/chi2_token:np.float64(0.0009292459581047297) - rollout_corr/chi2_seq:np.float64(0.2774903466925025) - actor/pg_loss:np.float64(0.0008074860088527203) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002502697416275623) - actor/ppo_kl:np.float64(0.0002938043922213751) - 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(0.7578125) - self_distillation/token_reweight_w_mean:np.float64(147148.14364791103) - self_distillation/token_reweight_w_std:np.float64(1759607.3955868636) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0003166827373207) - self_distillation/token_reweight_w_clip_frac:np.float64(0.014095777121838182) - self_distillation/empty_target_batch:np.float64(0.19140625) - actor/grad_norm:np.float64(0.20968144107609987) - perf/mfu/actor:np.float64(0.0402602080853634) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.147705078125) - actor/lr:np.float64(1e-06) - training/global_step:99 - training/epoch:3 - critic/score/mean:0.7578125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7578125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.001307013095356524 - critic/advantages/max:0.875 - critic/advantages/min:-0.625 - critic/returns/mean:0.001307013095356524 - critic/returns/max:0.875 - critic/returns/min:-0.625 - response_length/mean:285.7929687
|
||
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
|
||
Given the following crystal structure parameters for the material RbLa9(IrO6)4 (Material ID: mp-560657), calculate the volume of the unit cell (in ų).
|
||
|
||
Lattice Parameters:
|
||
|
||
a = 7.82 Å
|
||
b = 7.82 Å
|
||
c = 17.88 Å
|
||
α = 90.00º
|
||
β = 90.00º
|
||
γ = 90.00º
|
||
|
||
|
||
|
||
A: 1025.67 ų
|
||
|
||
B: 1094.31 ų
|
||
|
||
C: 1200.45 ų
|
||
|
||
D: 1150.78 ų
|
||
|
||
Please reason step by step.
|
||
assistant
|
||
<think>
|
||
|
||
</think>
|
||
|
||
|
||
[response] <reasoning>
|
||
The volume $ V $ of a unit cell in a crystal with given lattice parameters $ a, b, c $ and angles $ \alpha, \beta, \gamma $ is calculated using the formula:
|
||
|
||
$$
|
||
V = abc \sqrt{1 + 2\cos\alpha\cos\beta\cos\gamma - \cos^2\alpha - \cos^2\beta - \cos^2\gamma}
|
||
$$
|
||
|
||
Given:
|
||
- $ a = 7.82 $ Å
|
||
- $ b = 7.82 $ Å
|
||
- $ c = 17.88 $ Å
|
||
- $ \alpha = 90.00^\circ $
|
||
- $ \beta = 90.00^\circ $
|
||
- $ \gamma = 90.00^\circ $
|
||
|
||
Since all angles are 90°, their cosines are 0. Substituting into the formula:
|
||
|
||
$$
|
||
V = abc \sqrt{1 + 2(0)(0)(0) - (0)^2 - (0)^2 - (0)^2} = abc \sqrt{1} = abc
|
||
$$
|
||
|
||
Now calculate:
|
||
|
||
$$
|
||
V = 7.82 \times 7.82 \times 17.88
|
||
$$
|
||
|
||
First, calculate $ 7.82 \times 7.82 $:
|
||
|
||
$$
|
||
7.82^2 = 61.1524
|
||
$$
|
||
|
||
Then multiply by 17.88:
|
||
|
||
$$
|
||
61.1524 \times 17.88 \approx 1094.31
|
||
$$
|
||
|
||
Thus, the volume is approximately 1094.31 ų.
|
||
</reasoning>
|
||
<answer>
|
||
B
|
||
</answer>
|
||
[ground_truth] B
|
||
[score] 1.0
|
||
[acc] 1.0
|
||
[pred] B
|
||
[incorrect_format] 1
|
||
[feedback]
|
||
local_global_step_folder: /mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-4B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_100
|
||
step:100 - global_seqlen/min:18525 - global_seqlen/max:24595 - global_seqlen/minmax_diff:6070 - global_seqlen/balanced_min:21455 - global_seqlen/balanced_max:21459 - global_seqlen/mean:21457.25 - actor/entropy:0.052000705152750015 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.30163344740867615 - training/rollout_probs_diff_mean:0.00194946164265275 - training/rollout_probs_diff_std:0.010225724428892136 - training/rollout_actor_probs_pearson_corr:0.9965116381645203 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71875 - 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:0.71875 - rollout_corr/rollout_is_mean:0.9999290108680725 - rollout_corr/rollout_is_ratio_fraction_high:1.3980147741676774e-05 - rollout_corr/rollout_is_ratio_fraction_low:6.990074325585738e-05 - rollout_corr/rollout_is_max:2.706778049468994 - rollout_corr/rollout_is_min:0.4680381119251251 - rollout_corr/rollout_is_std:0.027313215658068657 - rollout_corr/rollout_is_eff_sample_size:0.9992544253530163 - rollout_corr/rollout_is_seq_mean:0.9999463558197021 - rollout_corr/rollout_is_seq_std:0.0017657832941040397 - rollout_corr/rollout_is_seq_max:1.0095514059066772 - rollout_corr/rollout_is_seq_min:0.996039092540741 - rollout_corr/rollout_is_seq_max_deviation:0.009551405906677246 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.056842735502869) - rollout_corr/training_log_ppl:np.float64(0.053712156926849275) - rollout_corr/kl:np.float64(0.0006709553122909995) - rollout_corr/k3_kl:np.float64(0.0005436672919252317) - rollout_corr/rollout_ppl:np.float64(1.0561046437360346) - rollout_corr/rollout_log_ppl:np.float64(0.05304120097935083) - rollout_corr/log_ppl_diff:np.float64(0.0006709559474984417) - rollout_corr/log_ppl_abs_diff:np.float64(0.001628608099053963) - rollout_corr/log_ppl_diff_max:np.float64(0.01552504301071167) - rollout_corr/log_ppl_diff_min:np.float64(-0.005156993865966797) - rollout_corr/ppl_ratio:np.float64(1.0006739916279912) - rollout_corr/chi2_token:np.float64(0.0007216199301183224) - rollout_corr/chi2_seq:np.float64(0.15093162842094898) - actor/pg_loss:np.float64(0.0006925989873707294) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00034995409805560485) - actor/ppo_kl:np.float64(0.00022765648916411863) - 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(0.71875) - self_distillation/token_reweight_context_sample_fraction:np.float64(0.71875) - self_distillation/token_reweight_correct_sample_fraction:np.float64(0.68359375) - self_distillation/token_reweight_w_mean:np.float64(32783.957445921376) - self_distillation/token_reweight_w_std:np.float64(431654.97193862277) - self_distillation/token_reweight_w_clipped_mean:np.float64(1.0002633819822222) - self_distillation/token_reweight_w_clip_frac:np.float64(0.012708691239822656) - self_distillation/empty_target_batch:np.float64(0.28125) - actor/grad_norm:np.float64(0.21034935116767883) - perf/mfu/actor:np.float64(0.03965654144334235) - perf/max_memory_allocated_gb:np.float64(117.12247371673584) - perf/max_memory_reserved_gb:np.float64(121.533203125) - perf/cpu_memory_used_gb:np.float64(147.16178703308105) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7686170212765957) - val-aux/sciknoweval/reward/std@16:np.float64(0.04264975776630575) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7854148936170213) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.02918137925697454) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.7512340425531915) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.042424831149873776) - val-aux/sciknoweval/
|
||
("Final validation metrics: {'val-aux/sciknoweval/reward/mean@16': "
|
||
"np.float64(0.7686170212765957), 'val-aux/sciknoweval/reward/std@16': "
|
||
"np.float64(0.04264975776630575), 'val-aux/sciknoweval/reward/best@2/mean': "
|
||
"np.float64(0.7854148936170213), 'val-aux/sciknoweval/reward/best@2/std': "
|
||
"np.float64(0.02918137925697454), 'val-aux/sciknoweval/reward/worst@2/mean': "
|
||
"np.float64(0.7512340425531915), 'val-aux/sciknoweval/reward/worst@2/std': "
|
||
"np.float64(0.042424831149873776), 'val-aux/sciknoweval/reward/maj@2/mean': "
|
||
"np.float64(0.7684468085106384), 'val-aux/sciknoweval/reward/maj@2/std': "
|
||
"np.float64(0.04273575303351148), 'val-aux/sciknoweval/reward/best@4/mean': "
|
||
"np.float64(0.7959787234042554), 'val-aux/sciknoweval/reward/best@4/std': "
|
||
"np.float64(0.01639341752615574), 'val-aux/sciknoweval/reward/worst@4/mean': "
|
||
"np.float64(0.7343617021276596), 'val-aux/sciknoweval/reward/worst@4/std': "
|
||
"np.float64(0.03784947100976082), 'val-aux/sciknoweval/reward/maj@4/mean': "
|
||
"np.float64(0.7722765957446808), 'val-aux/sciknoweval/reward/maj@4/std': "
|
||
"np.float64(0.031426065862592674), 'val-aux/sciknoweval/reward/best@8/mean': "
|
||
"np.float64(0.801563829787234), 'val-aux/sciknoweval/reward/best@8/std': "
|
||
"np.float64(0.008774628971156027), 'val-aux/sciknoweval/reward/worst@8/mean': "
|
||
"np.float64(0.7189574468085107), 'val-aux/sciknoweval/reward/worst@8/std': "
|
||
"np.float64(0.03147132698227977), 'val-aux/sciknoweval/reward/maj@8/mean': "
|
||
"np.float64(0.7746063829787234), 'val-aux/sciknoweval/reward/maj@8/std': "
|
||
"np.float64(0.021830599810403712), 'val-aux/sciknoweval/reward/best@16/mean': "
|
||
"np.float64(0.8047021276595744), 'val-aux/sciknoweval/reward/best@16/std': "
|
||
'np.float64(0.005902094678345516), '
|
||
"'val-aux/sciknoweval/reward/worst@16/mean': np.float64(0.705404255319149), "
|
||
"'val-aux/sciknoweval/reward/worst@16/std': np.float64(0.02298913016887447), "
|
||
"'val-aux/sciknoweval/reward/maj@16/mean': np.float64(0.7758829787234042), "
|
||
"'val-aux/sciknoweval/reward/maj@16/std': np.float64(0.014657220600786369), "
|
||
"'val-aux/sciknoweval/score/mean@16': np.float64(0.7686170212765957), "
|
||
"'val-aux/sciknoweval/score/std@16': np.float64(0.04264975776630575), "
|
||
"'val-aux/sciknoweval/score/best@2/mean': np.float64(0.7854148936170213), "
|
||
"'val-aux/sciknoweval/score/best@2/std': np.float64(0.02918137925697454), "
|
||
"'val-aux/sciknoweval/score/worst@2/mean': np.float64(0.7512340425531915), "
|
||
"'val-aux/sciknoweval/score/worst@2/std': np.float64(0.042424831149873776), "
|
||
"'val-aux/sciknoweval/score/maj@2/mean': np.float64(0.7684468085106384), "
|
||
"'val-aux/sciknoweval/score/maj@2/std': np.float64(0.04273575303351148), "
|
||
"'val-aux/sciknoweval/score/best@4/mean': np.float64(0.7959787234042554), "
|
||
"'val-aux/sciknoweval/score/best@4/std': np.float64(0.01639341752615574), "
|
||
"'val-aux/sciknoweval/score/worst@4/mean': np.float64(0.7343617021276596), "
|
||
"'val-aux/sciknoweval/score/worst@4/std': np.float64(0.03784947100976082), "
|
||
"'val-aux/sciknoweval/score/maj@4/mean': np.float64(0.7722765957446808), "
|
||
"'val-aux/sciknoweval/score/maj@4/std': np.float64(0.031426065862592674), "
|
||
"'val-aux/sciknoweval/score/best@8/mean': np.float64(0.801563829787234), "
|
||
"'val-aux/sciknoweval/score/best@8/std': np.float64(0.008774628971156027), "
|
||
"'val-aux/sciknoweval/score/worst@8/mean': np.float64(0.7189574468085107), "
|
||
"'val-aux/sciknoweval/score/worst@8/std': np.float64(0.03147132698227977), "
|
||
"'val-aux/sciknoweval/score/maj@8/mean': np.float64(0.7746063829787234), "
|
||
"'val-aux/sciknoweval/score/maj@8/std': np.float64(0.021830599810403712), "
|
||
"'val-aux/sciknoweval/score/best@16/mean': np.float64(0.8047021276595744), "
|
||
"'val-aux/sciknoweval/score/best@16/std': np.float64(0.005902094678345516), "
|
||
"'val-aux/sciknoweval/score/worst@16/mean': np.float64(0.705404255319149), "
|
||
"'val-aux/sciknoweval/score/worst@16/std': np.float64(0.02298913016887447), "
|
||
"'val-aux/sciknoweval/score/maj@16/mean': np.float64(0.7758829787234042), "
|
||
"'val-aux/sciknoweval/score/maj@16/std': np.float64(0.014657220600786369), "
|
||
"'val-core/sciknoweval/acc/mean@16': np.float64(0.7686170212765957), "
|
||
"'val-aux/sciknoweval/acc/std@16': np.float64(0.04264975776630575), "
|
||
"'val-aux/sciknoweval/acc/best@2/mean': np.float64(0.7854148936170213), "
|
||
"'val-aux/sciknoweval/acc/best@2/std': np.float64(0.02918137925697454), "
|
||
"'val-aux/sciknoweval/acc/worst@2/mean': np.float64(0.7512340425531915), "
|
||
"'val-aux/sciknoweval/acc/worst@2/std': np.float64(0.042424831149873776), "
|
||
"'val-aux/sciknoweval/acc/maj@2/mean': np.float64(0.7684468085106384), "
|
||
"'val-aux/sciknoweval/acc/maj@2/std': np.float64(0.04273575303351148), "
|
||
"'val-aux/sciknoweval/acc/best@4/mean': np.float64(0.7959787234042554), "
|
||
"'val-aux/sciknoweval/acc/best@4/std': np.float64(0.01639341752615574), "
|
||
"'val-aux/sciknoweval/acc/worst@4/mean': np.float64(0.7343617021276596), "
|
||
"'val-aux/sciknoweval/acc/worst@4/std': np.float64(0.03784947100976082), "
|
||
"'val-aux/sciknoweval/acc/maj@4/mean': np.float64(0.7722765957446808), "
|
||
"'val-aux/sciknoweval/acc/maj@4/std': np.float64(0.031426065862592674), "
|
||
"'val-aux/sciknoweval/acc/best@8/mean': np.float64(0.801563829787234), "
|
||
"'val-aux/sciknoweval/acc/best@8/std': np.float64(0.008774628971156027), "
|
||
"'val-aux/sciknoweval/acc/worst@8/mean': np.float64(0.7189574468085107), "
|
||
"'val-aux/sciknoweval/acc/worst@8/std': np.float64(0.03147132698227977), "
|
||
"'val-aux/sciknoweval/acc/maj@8/mean': np.float64(0.7746063829787234), "
|
||
"'val-aux/sciknoweval/acc/maj@8/std': np.float64(0.021830599810403712), "
|
||
"'val-core/sciknoweval/acc/best@16/mean': np.float64(0.8047021276595744), "
|
||
"'val-core/sciknoweval/acc/best@16/std': np.float64(0.005902094678345516), "
|
||
"'val-aux/sciknoweval/acc/worst@16/mean': np.float64(0.705404255319149), "
|
||
"'val-aux/sciknoweval/acc/worst@16/std': np.float64(0.02298913016887447), "
|
||
"'val-core/sciknoweval/acc/maj@16/mean': np.float64(0.7758829787234042), "
|
||
"'val-core/sciknoweval/acc/maj@16/std': np.float64(0.014657220600786369), "
|
||
"'val-aux/sciknoweval/incorrect_format/mean@16': np.float64(1.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/std@16': np.float64(0.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/best@2/mean': np.float64(1.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/best@2/std': np.float64(0.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/worst@2/mean': np.float64(1.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/worst@2/std': np.float64(0.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/maj@2/mean': np.float64(1.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/maj@2/std': np.float64(0.0), "
|
||
"'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(1.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/worst@4/std': np.float64(0.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/maj@4/mean': np.float64(1.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/maj@4/std': np.float64(0.0), "
|
||
"'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(1.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/worst@8/std': np.float64(0.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/maj@8/mean': np.float64(1.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/maj@8/std': np.float64(0.0), "
|
||
"'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(1.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/worst@16/std': np.float64(0.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/maj@16/mean': np.float64(1.0), "
|
||
"'val-aux/sciknoweval/incorrect_format/maj@16/std': np.float64(0.0), "
|
||
"'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)}")
|