1048 lines
428 KiB
Plaintext
1048 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-GRPO-Qwen-Qwen3-8B-mbs8-train32-rollout8-lr1e-6-vllm0.8/latest_checkpointed_iteration.txt
|
||
Training from scratch
|
||
Training Progress: 9%|▉ | 9/100 [04:55<46:46, 30.84s/it]
|
||
step:1 - global_seqlen/min:18271 - global_seqlen/max:28190 - global_seqlen/minmax_diff:9919 - global_seqlen/balanced_min:23425 - global_seqlen/balanced_max:23428 - global_seqlen/mean:23426.0 - actor/entropy:0.13164295256137848 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5248507261276245 - training/rollout_probs_diff_mean:0.0027477943804115057 - training/rollout_probs_diff_std:0.01053619384765625 - training/rollout_actor_probs_pearson_corr:0.9984213709831238 - rollout_corr/rollout_is_mean:1.0000320672988892 - rollout_corr/rollout_is_ratio_fraction_high:2.128384221578017e-05 - rollout_corr/rollout_is_ratio_fraction_low:2.128384221578017e-05 - rollout_corr/rollout_is_max:2.399998903274536 - rollout_corr/rollout_is_min:0.33881837129592896 - rollout_corr/rollout_is_std:0.02844212017953396 - rollout_corr/rollout_is_eff_sample_size:0.9991918186557794 - rollout_corr/rollout_is_seq_mean:1.0000114440917969 - rollout_corr/rollout_is_seq_std:0.0015418194234371185 - rollout_corr/rollout_is_seq_max:1.0049630403518677 - rollout_corr/rollout_is_seq_min:0.9952175617218018 - rollout_corr/rollout_is_seq_max_deviation:0.004963040351867676 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1415632287971675) - rollout_corr/training_log_ppl:np.float64(0.1294490935615613) - rollout_corr/kl:np.float64(0.00041570499640330993) - rollout_corr/k3_kl:np.float64(0.000427163151385912) - rollout_corr/rollout_ppl:np.float64(1.141070518642664) - rollout_corr/rollout_log_ppl:np.float64(0.12903338878822979) - rollout_corr/log_ppl_diff:np.float64(0.00041570477333152667) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011931666304008104) - rollout_corr/log_ppl_diff_max:np.float64(0.005768194794654846) - rollout_corr/log_ppl_diff_min:np.float64(-0.004562102258205414) - rollout_corr/ppl_ratio:np.float64(1.000416962429881) - rollout_corr/chi2_token:np.float64(0.0009065121412277222) - rollout_corr/chi2_seq:np.float64(0.4969969654921442) - actor/pg_loss:np.float64(-0.00010641675908118486) - 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) - actor/grad_norm:np.float64(0.20275606587529182) - perf/mfu/actor:np.float64(0.09007042783248102) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(108.43737363815308) - actor/lr:np.float64(0.0) - training/global_step:1 - training/epoch:0 - 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:0.0022427847143262625 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.0022427847143262625 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:367.0625 - response_length/max:775.0 - response_length/min:149.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:367.0625 - response_length_non_aborted/max:775.0 - response_length_non_aborted/min:149.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:365.0 - prompt_length/max:466.0 - prompt_length/min:197.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.0002685021609067917 - timing_s/agent_loop/generate_sequences/min:np.float64(12.760501340031624) - timing_s/agent_loop/generate_sequences/max:np.float64(22.61353741772473) - timing_s/agent_loop/generate_sequences/mean:np.float64(17.434483396420546) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(22.61353741772473) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/prompt_length:266
|
||
step:2 - global_seqlen/min:20061 - global_seqlen/max:28595 - global_seqlen/minmax_diff:8534 - global_seqlen/balanced_min:23472 - global_seqlen/balanced_max:23477 - global_seqlen/mean:23475.625 - actor/entropy:0.14814899861812592 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4770219326019287 - training/rollout_probs_diff_mean:0.0028117208275943995 - training/rollout_probs_diff_std:0.01036490686237812 - training/rollout_actor_probs_pearson_corr:0.998557984828949 - rollout_corr/rollout_is_mean:0.9999970197677612 - rollout_corr/rollout_is_ratio_fraction_high:1.1147153600177262e-05 - rollout_corr/rollout_is_ratio_fraction_low:3.344146171002649e-05 - rollout_corr/rollout_is_max:3.167215347290039 - rollout_corr/rollout_is_min:0.3864081799983978 - rollout_corr/rollout_is_std:0.029656805098056793 - rollout_corr/rollout_is_eff_sample_size:0.9991210087484641 - rollout_corr/rollout_is_seq_mean:1.0000070333480835 - rollout_corr/rollout_is_seq_std:0.0019922589417546988 - rollout_corr/rollout_is_seq_max:1.0065020322799683 - rollout_corr/rollout_is_seq_min:0.9941989183425903 - rollout_corr/rollout_is_seq_max_deviation:0.006502032279968262 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1674070176668465) - rollout_corr/training_log_ppl:np.float64(0.15027856942469953) - rollout_corr/kl:np.float64(0.0003824043578806169) - rollout_corr/k3_kl:np.float64(0.00046587164479738874) - rollout_corr/rollout_ppl:np.float64(1.1669510854408145) - rollout_corr/rollout_log_ppl:np.float64(0.14989616414823104) - rollout_corr/log_ppl_diff:np.float64(0.0003824052764684893) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013567420028266497) - rollout_corr/log_ppl_diff_max:np.float64(0.009369313716888428) - rollout_corr/log_ppl_diff_min:np.float64(-0.005993887782096863) - rollout_corr/ppl_ratio:np.float64(1.0003841957077384) - rollout_corr/chi2_token:np.float64(0.0010940337087959051) - rollout_corr/chi2_seq:np.float64(0.27721586590632796) - actor/pg_loss:np.float64(-4.244828596711159e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0001937607421496068) - actor/ppo_kl:np.float64(-7.490384608743739e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.35917383059859276) - perf/mfu/actor:np.float64(0.09536108344592256) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(107.28438568115234) - actor/lr:np.float64(1e-07) - training/global_step:2 - training/epoch:0 - critic/score/mean:0.421875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.421875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.00117184454575181 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00117184454575181 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:350.42578125 - response_length/max:815.0 - response_length/min:143.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:350.42578125 - response_length_non_aborted/max:815.0 - response_length_non_aborted/min:143.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:383.1875 - prompt_length/max:494.0 - prompt_length/min:197.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00014822930097579956 - timing_s/agent_loop/generate_sequences/min:np.float64(1.828049125149846) - timing_s/agent_loop/generate_sequences/max:np.float64(7.133984874933958) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.004583149573591) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.133984874933958) - timing_s/agent_loop/slowest/tool_calls:np.float64
|
||
step:3 - global_seqlen/min:18861 - global_seqlen/max:27601 - global_seqlen/minmax_diff:8740 - global_seqlen/balanced_min:23577 - global_seqlen/balanced_max:23579 - global_seqlen/mean:23578.0 - actor/entropy:0.15725769102573395 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.8945247530937195 - training/rollout_probs_diff_mean:0.0028085215017199516 - training/rollout_probs_diff_std:0.010637396946549416 - training/rollout_actor_probs_pearson_corr:0.998668372631073 - rollout_corr/rollout_is_mean:1.0000150203704834 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:3.321229451103136e-05 - rollout_corr/rollout_is_max:1.644019603729248 - rollout_corr/rollout_is_min:0.02608698606491089 - rollout_corr/rollout_is_std:0.029119335114955902 - rollout_corr/rollout_is_eff_sample_size:0.9991529017165889 - rollout_corr/rollout_is_seq_mean:0.9999891519546509 - rollout_corr/rollout_is_seq_std:0.001679171109572053 - rollout_corr/rollout_is_seq_max:1.0052025318145752 - rollout_corr/rollout_is_seq_min:0.9923624396324158 - rollout_corr/rollout_is_seq_max_deviation:0.0076375603675842285 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1635099058039486) - rollout_corr/training_log_ppl:np.float64(0.14788767747813836) - rollout_corr/kl:np.float64(0.0006985796255651877) - rollout_corr/k3_kl:np.float64(0.0004934847250837038) - rollout_corr/rollout_ppl:np.float64(1.1626955484971404) - rollout_corr/rollout_log_ppl:np.float64(0.14718909670045832) - rollout_corr/log_ppl_diff:np.float64(0.0006985807776800357) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015300126760848798) - rollout_corr/log_ppl_diff_max:np.float64(0.020397737622261047) - rollout_corr/log_ppl_diff_min:np.float64(-0.004551559686660767) - rollout_corr/ppl_ratio:np.float64(1.00070138880983) - rollout_corr/chi2_token:np.float64(0.0004848798271268606) - rollout_corr/chi2_seq:np.float64(0.2387628396973014) - actor/pg_loss:np.float64(3.924837801605463e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003787184837165114) - actor/ppo_kl:np.float64(0.0001540237209205486) - actor/pg_clipfrac_lower:np.float64(2.411265450064093e-05) - actor/grad_norm:np.float64(0.22798091545701027) - perf/mfu/actor:np.float64(0.0946753525771929) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.40783309936523) - actor/lr:np.float64(2e-07) - training/global_step:3 - training/epoch:0 - critic/score/mean:0.48828125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.48828125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.002475699642673135 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.002475699642673135 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:352.84375 - response_length/max:778.0 - response_length/min:139.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:352.84375 - response_length_non_aborted/max:778.0 - response_length_non_aborted/min:139.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:383.96875 - prompt_length/max:502.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:9.593181312084198e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8824504688382149) - timing_s/agent_loop/generate_sequences/max:np.float64(7.000263413414359) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.041497490667098) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.000263413414359) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - t
|
||
step:4 - global_seqlen/min:18454 - global_seqlen/max:25358 - global_seqlen/minmax_diff:6904 - global_seqlen/balanced_min:22418 - global_seqlen/balanced_max:22423 - global_seqlen/mean:22421.25 - actor/entropy:0.16384702920913696 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2932341396808624 - training/rollout_probs_diff_mean:0.0029966789297759533 - training/rollout_probs_diff_std:0.010196198709309101 - training/rollout_actor_probs_pearson_corr:0.9987289309501648 - rollout_corr/rollout_is_mean:0.9998806715011597 - rollout_corr/rollout_is_ratio_fraction_high:1.2197204341646284e-05 - rollout_corr/rollout_is_ratio_fraction_low:3.659161302493885e-05 - rollout_corr/rollout_is_max:2.14375901222229 - rollout_corr/rollout_is_min:0.38381150364875793 - rollout_corr/rollout_is_std:0.029936857521533966 - rollout_corr/rollout_is_eff_sample_size:0.999104349052325 - rollout_corr/rollout_is_seq_mean:0.9998318552970886 - rollout_corr/rollout_is_seq_std:0.0017542371060699224 - rollout_corr/rollout_is_seq_max:1.0063813924789429 - rollout_corr/rollout_is_seq_min:0.9924446940422058 - rollout_corr/rollout_is_seq_max_deviation:0.0075553059577941895 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1746212770231068) - rollout_corr/training_log_ppl:np.float64(0.15730432402779115) - rollout_corr/kl:np.float64(0.000751812893312831) - rollout_corr/k3_kl:np.float64(0.0005046583516019609) - rollout_corr/rollout_ppl:np.float64(1.1737040071748197) - rollout_corr/rollout_log_ppl:np.float64(0.15655251036514528) - rollout_corr/log_ppl_diff:np.float64(0.0007518136626458727) - rollout_corr/log_ppl_abs_diff:np.float64(0.001456380974559579) - rollout_corr/log_ppl_diff_max:np.float64(0.014719516038894653) - rollout_corr/log_ppl_diff_min:np.float64(-0.0057411715388298035) - rollout_corr/ppl_ratio:np.float64(1.0007540648803115) - rollout_corr/chi2_token:np.float64(0.0004748636856675148) - rollout_corr/chi2_seq:np.float64(0.012265707831829786) - actor/pg_loss:np.float64(8.195138070732355e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00033394731326552574) - actor/ppo_kl:np.float64(0.00012522640624190728) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.3742156885564327) - perf/mfu/actor:np.float64(0.09129534392392291) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(105.75098037719727) - actor/lr:np.float64(3e-07) - training/global_step:4 - training/epoch:0 - critic/score/mean:0.5 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0005473495693877339 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0005473495693877339 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:320.2578125 - response_length/max:637.0 - response_length/min:112.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:320.2578125 - response_length_non_aborted/max:637.0 - response_length_non_aborted/min:112.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:380.40625 - prompt_length/max:481.0 - prompt_length/min:196.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.0001420937478542328 - timing_s/agent_loop/generate_sequences/min:np.float64(1.4736281912773848) - timing_s/agent_loop/generate_sequences/max:np.float64(6.098991813138127) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.7255679049631) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.098991813138127) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - ti
|
||
step:5 - global_seqlen/min:20596 - global_seqlen/max:28411 - global_seqlen/minmax_diff:7815 - global_seqlen/balanced_min:25319 - global_seqlen/balanced_max:25325 - global_seqlen/mean:25322.125 - actor/entropy:0.14316131174564362 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2733950614929199 - training/rollout_probs_diff_mean:0.0025193935725837946 - training/rollout_probs_diff_std:0.009297977201640606 - training/rollout_actor_probs_pearson_corr:0.9988449215888977 - rollout_corr/rollout_is_mean:0.9999367594718933 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.9515424966812134 - rollout_corr/rollout_is_min:0.5123908519744873 - rollout_corr/rollout_is_std:0.027232352644205093 - rollout_corr/rollout_is_eff_sample_size:0.9992588295415155 - rollout_corr/rollout_is_seq_mean:0.9999023675918579 - rollout_corr/rollout_is_seq_std:0.001504938816651702 - rollout_corr/rollout_is_seq_max:1.004489779472351 - rollout_corr/rollout_is_seq_min:0.9948562383651733 - rollout_corr/rollout_is_seq_max_deviation:0.00514376163482666 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1670298031531274) - rollout_corr/training_log_ppl:np.float64(0.1497556624308345) - rollout_corr/kl:np.float64(0.00048706240530016487) - rollout_corr/k3_kl:np.float64(0.0004268709189929609) - rollout_corr/rollout_ppl:np.float64(1.1664286046288908) - rollout_corr/rollout_log_ppl:np.float64(0.14926859955812688) - rollout_corr/log_ppl_diff:np.float64(0.00048706287270761095) - rollout_corr/log_ppl_abs_diff:np.float64(0.001265541370230494) - rollout_corr/log_ppl_diff_max:np.float64(0.006666272878646851) - rollout_corr/log_ppl_diff_min:np.float64(-0.0046197474002838135) - rollout_corr/ppl_ratio:np.float64(1.0004885597154498) - rollout_corr/chi2_token:np.float64(0.0007471756543964148) - rollout_corr/chi2_seq:np.float64(0.32634232798591256) - actor/pg_loss:np.float64(-6.317137740552425e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(9.448716127735679e-05) - actor/ppo_kl:np.float64(-2.549028512888185e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.20024448819458485) - perf/mfu/actor:np.float64(0.10293015089772223) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(105.27999114990234) - actor/lr:np.float64(4e-07) - training/global_step:5 - training/epoch:0 - critic/score/mean:0.54296875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.54296875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.008730988018214703 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.008730988018214703 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:396.28515625 - response_length/max:770.0 - response_length/min:153.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:396.28515625 - response_length_non_aborted/max:770.0 - response_length_non_aborted/min:153.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:395.03125 - prompt_length/max:512.0 - prompt_length/min:196.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00010201521217823029 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8949396945536137) - timing_s/agent_loop/generate_sequences/max:np.float64(7.308365289121866) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.612634117438574) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.308365289121866) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowe
|
||
step:6 - global_seqlen/min:22162 - global_seqlen/max:26539 - global_seqlen/minmax_diff:4377 - global_seqlen/balanced_min:24656 - global_seqlen/balanced_max:24659 - global_seqlen/mean:24657.25 - actor/entropy:0.17344678938388824 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4341905415058136 - training/rollout_probs_diff_mean:0.002906338544562459 - training/rollout_probs_diff_std:0.00988501962274313 - training/rollout_actor_probs_pearson_corr:0.9988127946853638 - rollout_corr/rollout_is_mean:0.9997435808181763 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:4.1787676309468225e-05 - rollout_corr/rollout_is_max:1.6279510259628296 - rollout_corr/rollout_is_min:0.11007747054100037 - rollout_corr/rollout_is_std:0.029173526912927628 - rollout_corr/rollout_is_eff_sample_size:0.9991490934939479 - rollout_corr/rollout_is_seq_mean:0.9997667074203491 - rollout_corr/rollout_is_seq_std:0.0015804192516952753 - rollout_corr/rollout_is_seq_max:1.0045722723007202 - rollout_corr/rollout_is_seq_min:0.9951204657554626 - rollout_corr/rollout_is_seq_max_deviation:0.0048795342445373535 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1923689707182348) - rollout_corr/training_log_ppl:np.float64(0.16982925602496834) - rollout_corr/kl:np.float64(0.0005402864753527581) - rollout_corr/k3_kl:np.float64(0.0004851298600740961) - rollout_corr/rollout_ppl:np.float64(1.191715772729367) - rollout_corr/rollout_log_ppl:np.float64(0.16928896902390989) - rollout_corr/log_ppl_diff:np.float64(0.0005402870010584593) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013106374244671315) - rollout_corr/log_ppl_diff_max:np.float64(0.0053055137395858765) - rollout_corr/log_ppl_diff_min:np.float64(-0.003997787833213806) - rollout_corr/ppl_ratio:np.float64(1.000541717512533) - rollout_corr/chi2_token:np.float64(0.0010020597837865353) - rollout_corr/chi2_seq:np.float64(0.2143058932852) - actor/pg_loss:np.float64(0.00018408475443720818) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002850766641131486) - actor/ppo_kl:np.float64(-0.0001537304164536124) - actor/pg_clipfrac_lower:np.float64(2.1961745005683042e-05) - actor/grad_norm:np.float64(0.2505660653114319) - perf/mfu/actor:np.float64(0.10075013742375999) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(105.11516189575195) - actor/lr:np.float64(5e-07) - training/global_step:6 - training/epoch:0 - critic/score/mean:0.48828125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.48828125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.006038319319486618 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.006038319319486618 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:373.9140625 - response_length/max:889.0 - response_length/min:167.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:373.9140625 - response_length_non_aborted/max:889.0 - response_length_non_aborted/min:167.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:396.625 - prompt_length/max:477.0 - prompt_length/min:190.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00010812468826770782 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9656961932778358) - timing_s/agent_loop/generate_sequences/max:np.float64(7.710826674476266) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.21906588661659) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.710826674476266) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.
|
||
step:7 - global_seqlen/min:20792 - global_seqlen/max:24924 - global_seqlen/minmax_diff:4132 - global_seqlen/balanced_min:22999 - global_seqlen/balanced_max:23003 - global_seqlen/mean:23001.25 - actor/entropy:0.1555243581533432 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6118559837341309 - training/rollout_probs_diff_mean:0.0028650695458054543 - training/rollout_probs_diff_std:0.010248509235680103 - training/rollout_actor_probs_pearson_corr:0.9986856579780579 - rollout_corr/rollout_is_mean:0.9999462366104126 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:5.987020267639309e-05 - rollout_corr/rollout_is_max:1.6872659921646118 - rollout_corr/rollout_is_min:0.21203859150409698 - rollout_corr/rollout_is_std:0.029119335114955902 - rollout_corr/rollout_is_eff_sample_size:0.9991526637018234 - rollout_corr/rollout_is_seq_mean:0.9998912215232849 - rollout_corr/rollout_is_seq_std:0.0016600432572886348 - rollout_corr/rollout_is_seq_max:1.0042778253555298 - rollout_corr/rollout_is_seq_min:0.9954917430877686 - rollout_corr/rollout_is_seq_max_deviation:0.004508256912231445 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1648076949641109) - rollout_corr/training_log_ppl:np.float64(0.14963223517406732) - rollout_corr/kl:np.float64(0.0005729907538167822) - rollout_corr/k3_kl:np.float64(0.0005188208063344746) - rollout_corr/rollout_ppl:np.float64(1.1641330625861883) - rollout_corr/rollout_log_ppl:np.float64(0.1490592439440661) - rollout_corr/log_ppl_diff:np.float64(0.0005729912300012074) - rollout_corr/log_ppl_abs_diff:np.float64(0.001507752669567708) - rollout_corr/log_ppl_diff_max:np.float64(0.013988256454467773) - rollout_corr/log_ppl_diff_min:np.float64(-0.0046309977769851685) - rollout_corr/ppl_ratio:np.float64(1.0005752036813647) - rollout_corr/chi2_token:np.float64(0.0009152269922196865) - rollout_corr/chi2_seq:np.float64(0.3741769967600703) - actor/pg_loss:np.float64(-0.00010460847988724709) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003684248868012219) - actor/ppo_kl:np.float64(3.7074248737667403e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.24469587951898575) - perf/mfu/actor:np.float64(0.09409411841953638) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(105.16532897949219) - actor/lr:np.float64(6e-07) - training/global_step:7 - 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:0.00369848171249032 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00369848171249032 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:326.2265625 - response_length/max:685.0 - response_length/min:136.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:326.2265625 - response_length_non_aborted/max:685.0 - response_length_non_aborted/min:136.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:392.5625 - prompt_length/max:477.0 - prompt_length/min:267.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00010204501450061798 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8541261740028858) - timing_s/agent_loop/generate_sequences/max:np.float64(6.407487591728568) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.88140053445386) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.407487591728568) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/ag
|
||
step:8 - global_seqlen/min:23566 - global_seqlen/max:29329 - global_seqlen/minmax_diff:5763 - global_seqlen/balanced_min:25937 - global_seqlen/balanced_max:25943 - global_seqlen/mean:25940.625 - actor/entropy:0.13531209528446198 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.44297340512275696 - training/rollout_probs_diff_mean:0.0025063324719667435 - training/rollout_probs_diff_std:0.009608217515051365 - training/rollout_actor_probs_pearson_corr:0.9987505078315735 - rollout_corr/rollout_is_mean:0.9998748898506165 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.9733790395548567e-05 - rollout_corr/rollout_is_max:1.68257474899292 - rollout_corr/rollout_is_min:0.21203859150409698 - rollout_corr/rollout_is_std:0.02694188803434372 - rollout_corr/rollout_is_eff_sample_size:0.9992746611342851 - rollout_corr/rollout_is_seq_mean:0.9998258948326111 - rollout_corr/rollout_is_seq_std:0.0016792099922895432 - rollout_corr/rollout_is_seq_max:1.0077191591262817 - rollout_corr/rollout_is_seq_min:0.995216965675354 - rollout_corr/rollout_is_seq_max_deviation:0.007719159126281738 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1455142376944423) - rollout_corr/training_log_ppl:np.float64(0.13238640106283128) - rollout_corr/kl:np.float64(0.0006064566179098119) - rollout_corr/k3_kl:np.float64(0.00042834825232063167) - rollout_corr/rollout_ppl:np.float64(1.1448127096518874) - rollout_corr/rollout_log_ppl:np.float64(0.13177994473517174) - rollout_corr/log_ppl_diff:np.float64(0.0006064563276595436) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013034501535003074) - rollout_corr/log_ppl_diff_max:np.float64(0.007684662938117981) - rollout_corr/log_ppl_diff_min:np.float64(-0.0044105201959609985) - rollout_corr/ppl_ratio:np.float64(1.0006079657468945) - rollout_corr/chi2_token:np.float64(0.0004968128632754087) - rollout_corr/chi2_seq:np.float64(0.217244369443506) - actor/pg_loss:np.float64(-2.9583810828626156e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0003115442459602491) - actor/ppo_kl:np.float64(3.74101371178881e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1923805233091116) - perf/mfu/actor:np.float64(0.10503296560951474) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(105.0584716796875) - actor/lr:np.float64(7e-07) - training/global_step:8 - training/epoch:0 - critic/score/mean:0.54296875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.54296875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0034040790051221848 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0034040790051221848 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:395.89453125 - response_length/max:825.0 - response_length/min:154.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:395.89453125 - response_length_non_aborted/max:825.0 - response_length_non_aborted/min:154.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:414.75 - prompt_length/max:480.0 - prompt_length/min:264.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00011474452912807465 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9491427764296532) - timing_s/agent_loop/generate_sequences/max:np.float64(7.612683465704322) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.6518478234283975) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.612683465704322) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_
|
||
step:9 - global_seqlen/min:21362 - global_seqlen/max:27415 - global_seqlen/minmax_diff:6053 - global_seqlen/balanced_min:25512 - global_seqlen/balanced_max:25520 - global_seqlen/mean:25517.625 - actor/entropy:0.13916555047035217 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.289775013923645 - training/rollout_probs_diff_mean:0.0025533048901706934 - training/rollout_probs_diff_std:0.00933651439845562 - training/rollout_actor_probs_pearson_corr:0.9987970590591431 - rollout_corr/rollout_is_mean:0.9998382329940796 - rollout_corr/rollout_is_ratio_fraction_high:1.0290924365108367e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.0290924365108367e-05 - rollout_corr/rollout_is_max:2.3356175422668457 - rollout_corr/rollout_is_min:0.48794692754745483 - rollout_corr/rollout_is_std:0.02705448307096958 - rollout_corr/rollout_is_eff_sample_size:0.9992684712788912 - rollout_corr/rollout_is_seq_mean:0.9998385906219482 - rollout_corr/rollout_is_seq_std:0.0013983597746118903 - rollout_corr/rollout_is_seq_max:1.005619764328003 - rollout_corr/rollout_is_seq_min:0.9954448342323303 - rollout_corr/rollout_is_seq_max_deviation:0.00561976432800293 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1501446971669793) - rollout_corr/training_log_ppl:np.float64(0.13680826713243732) - rollout_corr/kl:np.float64(0.0007333146883379271) - rollout_corr/k3_kl:np.float64(0.0005522432770845853) - rollout_corr/rollout_ppl:np.float64(1.149286507628858) - rollout_corr/rollout_log_ppl:np.float64(0.13607495112955803) - rollout_corr/log_ppl_diff:np.float64(0.0007333160028792918) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014585833123419434) - rollout_corr/log_ppl_diff_max:np.float64(0.010762542486190796) - rollout_corr/log_ppl_diff_min:np.float64(-0.005348235368728638) - rollout_corr/ppl_ratio:np.float64(1.0007353429682553) - rollout_corr/chi2_token:np.float64(0.0008337069302797318) - rollout_corr/chi2_seq:np.float64(0.378167666727677) - actor/pg_loss:np.float64(7.547589484602213e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0006460250629061193) - actor/ppo_kl:np.float64(0.00018914890105747872) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.2567829228937626) - perf/mfu/actor:np.float64(0.10245365672343118) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(105.11789321899414) - actor/lr:np.float64(8e-07) - training/global_step:9 - training/epoch:0 - 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:0.00803849846124649 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00803849846124649 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:379.58203125 - response_length/max:680.0 - response_length/min:145.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:379.58203125 - response_length_non_aborted/max:680.0 - response_length_non_aborted/min:145.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:417.84375 - prompt_length/max:541.0 - prompt_length/min:262.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00014753639698028564 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9016973823308945) - timing_s/agent_loop/generate_sequences/max:np.float64(7.0271569583565) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.523215597298986) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.0271569583565) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - ti
|
||
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 given crystal structure is orthorhombic, as all the angles (α, β, γ) are 90.00º. The volume $ V $ of a unit cell in an orthorhombic system is calculated using the formula:
|
||
|
||
$$
|
||
V = a \times b \times c
|
||
$$
|
||
|
||
Where:
|
||
- $ 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 $:
|
||
|
||
$$
|
||
7.82 \times 7.82 = 61.1524
|
||
$$
|
||
|
||
Now multiply by $ 17.88 $:
|
||
|
||
$$
|
||
61.1524 \times 17.88 = 1094.31
|
||
$$
|
||
|
||
Thus, the volume of the unit cell is $ 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-GRPO-Qwen-Qwen3-8B-mbs8-train32-rollout8-lr1e-6-vllm0.8/global_step_10
|
||
step:10 - global_seqlen/min:21528 - global_seqlen/max:30348 - global_seqlen/minmax_diff:8820 - global_seqlen/balanced_min:25597 - global_seqlen/balanced_max:25600 - global_seqlen/mean:25598.625 - actor/entropy:0.16920506954193115 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6033204197883606 - training/rollout_probs_diff_mean:0.0029068614821881056 - training/rollout_probs_diff_std:0.009937559254467487 - training/rollout_actor_probs_pearson_corr:0.9988886713981628 - rollout_corr/rollout_is_mean:0.999866783618927 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.9597660866566002e-05 - rollout_corr/rollout_is_max:1.6955915689468384 - rollout_corr/rollout_is_min:0.11169125884771347 - rollout_corr/rollout_is_std:0.028244443237781525 - rollout_corr/rollout_is_eff_sample_size:0.9992025302719866 - rollout_corr/rollout_is_seq_mean:0.9998682737350464 - rollout_corr/rollout_is_seq_std:0.0015377062372863293 - rollout_corr/rollout_is_seq_max:1.004624605178833 - rollout_corr/rollout_is_seq_min:0.994768500328064 - rollout_corr/rollout_is_seq_max_deviation:0.005231499671936035 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1872184653766453) - rollout_corr/training_log_ppl:np.float64(0.16784915810421808) - rollout_corr/kl:np.float64(0.000722991060116307) - rollout_corr/k3_kl:np.float64(0.0005227115022989892) - rollout_corr/rollout_ppl:np.float64(1.1863347007893026) - rollout_corr/rollout_log_ppl:np.float64(0.16712616474978859) - rollout_corr/log_ppl_diff:np.float64(0.0007229933544294909) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014674967678729445) - rollout_corr/log_ppl_diff_max:np.float64(0.006441354751586914) - rollout_corr/log_ppl_diff_min:np.float64(-0.003609776496887207) - rollout_corr/ppl_ratio:np.float64(1.0007247407920659) - rollout_corr/chi2_token:np.float64(0.0006014511454850435) - rollout_corr/chi2_seq:np.float64(0.217362450202927) - actor/pg_loss:np.float64(-0.00012342154514044523) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00042188452925984166) - actor/ppo_kl:np.float64(0.0001569013978262035) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.241903368383646) - perf/mfu/actor:np.float64(0.10273077133253324) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(105.04884338378906) - actor/lr:np.float64(9e-07) - val-aux/sciknoweval/reward/mean@16:np.float64(0.6795212765957447) - val-aux/sciknoweval/reward/std@16:np.float64(0.12984114398296573) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7279042553191489) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.08743747296198659) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6308191489361703) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.12907308985602772) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.6793829787234041) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.13009488275314807) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7579255319148935) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.05743236777991458) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.580531914893617) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.12283448124701997) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.697404255319149) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.09072076266889331) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7801914893617022) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.039544738199845254) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.526968085106383) - val-aux/sciknoweval/reward/worst@8/std:np.float64(0.10635135878078301) - val-aux/sciknoweval/reward/maj@8/mean:np.float64(0.7053510638297872) - val-aux/sciknoweval/reward/maj@8/std:np.float64(0.05191007682205249) - val-aux/sciknoweval/reward/best@16/mean:np.float64(0.7965425531914894) - val-aux/scikno
|
||
step:11 - global_seqlen/min:23579 - global_seqlen/max:28112 - global_seqlen/minmax_diff:4533 - global_seqlen/balanced_min:26452 - global_seqlen/balanced_max:26455 - global_seqlen/mean:26453.625 - actor/entropy:0.1441550999879837 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5545997023582458 - training/rollout_probs_diff_mean:0.0024969361256808043 - training/rollout_probs_diff_std:0.009402581490576267 - training/rollout_actor_probs_pearson_corr:0.9987636208534241 - rollout_corr/rollout_is_mean:1.0001752376556396 - rollout_corr/rollout_is_ratio_fraction_high:8.951189556682948e-06 - rollout_corr/rollout_is_ratio_fraction_low:2.6853567760554142e-05 - rollout_corr/rollout_is_max:3.7230892181396484 - rollout_corr/rollout_is_min:0.25212857127189636 - rollout_corr/rollout_is_std:0.02670188620686531 - rollout_corr/rollout_is_eff_sample_size:0.9992878743513974 - rollout_corr/rollout_is_seq_mean:1.0002554655075073 - rollout_corr/rollout_is_seq_std:0.0014197564451023936 - rollout_corr/rollout_is_seq_max:1.004997730255127 - rollout_corr/rollout_is_seq_min:0.9965838193893433 - rollout_corr/rollout_is_seq_max_deviation:0.004997730255126953 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1556855565868318) - rollout_corr/training_log_ppl:np.float64(0.14192374088452198) - rollout_corr/kl:np.float64(0.0001279704757504163) - rollout_corr/k3_kl:np.float64(0.0004536967828556726) - rollout_corr/rollout_ppl:np.float64(1.1555262096226215) - rollout_corr/rollout_log_ppl:np.float64(0.14179576889728196) - rollout_corr/log_ppl_diff:np.float64(0.00012797198724001646) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011966362362727523) - rollout_corr/log_ppl_diff_max:np.float64(0.005566880106925964) - rollout_corr/log_ppl_diff_min:np.float64(-0.005283966660499573) - rollout_corr/ppl_ratio:np.float64(1.0001292265951633) - rollout_corr/chi2_token:np.float64(0.0016099370550364256) - rollout_corr/chi2_seq:np.float64(0.5782006285153329) - actor/pg_loss:np.float64(5.143752787262201e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00019089993247689563) - actor/ppo_kl:np.float64(5.072445345533083e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.12007458135485649) - perf/mfu/actor:np.float64(0.10334472437206974) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(108.11588668823242) - actor/lr:np.float64(1e-06) - training/global_step:11 - training/epoch:0 - critic/score/mean:0.50390625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.50390625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.003309702267870307 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.003309702267870307 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:436.39453125 - response_length/max:721.0 - response_length/min:160.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:436.39453125 - response_length_non_aborted/max:721.0 - response_length_non_aborted/min:160.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:390.28125 - prompt_length/max:481.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.0001160837709903717 - timing_s/agent_loop/generate_sequences/min:np.float64(2.0758140552788973) - timing_s/agent_loop/generate_sequences/max:np.float64(7.61514799669385) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.149380202659813) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.61514799669385) - timing_s/agent_loop/slowest/tool_calls:np.flo
|
||
step:12 - global_seqlen/min:22402 - global_seqlen/max:29419 - global_seqlen/minmax_diff:7017 - global_seqlen/balanced_min:26481 - global_seqlen/balanced_max:26487 - global_seqlen/mean:26485.25 - actor/entropy:0.1846352517604828 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3155989646911621 - training/rollout_probs_diff_mean:0.0028140994254499674 - training/rollout_probs_diff_std:0.009286953136324883 - training/rollout_actor_probs_pearson_corr:0.9989963173866272 - rollout_corr/rollout_is_mean:1.0000858306884766 - rollout_corr/rollout_is_ratio_fraction_high:8.870752935763448e-06 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.25616192817688 - rollout_corr/rollout_is_min:0.5524312257766724 - rollout_corr/rollout_is_std:0.027619199827313423 - rollout_corr/rollout_is_eff_sample_size:0.9992378802349873 - rollout_corr/rollout_is_seq_mean:1.0001217126846313 - rollout_corr/rollout_is_seq_std:0.0015678335912525654 - rollout_corr/rollout_is_seq_max:1.0049930810928345 - rollout_corr/rollout_is_seq_min:0.9935046434402466 - rollout_corr/rollout_is_seq_max_deviation:0.006495356559753418 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.197012455202639) - rollout_corr/training_log_ppl:np.float64(0.1768503768616938) - rollout_corr/kl:np.float64(0.0005232393070224717) - rollout_corr/k3_kl:np.float64(0.0008367889608820178) - rollout_corr/rollout_ppl:np.float64(1.1963813910260797) - rollout_corr/rollout_log_ppl:np.float64(0.17632713728380622) - rollout_corr/log_ppl_diff:np.float64(0.0005232395778875798) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015441769937751815) - rollout_corr/log_ppl_diff_max:np.float64(0.03289683163166046) - rollout_corr/log_ppl_diff_min:np.float64(-0.020482316613197327) - rollout_corr/ppl_ratio:np.float64(1.000529296696186) - rollout_corr/chi2_token:np.float64(0.0047122761607170105) - rollout_corr/chi2_seq:np.float64(5.705244723241776) - actor/pg_loss:np.float64(2.2202730178833008e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00040041032389126485) - actor/ppo_kl:np.float64(0.0002610394977891062) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.2557632587850094) - perf/mfu/actor:np.float64(0.10529765568035229) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.64424991607666) - actor/lr:np.float64(1e-06) - training/global_step:12 - training/epoch:0 - critic/score/mean:0.60546875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.60546875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0005666193319484591 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0005666193319484591 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:440.3515625 - response_length/max:797.0 - response_length/min:154.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:440.3515625 - response_length_non_aborted/max:797.0 - response_length_non_aborted/min:154.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:387.3125 - prompt_length/max:482.0 - prompt_length/min:261.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:8.065253496170044e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.6090983245521784) - timing_s/agent_loop/generate_sequences/max:np.float64(7.733725985512137) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.158535132286488) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.733725985512137) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent
|
||
step:13 - global_seqlen/min:25377 - global_seqlen/max:30122 - global_seqlen/minmax_diff:4745 - global_seqlen/balanced_min:27079 - global_seqlen/balanced_max:27081 - global_seqlen/mean:27079.625 - actor/entropy:0.1729048788547516 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2444952130317688 - training/rollout_probs_diff_mean:0.0027501848526299 - training/rollout_probs_diff_std:0.009192431345582008 - training/rollout_actor_probs_pearson_corr:0.9989957809448242 - rollout_corr/rollout_is_mean:0.9998618960380554 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6526812314987183 - rollout_corr/rollout_is_min:0.5106894969940186 - rollout_corr/rollout_is_std:0.027012590318918228 - rollout_corr/rollout_is_eff_sample_size:0.9992704948770252 - rollout_corr/rollout_is_seq_mean:0.9998829364776611 - rollout_corr/rollout_is_seq_std:0.0015230097342282534 - rollout_corr/rollout_is_seq_max:1.006705641746521 - rollout_corr/rollout_is_seq_min:0.9953020215034485 - rollout_corr/rollout_is_seq_max_deviation:0.006705641746520996 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.188228433020413) - rollout_corr/training_log_ppl:np.float64(0.1695324133397662) - rollout_corr/kl:np.float64(0.0005957026775860186) - rollout_corr/k3_kl:np.float64(0.00043430524907250856) - rollout_corr/rollout_ppl:np.float64(1.1875129132531583) - rollout_corr/rollout_log_ppl:np.float64(0.1689367104190751) - rollout_corr/log_ppl_diff:np.float64(0.0005957029206911102) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012361184635665268) - rollout_corr/log_ppl_diff_max:np.float64(0.005819700658321381) - rollout_corr/log_ppl_diff_min:np.float64(-0.003224298357963562) - rollout_corr/ppl_ratio:np.float64(1.000597006175667) - rollout_corr/chi2_token:np.float64(0.0005501273553818464) - rollout_corr/chi2_seq:np.float64(0.09272657055407763) - actor/pg_loss:np.float64(-7.470371201634407e-07) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00014200475698089576) - actor/ppo_kl:np.float64(8.690033108127437e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.13939961977303028) - perf/mfu/actor:np.float64(0.10884410844931278) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.16784048080444) - actor/lr:np.float64(1e-06) - training/global_step:13 - training/epoch:0 - 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:-0.0007487679249607027 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0007487679249607027 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:430.39453125 - response_length/max:683.0 - response_length/min:151.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:430.39453125 - response_length_non_aborted/max:683.0 - response_length_non_aborted/min:151.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:415.84375 - prompt_length/max:493.0 - prompt_length/min:264.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:2.849474549293518e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.5356736686080694) - timing_s/agent_loop/generate_sequences/max:np.float64(7.184362847357988) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.028661246171396) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.184362847357988) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/
|
||
step:14 - global_seqlen/min:20977 - global_seqlen/max:29238 - global_seqlen/minmax_diff:8261 - global_seqlen/balanced_min:25164 - global_seqlen/balanced_max:25170 - global_seqlen/mean:25167.5 - actor/entropy:0.16507954895496368 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2553977966308594 - training/rollout_probs_diff_mean:0.0026576712261885405 - training/rollout_probs_diff_std:0.00907220970839262 - training/rollout_actor_probs_pearson_corr:0.9990447759628296 - rollout_corr/rollout_is_mean:1.000095248222351 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:9.685792974778451e-06 - rollout_corr/rollout_is_max:1.6395798921585083 - rollout_corr/rollout_is_min:0.4166649878025055 - rollout_corr/rollout_is_std:0.0269794724881649 - rollout_corr/rollout_is_eff_sample_size:0.9992728755912055 - rollout_corr/rollout_is_seq_mean:1.0000946521759033 - rollout_corr/rollout_is_seq_std:0.0014211699599400163 - rollout_corr/rollout_is_seq_max:1.0052399635314941 - rollout_corr/rollout_is_seq_min:0.995132327079773 - rollout_corr/rollout_is_seq_max_deviation:0.005239963531494141 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1816609147936106) - rollout_corr/training_log_ppl:np.float64(0.1643792792747263) - rollout_corr/kl:np.float64(0.0004290493978587051) - rollout_corr/k3_kl:np.float64(0.0004259060699780548) - rollout_corr/rollout_ppl:np.float64(1.1811475306749344) - rollout_corr/rollout_log_ppl:np.float64(0.16395022800861625) - rollout_corr/log_ppl_diff:np.float64(0.0004290512661100365) - rollout_corr/log_ppl_abs_diff:np.float64(0.001253184695087839) - rollout_corr/log_ppl_diff_max:np.float64(0.006546080112457275) - rollout_corr/log_ppl_diff_min:np.float64(-0.004463165998458862) - rollout_corr/ppl_ratio:np.float64(1.0004304558970034) - rollout_corr/chi2_token:np.float64(0.0008710860274732113) - rollout_corr/chi2_seq:np.float64(0.33716148510575294) - actor/pg_loss:np.float64(-3.031059168279171e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00016462049688925617) - actor/ppo_kl:np.float64(0.00014458334416866947) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.12229474447667599) - perf/mfu/actor:np.float64(0.10208644290646234) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.20569229125977) - actor/lr:np.float64(1e-06) - training/global_step:14 - training/epoch:0 - critic/score/mean:0.77734375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.77734375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0031842044554650784 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0031842044554650784 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:403.296875 - response_length/max:936.0 - response_length/min:166.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:403.296875 - response_length_non_aborted/max:936.0 - response_length_non_aborted/min:166.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:383.1875 - prompt_length/max:519.0 - prompt_length/min:258.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:2.5698915123939514e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.1408378798514605) - timing_s/agent_loop/generate_sequences/max:np.float64(8.30212122388184) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.7140189427445875) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.30212122388184) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agen
|
||
step:15 - global_seqlen/min:22220 - global_seqlen/max:28895 - global_seqlen/minmax_diff:6675 - global_seqlen/balanced_min:26398 - global_seqlen/balanced_max:26402 - global_seqlen/mean:26400.5 - actor/entropy:0.15334376692771912 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.31862038373947144 - training/rollout_probs_diff_mean:0.002467168727889657 - training/rollout_probs_diff_std:0.008850067853927612 - training/rollout_actor_probs_pearson_corr:0.99895840883255 - rollout_corr/rollout_is_mean:0.999837338924408 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.8703776597976685 - rollout_corr/rollout_is_min:0.5231994986534119 - rollout_corr/rollout_is_std:0.025291837751865387 - rollout_corr/rollout_is_eff_sample_size:0.9993606128205051 - rollout_corr/rollout_is_seq_mean:0.9998326897621155 - rollout_corr/rollout_is_seq_std:0.0014355891617015004 - rollout_corr/rollout_is_seq_max:1.0044710636138916 - rollout_corr/rollout_is_seq_min:0.9946416020393372 - rollout_corr/rollout_is_seq_max_deviation:0.005358397960662842 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1715559563599527) - rollout_corr/training_log_ppl:np.float64(0.15571669759083306) - rollout_corr/kl:np.float64(0.0006178469196100878) - rollout_corr/k3_kl:np.float64(0.00039293118835814766) - rollout_corr/rollout_ppl:np.float64(1.1708267177455127) - rollout_corr/rollout_log_ppl:np.float64(0.1550988501185202) - rollout_corr/log_ppl_diff:np.float64(0.0006178474723128602) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012598117464222014) - rollout_corr/log_ppl_diff_max:np.float64(0.006419658660888672) - rollout_corr/log_ppl_diff_min:np.float64(-0.004078000783920288) - rollout_corr/ppl_ratio:np.float64(1.0006193004082888) - rollout_corr/chi2_token:np.float64(0.00034064752981066704) - rollout_corr/chi2_seq:np.float64(0.33131241356022656) - actor/pg_loss:np.float64(7.49519094824791e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0001743871712278633) - actor/ppo_kl:np.float64(9.903572419744933e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.2359774075448513) - perf/mfu/actor:np.float64(0.10741188729449186) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.21630764007568) - actor/lr:np.float64(1e-06) - training/global_step:15 - 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:0.0016187583096325397 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0016187583096325397 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:435.265625 - response_length/max:741.0 - response_length/min:152.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:435.265625 - response_length_non_aborted/max:741.0 - response_length_non_aborted/min:152.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:389.75 - prompt_length/max:512.0 - prompt_length/min:263.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:4.758872091770172e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.033079355955124) - timing_s/agent_loop/generate_sequences/max:np.float64(7.490394292399287) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.156952525649103) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.490394292399287) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/prompt
|
||
step:16 - global_seqlen/min:23242 - global_seqlen/max:29683 - global_seqlen/minmax_diff:6441 - global_seqlen/balanced_min:26899 - global_seqlen/balanced_max:26903 - global_seqlen/mean:26901.125 - actor/entropy:0.17101728916168213 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.31862038373947144 - training/rollout_probs_diff_mean:0.0025425369385629892 - training/rollout_probs_diff_std:0.008662575855851173 - training/rollout_actor_probs_pearson_corr:0.999067485332489 - rollout_corr/rollout_is_mean:1.0001250505447388 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:8.979804078990128e-06 - rollout_corr/rollout_is_max:1.8446787595748901 - rollout_corr/rollout_is_min:0.27226293087005615 - rollout_corr/rollout_is_std:0.02593412436544895 - rollout_corr/rollout_is_eff_sample_size:0.9993281113448161 - rollout_corr/rollout_is_seq_mean:1.0001810789108276 - rollout_corr/rollout_is_seq_std:0.0015763085102662444 - rollout_corr/rollout_is_seq_max:1.0065312385559082 - rollout_corr/rollout_is_seq_min:0.9940209984779358 - rollout_corr/rollout_is_seq_max_deviation:0.006531238555908203 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.193248980678618) - rollout_corr/training_log_ppl:np.float64(0.17239582387992414) - rollout_corr/kl:np.float64(0.00033440753456126515) - rollout_corr/k3_kl:np.float64(0.00037891795724220856) - rollout_corr/rollout_ppl:np.float64(1.1928353495895863) - rollout_corr/rollout_log_ppl:np.float64(0.17206141452334123) - rollout_corr/log_ppl_diff:np.float64(0.0003344093565829098) - rollout_corr/log_ppl_abs_diff:np.float64(0.001199286969495006) - rollout_corr/log_ppl_diff_max:np.float64(0.009442910552024841) - rollout_corr/log_ppl_diff_min:np.float64(-0.005542337894439697) - rollout_corr/ppl_ratio:np.float64(1.0003358356188983) - rollout_corr/chi2_token:np.float64(0.0008554125670343637) - rollout_corr/chi2_seq:np.float64(0.4001388291362673) - actor/pg_loss:np.float64(-1.7592916265130043e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(8.427388002019143e-05) - actor/ppo_kl:np.float64(0.0001517259071359689) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1951090656220913) - perf/mfu/actor:np.float64(0.10736358975617949) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.19661712646484) - actor/lr:np.float64(1e-06) - training/global_step:16 - training/epoch:0 - critic/score/mean:0.640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.002055252669379115 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.002055252669379115 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:435.00390625 - response_length/max:835.0 - response_length/min:146.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:435.00390625 - response_length_non_aborted/max:835.0 - response_length_non_aborted/min:146.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:405.65625 - prompt_length/max:495.0 - prompt_length/min:268.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:9.904615581035614e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1833369322121143) - timing_s/agent_loop/generate_sequences/max:np.float64(7.8781713005155325) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.695604027634545) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.8781713005155325) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s
|
||
step:17 - global_seqlen/min:22126 - global_seqlen/max:30070 - global_seqlen/minmax_diff:7944 - global_seqlen/balanced_min:25601 - global_seqlen/balanced_max:25610 - global_seqlen/mean:25606.375 - actor/entropy:0.16802534461021423 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4241411089897156 - training/rollout_probs_diff_mean:0.0025332875084131956 - training/rollout_probs_diff_std:0.008655770681798458 - training/rollout_actor_probs_pearson_corr:0.9990943670272827 - rollout_corr/rollout_is_mean:0.9999657869338989 - rollout_corr/rollout_is_ratio_fraction_high:9.704686817713082e-06 - rollout_corr/rollout_is_ratio_fraction_low:9.704686817713082e-06 - rollout_corr/rollout_is_max:2.0695576667785645 - rollout_corr/rollout_is_min:0.48783329129219055 - rollout_corr/rollout_is_std:0.025280050933361053 - rollout_corr/rollout_is_eff_sample_size:0.9993612081053422 - rollout_corr/rollout_is_seq_mean:0.9999219179153442 - rollout_corr/rollout_is_seq_std:0.0015641668578609824 - rollout_corr/rollout_is_seq_max:1.0051617622375488 - rollout_corr/rollout_is_seq_min:0.9921011924743652 - rollout_corr/rollout_is_seq_max_deviation:0.007898807525634766 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.188482933677733) - rollout_corr/training_log_ppl:np.float64(0.16806937995715998) - rollout_corr/kl:np.float64(0.0004929015229606648) - rollout_corr/k3_kl:np.float64(0.00041387354062649706) - rollout_corr/rollout_ppl:np.float64(1.187876682728529) - rollout_corr/rollout_log_ppl:np.float64(0.16757647686608834) - rollout_corr/log_ppl_diff:np.float64(0.0004929030910716392) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013073813388473354) - rollout_corr/log_ppl_diff_max:np.float64(0.008626759052276611) - rollout_corr/log_ppl_diff_min:np.float64(-0.0042053088545799255) - rollout_corr/ppl_ratio:np.float64(1.0004944938700646) - rollout_corr/chi2_token:np.float64(0.0006821004208177328) - rollout_corr/chi2_seq:np.float64(0.3241713694296777) - actor/pg_loss:np.float64(9.176251478493214e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00010633002148097148) - actor/ppo_kl:np.float64(5.967334392664725e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.16828186437487602) - perf/mfu/actor:np.float64(0.10254237405965684) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.18576049804688) - actor/lr:np.float64(1e-06) - training/global_step:17 - training/epoch:0 - critic/score/mean:0.56640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.56640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0036283396184444427 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0036283396184444427 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:402.51171875 - response_length/max:755.0 - response_length/min:113.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:402.51171875 - response_length_non_aborted/max:755.0 - response_length_non_aborted/min:113.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:397.6875 - prompt_length/max:518.0 - prompt_length/min:197.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00010741129517555237 - timing_s/agent_loop/generate_sequences/min:np.float64(1.4098624978214502) - timing_s/agent_loop/generate_sequences/max:np.float64(7.328284703195095) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.6529731439222815) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.328284703195095) - timing_s/agent_loop/slowest/tool_calls:
|
||
step:18 - global_seqlen/min:22716 - global_seqlen/max:27857 - global_seqlen/minmax_diff:5141 - global_seqlen/balanced_min:25547 - global_seqlen/balanced_max:25558 - global_seqlen/mean:25553.75 - actor/entropy:0.15348316729068756 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.41023707389831543 - training/rollout_probs_diff_mean:0.002396379830315709 - training/rollout_probs_diff_std:0.008675065822899342 - training/rollout_actor_probs_pearson_corr:0.9990106225013733 - rollout_corr/rollout_is_mean:1.0001230239868164 - rollout_corr/rollout_is_ratio_fraction_high:9.547268746246118e-06 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.525377035140991 - rollout_corr/rollout_is_min:0.535515308380127 - rollout_corr/rollout_is_std:0.024909522384405136 - rollout_corr/rollout_is_eff_sample_size:0.9993800194714868 - rollout_corr/rollout_is_seq_mean:1.000164270401001 - rollout_corr/rollout_is_seq_std:0.001482548424974084 - rollout_corr/rollout_is_seq_max:1.006403923034668 - rollout_corr/rollout_is_seq_min:0.9943401217460632 - rollout_corr/rollout_is_seq_max_deviation:0.006403923034667969 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1755624981597066) - rollout_corr/training_log_ppl:np.float64(0.1577926207028213) - rollout_corr/kl:np.float64(0.0004000797638387399) - rollout_corr/k3_kl:np.float64(0.00042349859928947353) - rollout_corr/rollout_ppl:np.float64(1.175103010609746) - rollout_corr/rollout_log_ppl:np.float64(0.15739254072104814) - rollout_corr/log_ppl_diff:np.float64(0.0004000799817731604) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012859311391366646) - rollout_corr/log_ppl_diff_max:np.float64(0.006476789712905884) - rollout_corr/log_ppl_diff_min:np.float64(-0.005883336067199707) - rollout_corr/ppl_ratio:np.float64(1.0004015415906906) - rollout_corr/chi2_token:np.float64(0.0009252710733562708) - rollout_corr/chi2_seq:np.float64(0.45614964910782874) - actor/pg_loss:np.float64(9.575032163411379e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0002660133432073053) - actor/ppo_kl:np.float64(0.00022335322293542959) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.18165946006774902) - perf/mfu/actor:np.float64(0.10286695858314113) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.21480178833008) - actor/lr:np.float64(1e-06) - training/global_step:18 - 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:0.004518244881182909 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.004518244881182909 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:409.1484375 - response_length/max:880.0 - response_length/min:170.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:409.1484375 - response_length_non_aborted/max:880.0 - response_length_non_aborted/min:170.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:389.40625 - prompt_length/max:492.0 - prompt_length/min:206.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00012827478349208832 - timing_s/agent_loop/generate_sequences/min:np.float64(2.1651914436370134) - timing_s/agent_loop/generate_sequences/max:np.float64(8.112665966153145) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.75771035838261) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.112665966153145) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent
|
||
step:19 - global_seqlen/min:24113 - global_seqlen/max:31499 - global_seqlen/minmax_diff:7386 - global_seqlen/balanced_min:27522 - global_seqlen/balanced_max:27526 - global_seqlen/mean:27523.875 - actor/entropy:0.18350636959075928 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4400348365306854 - training/rollout_probs_diff_mean:0.0025148489512503147 - training/rollout_probs_diff_std:0.008488977327942848 - training/rollout_actor_probs_pearson_corr:0.9992618560791016 - rollout_corr/rollout_is_mean:0.9999606609344482 - rollout_corr/rollout_is_ratio_fraction_high:8.207013706851285e-06 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.165530204772949 - rollout_corr/rollout_is_min:0.5154091119766235 - rollout_corr/rollout_is_std:0.02545158751308918 - rollout_corr/rollout_is_eff_sample_size:0.9993523979621741 - rollout_corr/rollout_is_seq_mean:0.9999680519104004 - rollout_corr/rollout_is_seq_std:0.0013802233152091503 - rollout_corr/rollout_is_seq_max:1.0054303407669067 - rollout_corr/rollout_is_seq_min:0.9942456483840942 - rollout_corr/rollout_is_seq_max_deviation:0.005754351615905762 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.210047378204763) - rollout_corr/training_log_ppl:np.float64(0.18686755485396134) - rollout_corr/kl:np.float64(0.0003091387440541382) - rollout_corr/k3_kl:np.float64(0.0003994344855300369) - rollout_corr/rollout_ppl:np.float64(1.2096817316487432) - rollout_corr/rollout_log_ppl:np.float64(0.18655841430881992) - rollout_corr/log_ppl_diff:np.float64(0.00030914054514141753) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011480295725050382) - rollout_corr/log_ppl_diff_max:np.float64(0.0067489296197891235) - rollout_corr/log_ppl_diff_min:np.float64(-0.00647541880607605) - rollout_corr/ppl_ratio:np.float64(1.0003103516064584) - rollout_corr/chi2_token:np.float64(0.0010224655270576477) - rollout_corr/chi2_seq:np.float64(0.5422194402199239) - actor/pg_loss:np.float64(-3.9114849641919136e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0001304197849094635) - actor/ppo_kl:np.float64(-7.434647464321653e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.17187580466270447) - perf/mfu/actor:np.float64(0.11026519680088508) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.16053438186646) - actor/lr:np.float64(1e-06) - training/global_step:19 - 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:-0.0009684276301413774 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0009684276301413774 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:475.96484375 - response_length/max:846.0 - response_length/min:173.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:475.96484375 - response_length_non_aborted/max:846.0 - response_length_non_aborted/min:173.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:384.15625 - prompt_length/max:494.0 - prompt_length/min:189.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00015503913164138794 - timing_s/agent_loop/generate_sequences/min:np.float64(2.144832994788885) - timing_s/agent_loop/generate_sequences/max:np.float64(8.308098182082176) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.560128854158393) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.308098182082176) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) -
|
||
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>
|
||
To calculate the volume of a unit cell in a crystal structure, we use the formula for the volume of a unit cell based on the lattice parameters:
|
||
|
||
$$ V = a \times b \times c \times \sqrt{1 + 2\cos\alpha\cos\beta\cos\gamma - \cos^2\alpha - \cos^2\beta - \cos^2\gamma} $$
|
||
|
||
In this case, the crystal structure is given with the following lattice parameters:
|
||
- $ 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 degrees, the crystal structure is orthorhombic, and the formula simplifies to:
|
||
|
||
$$ V = a \times b \times c $$
|
||
|
||
Now, substitute 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 $:
|
||
|
||
$$ 61.1524 \times 17.88 = 1094.31 $$
|
||
|
||
Therefore, the volume of the unit cell is $ 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-GRPO-Qwen-Qwen3-8B-mbs8-train32-rollout8-lr1e-6-vllm0.8/global_step_20
|
||
step:20 - global_seqlen/min:23527 - global_seqlen/max:31800 - global_seqlen/minmax_diff:8273 - global_seqlen/balanced_min:27764 - global_seqlen/balanced_max:27771 - global_seqlen/mean:27768.625 - actor/entropy:0.1863677203655243 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.25539880990982056 - training/rollout_probs_diff_mean:0.002553188009187579 - training/rollout_probs_diff_std:0.008304450660943985 - training/rollout_actor_probs_pearson_corr:0.9992569088935852 - rollout_corr/rollout_is_mean:0.9998255372047424 - rollout_corr/rollout_is_ratio_fraction_high:8.369811439479236e-06 - rollout_corr/rollout_is_ratio_fraction_low:8.369811439479236e-06 - rollout_corr/rollout_is_max:2.0695643424987793 - rollout_corr/rollout_is_min:0.40586209297180176 - rollout_corr/rollout_is_std:0.025391798466444016 - rollout_corr/rollout_is_eff_sample_size:0.9993553743445184 - rollout_corr/rollout_is_seq_mean:0.9998011589050293 - rollout_corr/rollout_is_seq_std:0.0013334499672055244 - rollout_corr/rollout_is_seq_max:1.0043513774871826 - rollout_corr/rollout_is_seq_min:0.9964183568954468 - rollout_corr/rollout_is_seq_max_deviation:0.004351377487182617 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.210513724014163) - rollout_corr/training_log_ppl:np.float64(0.18781593269523) - rollout_corr/kl:np.float64(0.0005289665862315829) - rollout_corr/k3_kl:np.float64(0.0003631379920179256) - rollout_corr/rollout_ppl:np.float64(1.2098644091747701) - rollout_corr/rollout_log_ppl:np.float64(0.18728696474136086) - rollout_corr/log_ppl_diff:np.float64(0.0005289679538691416) - rollout_corr/log_ppl_abs_diff:np.float64(0.001086680407752283) - rollout_corr/log_ppl_diff_max:np.float64(0.005179837346076965) - rollout_corr/log_ppl_diff_min:np.float64(-0.003173530101776123) - rollout_corr/ppl_ratio:np.float64(1.0005300352349877) - rollout_corr/chi2_token:np.float64(0.0004152453038841486) - rollout_corr/chi2_seq:np.float64(0.07480436656624079) - actor/pg_loss:np.float64(2.3769563995301723e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(7.079792703734711e-05) - actor/ppo_kl:np.float64(-1.0977940139156317e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1225764099508524) - perf/mfu/actor:np.float64(0.11097826177378524) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.25468444824219) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.706781914893617) - val-aux/sciknoweval/reward/std@16:np.float64(0.09874159305871691) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7449893617021276) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.0670867911174988) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6692127659574468) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.09798249294138696) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7065957446808511) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.10005103385584223) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7665851063829787) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.04321020147130527) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6289255319148936) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.09109460458607277) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7163404255319149) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.07276144717910131) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7829042553191489) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.03043673029185372) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.5888936170212765) - val-aux/sciknoweval/reward/worst@8/std:np.float64(0.07593238143528759) - val-aux/sciknoweval/reward/maj@8/mean:np.float64(0.7209148936170212) - val-aux/sciknoweval/reward/maj@8/std:np.float64(0.04468436283740652) - val-aux/sciknoweval/reward/best@16/mean:np.float64(0.7954787234042554
|
||
step:21 - global_seqlen/min:24020 - global_seqlen/max:30816 - global_seqlen/minmax_diff:6796 - global_seqlen/balanced_min:27042 - global_seqlen/balanced_max:27050 - global_seqlen/mean:27046.625 - actor/entropy:0.20988589525222778 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23094487190246582 - training/rollout_probs_diff_mean:0.002792060375213623 - training/rollout_probs_diff_std:0.008543950505554676 - training/rollout_actor_probs_pearson_corr:0.9993190765380859 - rollout_corr/rollout_is_mean:0.9999420642852783 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.816615343093872 - rollout_corr/rollout_is_min:0.6647822260856628 - rollout_corr/rollout_is_std:0.026228930801153183 - rollout_corr/rollout_is_eff_sample_size:0.9993122780583822 - rollout_corr/rollout_is_seq_mean:0.9999385476112366 - rollout_corr/rollout_is_seq_std:0.00129312661010772 - rollout_corr/rollout_is_seq_max:1.0037506818771362 - rollout_corr/rollout_is_seq_min:0.9962911605834961 - rollout_corr/rollout_is_seq_max_deviation:0.0037506818771362305 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2395123722963035) - rollout_corr/training_log_ppl:np.float64(0.21087656814052025) - rollout_corr/kl:np.float64(0.00034929181821574673) - rollout_corr/k3_kl:np.float64(0.0003811120066359308) - rollout_corr/rollout_ppl:np.float64(1.239079408813268) - rollout_corr/rollout_log_ppl:np.float64(0.21052727483038325) - rollout_corr/log_ppl_diff:np.float64(0.00034929331013699993) - rollout_corr/log_ppl_abs_diff:np.float64(0.001078697205230128) - rollout_corr/log_ppl_diff_max:np.float64(0.004425495862960815) - rollout_corr/log_ppl_diff_min:np.float64(-0.0028105080127716064) - rollout_corr/ppl_ratio:np.float64(1.0003502799663693) - rollout_corr/chi2_token:np.float64(0.0008452925831079483) - rollout_corr/chi2_seq:np.float64(0.35264981212094426) - actor/pg_loss:np.float64(-6.608013063669205e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0001253350560546096) - actor/ppo_kl:np.float64(-7.613605790623978e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.15066424570977688) - perf/mfu/actor:np.float64(0.10792687021589811) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.82352495193481) - actor/lr:np.float64(1e-06) - training/global_step:21 - training/epoch:0 - critic/score/mean:0.56640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.56640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.004071439150720835 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.004071439150720835 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:451.05078125 - response_length/max:825.0 - response_length/min:157.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:451.05078125 - response_length_non_aborted/max:825.0 - response_length_non_aborted/min:157.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:394.15625 - prompt_length/max:524.0 - prompt_length/min:200.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:5.632452666759491e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.0553892366588116) - timing_s/agent_loop/generate_sequences/max:np.float64(8.023678202182055) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.208998567650269) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.023678202182055) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slo
|
||
step:22 - global_seqlen/min:20457 - global_seqlen/max:31713 - global_seqlen/minmax_diff:11256 - global_seqlen/balanced_min:26522 - global_seqlen/balanced_max:26527 - global_seqlen/mean:26525.25 - actor/entropy:0.1900346577167511 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2413552701473236 - training/rollout_probs_diff_mean:0.00256634084507823 - training/rollout_probs_diff_std:0.00823039747774601 - training/rollout_actor_probs_pearson_corr:0.9992367625236511 - rollout_corr/rollout_is_mean:0.9999353885650635 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:8.67016933625564e-06 - rollout_corr/rollout_is_max:1.6402086019515991 - rollout_corr/rollout_is_min:0.43521374464035034 - rollout_corr/rollout_is_std:0.024964498355984688 - rollout_corr/rollout_is_eff_sample_size:0.9993770429423398 - rollout_corr/rollout_is_seq_mean:0.9999351501464844 - rollout_corr/rollout_is_seq_std:0.0014069859171286225 - rollout_corr/rollout_is_seq_max:1.005880355834961 - rollout_corr/rollout_is_seq_min:0.995682418346405 - rollout_corr/rollout_is_seq_max_deviation:0.0058803558349609375 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.217024851590395) - rollout_corr/training_log_ppl:np.float64(0.19277588855766226) - rollout_corr/kl:np.float64(0.00034876816037332503) - rollout_corr/k3_kl:np.float64(0.00038947746364215163) - rollout_corr/rollout_ppl:np.float64(1.216602017171681) - rollout_corr/rollout_log_ppl:np.float64(0.19242711797414813) - rollout_corr/log_ppl_diff:np.float64(0.00034877058351412416) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010859863250516355) - rollout_corr/log_ppl_diff_max:np.float64(0.0049463361501693726) - rollout_corr/log_ppl_diff_min:np.float64(-0.005746215581893921) - rollout_corr/ppl_ratio:np.float64(1.0003498834557831) - rollout_corr/chi2_token:np.float64(0.0008665753994137049) - rollout_corr/chi2_seq:np.float64(0.2999505987390876) - actor/pg_loss:np.float64(9.697419591248035e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(9.08482979866676e-05) - actor/ppo_kl:np.float64(-6.385789393270613e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.12580287642776966) - perf/mfu/actor:np.float64(0.10640031863300181) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(110.78891372680664) - actor/lr:np.float64(1e-06) - training/global_step:22 - 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:0.0014565885066986084 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0014565885066986084 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:450.5390625 - response_length/max:790.0 - response_length/min:138.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:450.5390625 - response_length_non_aborted/max:790.0 - response_length_non_aborted/min:138.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:378.375 - prompt_length/max:509.0 - prompt_length/min:261.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:9.725242853164673e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.7559421602636576) - timing_s/agent_loop/generate_sequences/max:np.float64(7.862461443990469) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.1768025901401415) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.862461443990469) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agen
|
||
step:23 - global_seqlen/min:18473 - global_seqlen/max:30908 - global_seqlen/minmax_diff:12435 - global_seqlen/balanced_min:25645 - global_seqlen/balanced_max:25648 - global_seqlen/mean:25647.125 - actor/entropy:0.20922939479351044 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2637213468551636 - training/rollout_probs_diff_mean:0.0029920015949755907 - training/rollout_probs_diff_std:0.009136729873716831 - training/rollout_actor_probs_pearson_corr:0.9991100430488586 - rollout_corr/rollout_is_mean:1.000084638595581 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.9838535785675049 - rollout_corr/rollout_is_min:0.5637778639793396 - rollout_corr/rollout_is_std:0.02762783132493496 - rollout_corr/rollout_is_eff_sample_size:0.9992375231521491 - rollout_corr/rollout_is_seq_mean:1.0001695156097412 - rollout_corr/rollout_is_seq_std:0.0014690716052427888 - rollout_corr/rollout_is_seq_max:1.0058791637420654 - rollout_corr/rollout_is_seq_min:0.9964756965637207 - rollout_corr/rollout_is_seq_max_deviation:0.00587916374206543 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2296825596131384) - rollout_corr/training_log_ppl:np.float64(0.20483429163869005) - rollout_corr/kl:np.float64(0.00022616685472609532) - rollout_corr/k3_kl:np.float64(0.0003908713183022883) - rollout_corr/rollout_ppl:np.float64(1.2293974561616778) - rollout_corr/rollout_log_ppl:np.float64(0.20460812396777328) - rollout_corr/log_ppl_diff:np.float64(0.00022616767091676593) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011755102314054966) - rollout_corr/log_ppl_diff_max:np.float64(0.0060980916023254395) - rollout_corr/log_ppl_diff_min:np.float64(-0.0052372366189956665) - rollout_corr/ppl_ratio:np.float64(1.0002273020800203) - rollout_corr/chi2_token:np.float64(0.0011190951336175203) - rollout_corr/chi2_seq:np.float64(0.6697086337953806) - actor/pg_loss:np.float64(-6.491632666438818e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00018655569419934181) - actor/ppo_kl:np.float64(3.843692887528505e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1308762775734067) - perf/mfu/actor:np.float64(0.10322630604294634) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(110.69140243530273) - actor/lr:np.float64(1e-06) - training/global_step:23 - 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:0.002776920795440674 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.002776920795440674 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:417.78515625 - response_length/max:783.0 - response_length/min:179.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:417.78515625 - response_length_non_aborted/max:783.0 - response_length_non_aborted/min:179.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:383.6875 - prompt_length/max:493.0 - prompt_length/min:257.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:9.929202497005463e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.208122979849577) - timing_s/agent_loop/generate_sequences/max:np.float64(7.101036481559277) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.8324967114022) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.101036481559277) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slow
|
||
step:24 - global_seqlen/min:23618 - global_seqlen/max:30976 - global_seqlen/minmax_diff:7358 - global_seqlen/balanced_min:27454 - global_seqlen/balanced_max:27459 - global_seqlen/mean:27456.875 - actor/entropy:0.16937562823295593 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5486339330673218 - training/rollout_probs_diff_mean:0.0023563068825751543 - training/rollout_probs_diff_std:0.008330320939421654 - training/rollout_actor_probs_pearson_corr:0.9992015957832336 - rollout_corr/rollout_is_mean:1.000078558921814 - rollout_corr/rollout_is_ratio_fraction_high:8.243275260610972e-06 - rollout_corr/rollout_is_ratio_fraction_low:8.243275260610972e-06 - rollout_corr/rollout_is_max:3.039975881576538 - rollout_corr/rollout_is_min:0.0006037261919118464 - rollout_corr/rollout_is_std:0.02445065602660179 - rollout_corr/rollout_is_eff_sample_size:0.9994027607393757 - rollout_corr/rollout_is_seq_mean:1.0001696348190308 - rollout_corr/rollout_is_seq_std:0.0013404276687651873 - rollout_corr/rollout_is_seq_max:1.0070313215255737 - rollout_corr/rollout_is_seq_min:0.997212290763855 - rollout_corr/rollout_is_seq_max_deviation:0.0070313215255737305 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1992141208611429) - rollout_corr/training_log_ppl:np.float64(0.17708433591906214) - rollout_corr/kl:np.float64(0.00021467350635373528) - rollout_corr/k3_kl:np.float64(0.00040859091361511446) - rollout_corr/rollout_ppl:np.float64(1.198959435801953) - rollout_corr/rollout_log_ppl:np.float64(0.17686966038309038) - rollout_corr/log_ppl_diff:np.float64(0.00021467553597176448) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010494204252609052) - rollout_corr/log_ppl_diff_max:np.float64(0.016319751739501953) - rollout_corr/log_ppl_diff_min:np.float64(-0.003981977701187134) - rollout_corr/ppl_ratio:np.float64(1.0002161071170121) - rollout_corr/chi2_token:np.float64(0.0011550325434654951) - rollout_corr/chi2_seq:np.float64(0.5893677773419768) - actor/pg_loss:np.float64(8.695642463862896e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(7.039850561341154e-05) - actor/ppo_kl:np.float64(5.909442530871445e-07) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.14581617712974548) - perf/mfu/actor:np.float64(0.10952445859984875) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(110.68953323364258) - actor/lr:np.float64(1e-06) - training/global_step:24 - training/epoch:0 - 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:-0.0006563708302564919 - critic/advantages/max:0.5 - critic/advantages/min:-0.75 - critic/returns/mean:-0.0006563708302564919 - critic/returns/max:0.5 - critic/returns/min:-0.75 - response_length/mean:473.87109375 - response_length/max:818.0 - response_length/min:176.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:473.87109375 - response_length_non_aborted/max:818.0 - response_length_non_aborted/min:176.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:384.15625 - prompt_length/max:477.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:5.136430263519287e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.1059566028416157) - timing_s/agent_loop/generate_sequences/max:np.float64(7.97863014601171) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.456470786826685) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.97863014601171) - timing_s/agent_loop/slowest/tool_calls:np.float64(
|
||
step:25 - global_seqlen/min:22266 - global_seqlen/max:31914 - global_seqlen/minmax_diff:9648 - global_seqlen/balanced_min:26848 - global_seqlen/balanced_max:26857 - global_seqlen/mean:26854.5 - actor/entropy:0.2005494087934494 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.21512538194656372 - training/rollout_probs_diff_mean:0.002576504833996296 - training/rollout_probs_diff_std:0.008091717958450317 - training/rollout_actor_probs_pearson_corr:0.9993591904640198 - rollout_corr/rollout_is_mean:1.0000380277633667 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6483047008514404 - rollout_corr/rollout_is_min:0.5925564169883728 - rollout_corr/rollout_is_std:0.02495972253382206 - rollout_corr/rollout_is_eff_sample_size:0.9993775191858117 - rollout_corr/rollout_is_seq_mean:1.000016212463379 - rollout_corr/rollout_is_seq_std:0.0014132416108623147 - rollout_corr/rollout_is_seq_max:1.0057713985443115 - rollout_corr/rollout_is_seq_min:0.9941661953926086 - rollout_corr/rollout_is_seq_max_deviation:0.005833804607391357 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.247515902388841) - rollout_corr/training_log_ppl:np.float64(0.21253347316815052) - rollout_corr/kl:np.float64(0.0003586823201153777) - rollout_corr/k3_kl:np.float64(0.0004320936768209549) - rollout_corr/rollout_ppl:np.float64(1.2470650793984532) - rollout_corr/rollout_log_ppl:np.float64(0.2121747907367535) - rollout_corr/log_ppl_diff:np.float64(0.00035868243139702827) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011752601276384667) - rollout_corr/log_ppl_diff_max:np.float64(0.005879908800125122) - rollout_corr/log_ppl_diff_min:np.float64(-0.00401286780834198) - rollout_corr/ppl_ratio:np.float64(1.000359978992492) - rollout_corr/chi2_token:np.float64(0.001059497706592083) - rollout_corr/chi2_seq:np.float64(0.5972242541611195) - actor/pg_loss:np.float64(2.3624743334949017e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00012215516676405969) - actor/ppo_kl:np.float64(2.806563131052542e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1335355006158352) - perf/mfu/actor:np.float64(0.10680968031632504) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(110.6960678100586) - actor/lr:np.float64(1e-06) - training/global_step:25 - training/epoch:0 - 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:0.005718811880797148 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.005718811880797148 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:448.765625 - response_length/max:1180.0 - response_length/min:165.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:448.765625 - response_length_non_aborted/max:1180.0 - response_length_non_aborted/min:165.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:390.4375 - prompt_length/max:481.0 - prompt_length/min:196.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00014773011207580566 - timing_s/agent_loop/generate_sequences/min:np.float64(2.004378417506814) - timing_s/agent_loop/generate_sequences/max:np.float64(9.795474849641323) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.088747393274389) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(9.795474849641323) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/prompt_
|
||
step:26 - global_seqlen/min:25781 - global_seqlen/max:31695 - global_seqlen/minmax_diff:5914 - global_seqlen/balanced_min:28514 - global_seqlen/balanced_max:28518 - global_seqlen/mean:28515.875 - actor/entropy:0.20827430486679077 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2769121527671814 - training/rollout_probs_diff_mean:0.0026053066831082106 - training/rollout_probs_diff_std:0.00794726237654686 - training/rollout_actor_probs_pearson_corr:0.9993575811386108 - rollout_corr/rollout_is_mean:1.0002012252807617 - rollout_corr/rollout_is_ratio_fraction_high:7.904950507509056e-06 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.244551658630371 - rollout_corr/rollout_is_min:0.5874804258346558 - rollout_corr/rollout_is_std:0.024796804413199425 - rollout_corr/rollout_is_eff_sample_size:0.9993857344571517 - rollout_corr/rollout_is_seq_mean:1.0002319812774658 - rollout_corr/rollout_is_seq_std:0.0011953577632084489 - rollout_corr/rollout_is_seq_max:1.0041698217391968 - rollout_corr/rollout_is_seq_min:0.9968327283859253 - rollout_corr/rollout_is_seq_max_deviation:0.004169821739196777 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2440492357127368) - rollout_corr/training_log_ppl:np.float64(0.2152066015114542) - rollout_corr/kl:np.float64(0.00024418294762629245) - rollout_corr/k3_kl:np.float64(0.0003427424277333557) - rollout_corr/rollout_ppl:np.float64(1.2437414275482297) - rollout_corr/rollout_log_ppl:np.float64(0.2149624170269817) - rollout_corr/log_ppl_diff:np.float64(0.0002441844844724983) - rollout_corr/log_ppl_abs_diff:np.float64(0.000932207825826481) - rollout_corr/log_ppl_diff_max:np.float64(0.004497110843658447) - rollout_corr/log_ppl_diff_min:np.float64(-0.0034370869398117065) - rollout_corr/ppl_ratio:np.float64(1.0002449133899063) - rollout_corr/chi2_token:np.float64(0.0009027821943163872) - rollout_corr/chi2_seq:np.float64(0.36676088720560074) - actor/pg_loss:np.float64(3.524916246533394e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(4.288250374884228e-05) - actor/ppo_kl:np.float64(0.00015204964313486613) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.09116398356854916) - perf/mfu/actor:np.float64(0.113877495350513) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(110.65935134887695) - actor/lr:np.float64(1e-06) - training/global_step:26 - training/epoch:0 - 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:0.000627455476205796 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.000627455476205796 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:494.15234375 - response_length/max:880.0 - response_length/min:162.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:494.15234375 - response_length_non_aborted/max:880.0 - response_length_non_aborted/min:162.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:396.96875 - prompt_length/max:491.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00016456283628940582 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8963664006441832) - timing_s/agent_loop/generate_sequences/max:np.float64(8.646993298083544) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.699355000702781) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.646993298083544) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/ag
|
||
step:27 - global_seqlen/min:24354 - global_seqlen/max:30450 - global_seqlen/minmax_diff:6096 - global_seqlen/balanced_min:27918 - global_seqlen/balanced_max:27922 - global_seqlen/mean:27920.625 - actor/entropy:0.2055797427892685 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7309482097625732 - training/rollout_probs_diff_mean:0.0024637985043227673 - training/rollout_probs_diff_std:0.00815705768764019 - training/rollout_actor_probs_pearson_corr:0.9992928504943848 - rollout_corr/rollout_is_mean:1.0000125169754028 - rollout_corr/rollout_is_ratio_fraction_high:8.22930087451823e-06 - rollout_corr/rollout_is_ratio_fraction_low:1.645860174903646e-05 - rollout_corr/rollout_is_max:2.0586676597595215 - rollout_corr/rollout_is_min:0.0001314591645495966 - rollout_corr/rollout_is_std:0.02405497618019581 - rollout_corr/rollout_is_eff_sample_size:0.9994216927413265 - rollout_corr/rollout_is_seq_mean:1.0000035762786865 - rollout_corr/rollout_is_seq_std:0.0012129470705986023 - rollout_corr/rollout_is_seq_max:1.003065824508667 - rollout_corr/rollout_is_seq_min:0.9962401390075684 - rollout_corr/rollout_is_seq_max_deviation:0.0037598609924316406 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2440797071903944) - rollout_corr/training_log_ppl:np.float64(0.2131779525516322) - rollout_corr/kl:np.float64(0.0003574901133275299) - rollout_corr/k3_kl:np.float64(0.00035757935391700357) - rollout_corr/rollout_ppl:np.float64(1.2436211220920086) - rollout_corr/rollout_log_ppl:np.float64(0.21282046237320174) - rollout_corr/log_ppl_diff:np.float64(0.00035749017843045294) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010061207285616547) - rollout_corr/log_ppl_diff_max:np.float64(0.00536857545375824) - rollout_corr/log_ppl_diff_min:np.float64(-0.0038084089756011963) - rollout_corr/ppl_ratio:np.float64(1.0003584027290344) - rollout_corr/chi2_token:np.float64(0.0006836575921624899) - rollout_corr/chi2_seq:np.float64(0.28105390653945506) - actor/pg_loss:np.float64(-0.00014648318756371737) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(6.341250309560564e-05) - actor/ppo_kl:np.float64(-8.465370333166078e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.15089168027043343) - perf/mfu/actor:np.float64(0.11177476135023583) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.46063613891602) - actor/lr:np.float64(1e-06) - training/global_step:27 - training/epoch:1 - 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:0.0004742134769912809 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0004742134769912809 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:474.67578125 - response_length/max:811.0 - response_length/min:172.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:474.67578125 - response_length_non_aborted/max:811.0 - response_length_non_aborted/min:172.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:397.84375 - prompt_length/max:494.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00047997571527957916 - timing_s/agent_loop/generate_sequences/min:np.float64(2.266698367893696) - timing_s/agent_loop/generate_sequences/max:np.float64(7.956064598634839) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.491192605360993) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.956064598634839) - timing_s/agent_loop/slowest/tool_calls
|
||
step:28 - global_seqlen/min:20775 - global_seqlen/max:31390 - global_seqlen/minmax_diff:10615 - global_seqlen/balanced_min:26861 - global_seqlen/balanced_max:26867 - global_seqlen/mean:26864.875 - actor/entropy:0.24384665489196777 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5723066329956055 - training/rollout_probs_diff_mean:0.0027470295317471027 - training/rollout_probs_diff_std:0.008356913924217224 - training/rollout_actor_probs_pearson_corr:0.9994053840637207 - rollout_corr/rollout_is_mean:0.9999920129776001 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.5305564122390933e-05 - rollout_corr/rollout_is_max:1.5868679285049438 - rollout_corr/rollout_is_min:0.24166789650917053 - rollout_corr/rollout_is_std:0.025605684146285057 - rollout_corr/rollout_is_eff_sample_size:0.9993447785041603 - rollout_corr/rollout_is_seq_mean:0.9999786615371704 - rollout_corr/rollout_is_seq_std:0.0014453264884650707 - rollout_corr/rollout_is_seq_max:1.005670189857483 - rollout_corr/rollout_is_seq_min:0.9954844117164612 - rollout_corr/rollout_is_seq_max_deviation:0.00567018985748291 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2841918719932437) - rollout_corr/training_log_ppl:np.float64(0.24157115184061695) - rollout_corr/kl:np.float64(0.0003922118894976734) - rollout_corr/k3_kl:np.float64(0.0004578474525800402) - rollout_corr/rollout_ppl:np.float64(1.2836904847063124) - rollout_corr/rollout_log_ppl:np.float64(0.2411789383040741) - rollout_corr/log_ppl_diff:np.float64(0.0003922135365428403) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012401825661072508) - rollout_corr/log_ppl_diff_max:np.float64(0.00700107216835022) - rollout_corr/log_ppl_diff_min:np.float64(-0.005011111497879028) - rollout_corr/ppl_ratio:np.float64(1.000393595546484) - rollout_corr/chi2_token:np.float64(0.0010894446168094873) - rollout_corr/chi2_seq:np.float64(0.5639708044473082) - actor/pg_loss:np.float64(5.167769268155098e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(6.550479020006605e-05) - actor/ppo_kl:np.float64(1.7489767522249622e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.116465725004673) - perf/mfu/actor:np.float64(0.10695600784376426) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.37525177001953) - actor/lr:np.float64(1e-06) - training/global_step:28 - 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:-0.00028047000523656607 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00028047000523656607 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:463.08984375 - response_length/max:837.0 - response_length/min:169.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:463.08984375 - response_length_non_aborted/max:837.0 - response_length_non_aborted/min:169.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:376.4375 - prompt_length/max:478.0 - prompt_length/min:190.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.0001988764852285385 - timing_s/agent_loop/generate_sequences/min:np.float64(2.185948533937335) - timing_s/agent_loop/generate_sequences/max:np.float64(8.100449938327074) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.3857879123679595) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.100449938327074) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timin
|
||
step:29 - global_seqlen/min:22297 - global_seqlen/max:31545 - global_seqlen/minmax_diff:9248 - global_seqlen/balanced_min:28290 - global_seqlen/balanced_max:28300 - global_seqlen/mean:28296.125 - actor/entropy:0.1927935630083084 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.28604796528816223 - training/rollout_probs_diff_mean:0.0021387599408626556 - training/rollout_probs_diff_std:0.007081093732267618 - training/rollout_actor_probs_pearson_corr:0.9994828701019287 - rollout_corr/rollout_is_mean:0.999884843826294 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.611747145652771 - rollout_corr/rollout_is_min:0.5320075750350952 - rollout_corr/rollout_is_std:0.02203766442835331 - rollout_corr/rollout_is_eff_sample_size:0.9995145771041962 - rollout_corr/rollout_is_seq_mean:0.9998981952667236 - rollout_corr/rollout_is_seq_std:0.00132310192566365 - rollout_corr/rollout_is_seq_max:1.0052039623260498 - rollout_corr/rollout_is_seq_min:0.9921262264251709 - rollout_corr/rollout_is_seq_max_deviation:0.007873773574829102 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2352579105645418) - rollout_corr/training_log_ppl:np.float64(0.20490435884858016) - rollout_corr/kl:np.float64(0.0003952607846713363) - rollout_corr/k3_kl:np.float64(0.0003191322348712333) - rollout_corr/rollout_ppl:np.float64(1.2347394162788987) - rollout_corr/rollout_log_ppl:np.float64(0.20450909563805908) - rollout_corr/log_ppl_diff:np.float64(0.0003952632105210796) - rollout_corr/log_ppl_abs_diff:np.float64(0.001101043846574612) - rollout_corr/log_ppl_diff_max:np.float64(0.008444517850875854) - rollout_corr/log_ppl_diff_min:np.float64(-0.004731416702270508) - rollout_corr/ppl_ratio:np.float64(1.0003964479546994) - rollout_corr/chi2_token:np.float64(0.0005052480846643448) - rollout_corr/chi2_seq:np.float64(0.20438435021787882) - actor/pg_loss:np.float64(5.3944881074130535e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00011199153595953248) - actor/ppo_kl:np.float64(2.4016177389896143e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.11256356723606586) - perf/mfu/actor:np.float64(0.11208081941034986) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.4299087524414) - actor/lr:np.float64(1e-06) - training/global_step:29 - training/epoch:1 - 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:-0.002028820803388953 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.002028820803388953 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:479.66015625 - response_length/max:881.0 - response_length/min:177.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:479.66015625 - response_length_non_aborted/max:881.0 - response_length_non_aborted/min:177.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:404.59375 - prompt_length/max:512.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.0001045428216457367 - timing_s/agent_loop/generate_sequences/min:np.float64(2.2093354631215334) - timing_s/agent_loop/generate_sequences/max:np.float64(8.23103902488947) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.504767282072862) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.23103902488947) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/prom
|
||
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>
|
||
To calculate the volume of the unit cell for a crystal structure, we use the formula for the volume of a unit cell, which depends on the lattice parameters and the angles between them:
|
||
|
||
$$ V = a \cdot b \cdot c \cdot \sqrt{1 + 2\cos\alpha\cos\beta\cos\gamma - \cos^2\alpha - \cos^2\beta - \cos^2\gamma} $$
|
||
|
||
Given:
|
||
- $ a = 7.82 $ Å
|
||
- $ b = 7.82 $ Å
|
||
- $ c = 17.88 $ Å
|
||
- $ \alpha = 90^\circ $
|
||
- $ \beta = 90^\circ $
|
||
- $ \gamma = 90^\circ $
|
||
|
||
Since all angles are 90°, the crystal structure is orthorhombic. In this case, the formula simplifies to:
|
||
|
||
$$ V = a \cdot b \cdot c $$
|
||
|
||
Now, substitute the given values:
|
||
|
||
$$ V = 7.82 \cdot 7.82 \cdot 17.88 $$
|
||
|
||
First, calculate $ 7.82 \cdot 7.82 $:
|
||
|
||
$$ 7.82 \cdot 7.82 = 61.1524 $$
|
||
|
||
Now, multiply this by 17.88:
|
||
|
||
$$ 61.1524 \cdot 17.88 = 1094.31 $$
|
||
|
||
So, 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-GRPO-Qwen-Qwen3-8B-mbs8-train32-rollout8-lr1e-6-vllm0.8/global_step_30
|
||
step:30 - global_seqlen/min:23520 - global_seqlen/max:31889 - global_seqlen/minmax_diff:8369 - global_seqlen/balanced_min:29105 - global_seqlen/balanced_max:29106 - global_seqlen/mean:29105.25 - actor/entropy:0.22476468980312347 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.25539630651474 - training/rollout_probs_diff_mean:0.0022680300753563643 - training/rollout_probs_diff_std:0.0070181735791265965 - training/rollout_actor_probs_pearson_corr:0.9995337128639221 - rollout_corr/rollout_is_mean:1.000045895576477 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.7930431365966797 - rollout_corr/rollout_is_min:0.5083495378494263 - rollout_corr/rollout_is_std:0.02269851416349411 - rollout_corr/rollout_is_eff_sample_size:0.9994850427681393 - rollout_corr/rollout_is_seq_mean:1.0000736713409424 - rollout_corr/rollout_is_seq_std:0.001143696834333241 - rollout_corr/rollout_is_seq_max:1.0036154985427856 - rollout_corr/rollout_is_seq_min:0.9971887469291687 - rollout_corr/rollout_is_seq_max_deviation:0.0036154985427856445 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2817749911919236) - rollout_corr/training_log_ppl:np.float64(0.23585528284456814) - rollout_corr/kl:np.float64(0.00021687363476452504) - rollout_corr/k3_kl:np.float64(0.00031874458314717913) - rollout_corr/rollout_ppl:np.float64(1.2814569370821118) - rollout_corr/rollout_log_ppl:np.float64(0.2356384065351449) - rollout_corr/log_ppl_diff:np.float64(0.00021687630942324176) - rollout_corr/log_ppl_abs_diff:np.float64(0.0009794712605071254) - rollout_corr/log_ppl_diff_max:np.float64(0.005163669586181641) - rollout_corr/log_ppl_diff_min:np.float64(-0.0038784146308898926) - rollout_corr/ppl_ratio:np.float64(1.000217784428969) - rollout_corr/chi2_token:np.float64(0.0008400704246014357) - rollout_corr/chi2_seq:np.float64(0.4963743803091347) - actor/pg_loss:np.float64(-7.453944999724627e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(7.395594548142981e-05) - actor/ppo_kl:np.float64(2.2799680227336694e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.15676945261657238) - perf/mfu/actor:np.float64(0.11588638795918937) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.46013641357422) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7027925531914894) - val-aux/sciknoweval/reward/std@16:np.float64(0.1152265165634801) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7485106382978723) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.0846900615940614) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6569148936170213) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.10865230852578993) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7026063829787232) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.11444102586889436) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7797127659574468) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.056704653085655574) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.614436170212766) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.0941326641213177) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7107234042553192) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.08665034162618929) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.8018191489361703) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.035254136893894374) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.5756382978723404) - val-aux/sciknoweval/reward/worst@8/std:np.float64(0.07668049600750805) - val-aux/sciknoweval/reward/maj@8/mean:np.float64(0.7147765957446808) - val-aux/sciknoweval/reward/maj@8/std:np.float64(0.0579906950963037) - val-aux/sciknoweval/reward/best@16/mean:np.float64(0.8148297872340425) - val-aux/sciknoweval/reward/best
|
||
step:31 - global_seqlen/min:25888 - global_seqlen/max:30450 - global_seqlen/minmax_diff:4562 - global_seqlen/balanced_min:27424 - global_seqlen/balanced_max:27426 - global_seqlen/mean:27425.375 - actor/entropy:0.2341211438179016 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23105886578559875 - training/rollout_probs_diff_mean:0.0023879760410636663 - training/rollout_probs_diff_std:0.007143564056605101 - training/rollout_actor_probs_pearson_corr:0.9994761347770691 - rollout_corr/rollout_is_mean:0.9998959898948669 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:8.458877346129157e-06 - rollout_corr/rollout_is_max:1.7586430311203003 - rollout_corr/rollout_is_min:0.44323089718818665 - rollout_corr/rollout_is_std:0.023076131939888 - rollout_corr/rollout_is_eff_sample_size:0.9994676564348994 - rollout_corr/rollout_is_seq_mean:0.999855101108551 - rollout_corr/rollout_is_seq_std:0.0011561827268451452 - rollout_corr/rollout_is_seq_max:1.0038914680480957 - rollout_corr/rollout_is_seq_min:0.9953445792198181 - rollout_corr/rollout_is_seq_max_deviation:0.004655420780181885 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2752317558042705) - rollout_corr/training_log_ppl:np.float64(0.23740968311176402) - rollout_corr/kl:np.float64(0.00037503032777941314) - rollout_corr/k3_kl:np.float64(0.00029784045699443595) - rollout_corr/rollout_ppl:np.float64(1.2747451555915177) - rollout_corr/rollout_log_ppl:np.float64(0.23703464957361575) - rollout_corr/log_ppl_diff:np.float64(0.0003750335381482728) - rollout_corr/log_ppl_abs_diff:np.float64(0.0009644899037084542) - rollout_corr/log_ppl_diff_max:np.float64(0.0038155168294906616) - rollout_corr/log_ppl_diff_min:np.float64(-0.004125386476516724) - rollout_corr/ppl_ratio:np.float64(1.0003758335951716) - rollout_corr/chi2_token:np.float64(0.00044544809497892857) - rollout_corr/chi2_seq:np.float64(0.17121168388985097) - actor/pg_loss:np.float64(0.0001104114344343543) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00012691650317719905) - actor/ppo_kl:np.float64(-5.702109045202519e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.14219314232468605) - perf/mfu/actor:np.float64(0.10953375845039864) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(112.64505767822266) - actor/lr:np.float64(1e-06) - training/global_step:31 - training/epoch:1 - 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:0.008942090906202793 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.008942090906202793 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:461.79296875 - response_length/max:839.0 - response_length/min:173.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:461.79296875 - response_length_non_aborted/max:839.0 - response_length_non_aborted/min:173.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:395.25 - prompt_length/max:492.0 - prompt_length/min:266.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:9.103678166866302e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.0917833521962166) - timing_s/agent_loop/generate_sequences/max:np.float64(8.422423332929611) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.35807181410928) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.422423332929611) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_
|
||
step:32 - global_seqlen/min:21612 - global_seqlen/max:32901 - global_seqlen/minmax_diff:11289 - global_seqlen/balanced_min:27666 - global_seqlen/balanced_max:27672 - global_seqlen/mean:27669.5 - actor/entropy:0.2351425141096115 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.893662691116333 - training/rollout_probs_diff_mean:0.002326859161257744 - training/rollout_probs_diff_std:0.007395750377327204 - training/rollout_actor_probs_pearson_corr:0.9995218515396118 - rollout_corr/rollout_is_mean:0.9999954104423523 - rollout_corr/rollout_is_ratio_fraction_high:8.340283784491476e-06 - rollout_corr/rollout_is_ratio_fraction_low:8.340283784491476e-06 - rollout_corr/rollout_is_max:2.5332539081573486 - rollout_corr/rollout_is_min:0.012144971638917923 - rollout_corr/rollout_is_std:0.022933630272746086 - rollout_corr/rollout_is_eff_sample_size:0.9994742060099695 - rollout_corr/rollout_is_seq_mean:1.0000008344650269 - rollout_corr/rollout_is_seq_std:0.0012937926221638918 - rollout_corr/rollout_is_seq_max:1.0046977996826172 - rollout_corr/rollout_is_seq_min:0.9950659275054932 - rollout_corr/rollout_is_seq_max_deviation:0.004934072494506836 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2944361343979836) - rollout_corr/training_log_ppl:np.float64(0.24626350629114313) - rollout_corr/kl:np.float64(0.0002554487013840534) - rollout_corr/k3_kl:np.float64(0.0003589305630811168) - rollout_corr/rollout_ppl:np.float64(1.2940973713994026) - rollout_corr/rollout_log_ppl:np.float64(0.24600805527734337) - rollout_corr/log_ppl_diff:np.float64(0.00025545101379975677) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010613546473905444) - rollout_corr/log_ppl_diff_max:np.float64(0.01161763072013855) - rollout_corr/log_ppl_diff_min:np.float64(-0.0033280551433563232) - rollout_corr/ppl_ratio:np.float64(1.000256673200056) - rollout_corr/chi2_token:np.float64(0.0008953770156949759) - rollout_corr/chi2_seq:np.float64(0.5134820169769228) - actor/pg_loss:np.float64(6.652786396443844e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00010245629937344347) - actor/ppo_kl:np.float64(-5.820105452869484e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.16398444771766663) - perf/mfu/actor:np.float64(0.11033717682891869) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.75769805908203) - actor/lr:np.float64(1e-06) - training/global_step:32 - training/epoch:1 - 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:0.0005869474844075739 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.0005869474844075739 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:468.359375 - response_length/max:848.0 - response_length/min:160.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:468.359375 - response_length_non_aborted/max:848.0 - response_length_non_aborted/min:160.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:396.3125 - prompt_length/max:518.0 - prompt_length/min:197.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.0001023951917886734 - timing_s/agent_loop/generate_sequences/min:np.float64(1.942232746630907) - timing_s/agent_loop/generate_sequences/max:np.float64(8.150801707059145) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.316388803075824) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.150801707059145) - timing_s/agent_loop/slowest/tool_calls:np.float64(
|
||
step:33 - global_seqlen/min:20947 - global_seqlen/max:30495 - global_seqlen/minmax_diff:9548 - global_seqlen/balanced_min:27719 - global_seqlen/balanced_max:27728 - global_seqlen/mean:27724.125 - actor/entropy:0.25870415568351746 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.8123438954353333 - training/rollout_probs_diff_mean:0.0024514386896044016 - training/rollout_probs_diff_std:0.007493364159017801 - training/rollout_actor_probs_pearson_corr:0.9996020793914795 - rollout_corr/rollout_is_mean:0.9999789595603943 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.6236797819146886e-05 - rollout_corr/rollout_is_max:1.5526777505874634 - rollout_corr/rollout_is_min:0.0813368558883667 - rollout_corr/rollout_is_std:0.02380339801311493 - rollout_corr/rollout_is_eff_sample_size:0.9994336000282128 - rollout_corr/rollout_is_seq_mean:1.000024437904358 - rollout_corr/rollout_is_seq_std:0.001137533108703792 - rollout_corr/rollout_is_seq_max:1.0037413835525513 - rollout_corr/rollout_is_seq_min:0.9970559477806091 - rollout_corr/rollout_is_seq_max_deviation:0.0037413835525512695 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.318345540203154) - rollout_corr/training_log_ppl:np.float64(0.26373951183632016) - rollout_corr/kl:np.float64(0.0002792183334006104) - rollout_corr/k3_kl:np.float64(0.0003289724186430476) - rollout_corr/rollout_ppl:np.float64(1.3179421676322818) - rollout_corr/rollout_log_ppl:np.float64(0.2634602915350115) - rollout_corr/log_ppl_diff:np.float64(0.00027922030130866915) - rollout_corr/log_ppl_abs_diff:np.float64(0.0009092716936720535) - rollout_corr/log_ppl_diff_max:np.float64(0.009413182735443115) - rollout_corr/log_ppl_diff_min:np.float64(-0.003118470311164856) - rollout_corr/ppl_ratio:np.float64(1.0002800580114126) - rollout_corr/chi2_token:np.float64(0.0007109332364052534) - rollout_corr/chi2_seq:np.float64(0.21768948528915644) - actor/pg_loss:np.float64(-2.1165120415389538e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(4.417932996148011e-05) - actor/ppo_kl:np.float64(-4.76137685723188e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1778383795171976) - perf/mfu/actor:np.float64(0.11100855943794985) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.48153686523438) - actor/lr:np.float64(1e-06) - training/global_step:33 - training/epoch:1 - critic/score/mean:0.76171875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.76171875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.00239492766559124 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00239492766559124 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:481.16015625 - response_length/max:929.0 - response_length/min:168.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:481.16015625 - response_length_non_aborted/max:929.0 - response_length_non_aborted/min:168.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:385.21875 - prompt_length/max:480.0 - prompt_length/min:197.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:5.61252236366272e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.1162655036896467) - timing_s/agent_loop/generate_sequences/max:np.float64(8.431221760809422) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.50012135515135) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.431221760809422) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_
|
||
step:34 - global_seqlen/min:25779 - global_seqlen/max:32632 - global_seqlen/minmax_diff:6853 - global_seqlen/balanced_min:28623 - global_seqlen/balanced_max:28629 - global_seqlen/mean:28625.75 - actor/entropy:0.24488314986228943 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7310589551925659 - training/rollout_probs_diff_mean:0.00249440036714077 - training/rollout_probs_diff_std:0.0074654994532465935 - training/rollout_actor_probs_pearson_corr:0.9995710253715515 - rollout_corr/rollout_is_mean:1.000001072883606 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.6263600628008135e-05 - rollout_corr/rollout_is_max:1.951600193977356 - rollout_corr/rollout_is_min:0.26894107460975647 - rollout_corr/rollout_is_std:0.023488307371735573 - rollout_corr/rollout_is_eff_sample_size:0.9994484845358234 - rollout_corr/rollout_is_seq_mean:1.0000451803207397 - rollout_corr/rollout_is_seq_std:0.0012103441404178739 - rollout_corr/rollout_is_seq_max:1.0049504041671753 - rollout_corr/rollout_is_seq_min:0.9962252974510193 - rollout_corr/rollout_is_seq_max_deviation:0.004950404167175293 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3030735217034817) - rollout_corr/training_log_ppl:np.float64(0.2590878018454532) - rollout_corr/kl:np.float64(0.00033822323003818155) - rollout_corr/k3_kl:np.float64(0.00033817789738321835) - rollout_corr/rollout_ppl:np.float64(1.3026120658032596) - rollout_corr/rollout_log_ppl:np.float64(0.2587495753541589) - rollout_corr/log_ppl_diff:np.float64(0.0003382264912943356) - rollout_corr/log_ppl_abs_diff:np.float64(0.0009682540330686606) - rollout_corr/log_ppl_diff_max:np.float64(0.011813685297966003) - rollout_corr/log_ppl_diff_min:np.float64(-0.004161953926086426) - rollout_corr/ppl_ratio:np.float64(1.000339277787134) - rollout_corr/chi2_token:np.float64(0.0006227635312825441) - rollout_corr/chi2_seq:np.float64(0.22475781477987766) - actor/pg_loss:np.float64(0.00017607829067856073) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(3.003569008797058e-05) - actor/ppo_kl:np.float64(8.233531357149104e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.11507174372673035) - perf/mfu/actor:np.float64(0.1140994907811674) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.46148681640625) - actor/lr:np.float64(1e-06) - training/global_step:34 - training/epoch:1 - critic/score/mean:0.79296875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.79296875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0008680696482770145 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0008680696482770145 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:480.3671875 - response_length/max:934.0 - response_length/min:168.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:480.3671875 - response_length_non_aborted/max:934.0 - response_length_non_aborted/min:168.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:414.1875 - prompt_length/max:482.0 - prompt_length/min:271.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:9.508803486824036e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.029529247432947) - timing_s/agent_loop/generate_sequences/max:np.float64(8.856205441057682) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.579186692339135) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.856205441057682) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_
|
||
step:35 - global_seqlen/min:24227 - global_seqlen/max:29909 - global_seqlen/minmax_diff:5682 - global_seqlen/balanced_min:25979 - global_seqlen/balanced_max:25983 - global_seqlen/mean:25981.75 - actor/entropy:0.2596481442451477 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2622659206390381 - training/rollout_probs_diff_mean:0.0026135449297726154 - training/rollout_probs_diff_std:0.0073624346405267715 - training/rollout_actor_probs_pearson_corr:0.999534547328949 - rollout_corr/rollout_is_mean:1.0000660419464111 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5070068836212158 - rollout_corr/rollout_is_min:0.5062825679779053 - rollout_corr/rollout_is_std:0.02401033416390419 - rollout_corr/rollout_is_eff_sample_size:0.9994239551040021 - rollout_corr/rollout_is_seq_mean:1.0000636577606201 - rollout_corr/rollout_is_seq_std:0.0013239914551377296 - rollout_corr/rollout_is_seq_max:1.0058327913284302 - rollout_corr/rollout_is_seq_min:0.9960933327674866 - rollout_corr/rollout_is_seq_max_deviation:0.005832791328430176 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3059996077790856) - rollout_corr/training_log_ppl:np.float64(0.26158572202257346) - rollout_corr/kl:np.float64(0.0004447607697875) - rollout_corr/k3_kl:np.float64(0.00033729273330607157) - rollout_corr/rollout_ppl:np.float64(1.3054057322442532) - rollout_corr/rollout_log_ppl:np.float64(0.261140958769829) - rollout_corr/log_ppl_diff:np.float64(0.00044476325274445117) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011254070268478245) - rollout_corr/log_ppl_diff_max:np.float64(0.007861137390136719) - rollout_corr/log_ppl_diff_min:np.float64(-0.00546593964099884) - rollout_corr/ppl_ratio:np.float64(1.0004459931515157) - rollout_corr/chi2_token:np.float64(0.0004484797827899456) - rollout_corr/chi2_seq:np.float64(0.2356518420856446) - actor/pg_loss:np.float64(2.136023249477148e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(5.410318408394232e-06) - actor/ppo_kl:np.float64(0.00019875600776231295) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.09427732788026333) - perf/mfu/actor:np.float64(0.1048240639539772) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.44162034988403) - actor/lr:np.float64(1e-06) - training/global_step:35 - training/epoch:1 - critic/score/mean:0.76171875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.76171875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.006490063853561878 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.006490063853561878 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:427.3359375 - response_length/max:854.0 - response_length/min:151.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:427.3359375 - response_length_non_aborted/max:854.0 - response_length_non_aborted/min:151.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:384.59375 - prompt_length/max:459.0 - prompt_length/min:274.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00015152059495449066 - timing_s/agent_loop/generate_sequences/min:np.float64(1.993311908096075) - timing_s/agent_loop/generate_sequences/max:np.float64(7.891210563480854) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.914158031700936) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.891210563480854) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/pr
|
||
step:36 - global_seqlen/min:23449 - global_seqlen/max:31040 - global_seqlen/minmax_diff:7591 - global_seqlen/balanced_min:27587 - global_seqlen/balanced_max:27592 - global_seqlen/mean:27589.5 - actor/entropy:0.2709210216999054 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24491968750953674 - training/rollout_probs_diff_mean:0.002717098919674754 - training/rollout_probs_diff_std:0.007521712686866522 - training/rollout_actor_probs_pearson_corr:0.9994779825210571 - rollout_corr/rollout_is_mean:0.9999476075172424 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.64872407913208 - rollout_corr/rollout_is_min:0.6180930137634277 - rollout_corr/rollout_is_std:0.024854427203536034 - rollout_corr/rollout_is_eff_sample_size:0.9993825197696726 - rollout_corr/rollout_is_seq_mean:0.9999822974205017 - rollout_corr/rollout_is_seq_std:0.0013898462057113647 - rollout_corr/rollout_is_seq_max:1.00667142868042 - rollout_corr/rollout_is_seq_min:0.9957444667816162 - rollout_corr/rollout_is_seq_max_deviation:0.006671428680419922 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3220618744380772) - rollout_corr/training_log_ppl:np.float64(0.2742194219317753) - rollout_corr/kl:np.float64(0.00034166475395225326) - rollout_corr/k3_kl:np.float64(0.0003441765950782383) - rollout_corr/rollout_ppl:np.float64(1.321584774646908) - rollout_corr/rollout_log_ppl:np.float64(0.27387775406532455) - rollout_corr/log_ppl_diff:np.float64(0.00034166786645073444) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011080713447881863) - rollout_corr/log_ppl_diff_max:np.float64(0.006220310926437378) - rollout_corr/log_ppl_diff_min:np.float64(-0.005935758352279663) - rollout_corr/ppl_ratio:np.float64(1.0003428137861192) - rollout_corr/chi2_token:np.float64(0.0006969412788748741) - rollout_corr/chi2_seq:np.float64(0.2664695552084595) - actor/pg_loss:np.float64(-3.446557093411684e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(8.949851553552435e-05) - actor/ppo_kl:np.float64(4.116838032253156e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.13451017811894417) - perf/mfu/actor:np.float64(0.11006513992096655) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.43120193481445) - actor/lr:np.float64(1e-06) - training/global_step:36 - training/epoch:1 - critic/score/mean:0.59765625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.59765625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0015561387408524752 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0015561387408524752 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:464.390625 - response_length/max:878.0 - response_length/min:168.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:464.390625 - response_length_non_aborted/max:878.0 - response_length_non_aborted/min:168.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:397.78125 - prompt_length/max:509.0 - prompt_length/min:263.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00013710744678974152 - timing_s/agent_loop/generate_sequences/min:np.float64(1.3757042922079563) - timing_s/agent_loop/generate_sequences/max:np.float64(8.467374807223678) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.015290737050236) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.467374807223678) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/
|
||
step:37 - global_seqlen/min:20238 - global_seqlen/max:29602 - global_seqlen/minmax_diff:9364 - global_seqlen/balanced_min:25032 - global_seqlen/balanced_max:25035 - global_seqlen/mean:25033.0 - actor/entropy:0.2622256875038147 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23438845574855804 - training/rollout_probs_diff_mean:0.0027199105825275183 - training/rollout_probs_diff_std:0.0077196042984724045 - training/rollout_actor_probs_pearson_corr:0.9994896054267883 - rollout_corr/rollout_is_mean:0.9999966621398926 - rollout_corr/rollout_is_ratio_fraction_high:9.025270628626458e-06 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.0277390480041504 - rollout_corr/rollout_is_min:0.6365309953689575 - rollout_corr/rollout_is_std:0.02535303682088852 - rollout_corr/rollout_is_eff_sample_size:0.9993577554631589 - rollout_corr/rollout_is_seq_mean:0.9999998807907104 - rollout_corr/rollout_is_seq_std:0.0014201079029589891 - rollout_corr/rollout_is_seq_max:1.0045517683029175 - rollout_corr/rollout_is_seq_min:0.9944545030593872 - rollout_corr/rollout_is_seq_max_deviation:0.005545496940612793 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3239887948147953) - rollout_corr/training_log_ppl:np.float64(0.26963052261271514) - rollout_corr/kl:np.float64(0.00043052457150238865) - rollout_corr/k3_kl:np.float64(0.0004052748620466673) - rollout_corr/rollout_ppl:np.float64(1.3233865266665816) - rollout_corr/rollout_log_ppl:np.float64(0.26919999830715824) - rollout_corr/log_ppl_diff:np.float64(0.0004305243055569008) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012312442268012092) - rollout_corr/log_ppl_diff_max:np.float64(0.004594564437866211) - rollout_corr/log_ppl_diff_min:np.float64(-0.003984987735748291) - rollout_corr/ppl_ratio:np.float64(1.0004317718558013) - rollout_corr/chi2_token:np.float64(0.0007657709065824747) - rollout_corr/chi2_seq:np.float64(0.3679836820811033) - actor/pg_loss:np.float64(-2.334697637706995e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(5.749956244471832e-05) - actor/ppo_kl:np.float64(9.06747865041524e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.15217697247862816) - perf/mfu/actor:np.float64(0.10164436124329765) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.42990493774414) - actor/lr:np.float64(1e-06) - training/global_step:37 - training/epoch:1 - critic/score/mean:0.796875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.796875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0048601082526147366 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0048601082526147366 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:432.8125 - response_length/max:895.0 - response_length/min:173.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:432.8125 - response_length_non_aborted/max:895.0 - response_length_non_aborted/min:173.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:349.46875 - prompt_length/max:491.0 - prompt_length/min:203.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.0001987311989068985 - timing_s/agent_loop/generate_sequences/min:np.float64(2.1476458813995123) - timing_s/agent_loop/generate_sequences/max:np.float64(8.263966159895062) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.842065687218565) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.263966159895062) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_l
|
||
step:38 - global_seqlen/min:22776 - global_seqlen/max:30916 - global_seqlen/minmax_diff:8140 - global_seqlen/balanced_min:27403 - global_seqlen/balanced_max:27410 - global_seqlen/mean:27406.75 - actor/entropy:0.2778184711933136 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5546000003814697 - training/rollout_probs_diff_mean:0.0026332649867981672 - training/rollout_probs_diff_std:0.007570348680019379 - training/rollout_actor_probs_pearson_corr:0.9994733929634094 - rollout_corr/rollout_is_mean:1.0000190734863281 - rollout_corr/rollout_is_ratio_fraction_high:8.395601071242709e-06 - rollout_corr/rollout_is_ratio_fraction_low:1.6791202142485417e-05 - rollout_corr/rollout_is_max:2.175250768661499 - rollout_corr/rollout_is_min:0.20461466908454895 - rollout_corr/rollout_is_std:0.024789592251181602 - rollout_corr/rollout_is_eff_sample_size:0.9993858535200482 - rollout_corr/rollout_is_seq_mean:1.0000555515289307 - rollout_corr/rollout_is_seq_std:0.0012281483504921198 - rollout_corr/rollout_is_seq_max:1.0044065713882446 - rollout_corr/rollout_is_seq_min:0.9965072870254517 - rollout_corr/rollout_is_seq_max_deviation:0.004406571388244629 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3290739133954048) - rollout_corr/training_log_ppl:np.float64(0.27583638252690434) - rollout_corr/kl:np.float64(0.00030358191989154903) - rollout_corr/k3_kl:np.float64(0.00035663118822526485) - rollout_corr/rollout_ppl:np.float64(1.3286687950603664) - rollout_corr/rollout_log_ppl:np.float64(0.2755327984923497) - rollout_corr/log_ppl_diff:np.float64(0.0003035840345546603) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011323641519993544) - rollout_corr/log_ppl_diff_max:np.float64(0.005536884069442749) - rollout_corr/log_ppl_diff_min:np.float64(-0.0041558146476745605) - rollout_corr/ppl_ratio:np.float64(1.0003046747297049) - rollout_corr/chi2_token:np.float64(0.0008237855508923531) - rollout_corr/chi2_seq:np.float64(0.5890746030490845) - actor/pg_loss:np.float64(-7.543503306806087e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(8.630327238279278e-05) - actor/ppo_kl:np.float64(3.2657569320804214e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1603358257561922) - perf/mfu/actor:np.float64(0.10901964703649844) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.42809295654297) - actor/lr:np.float64(1e-06) - training/global_step:38 - training/epoch:1 - critic/score/mean:0.59765625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.59765625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.003431701799854636 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.003431701799854636 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:465.2734375 - response_length/max:1004.0 - response_length/min:154.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:465.2734375 - response_length_non_aborted/max:1004.0 - response_length_non_aborted/min:154.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:391.1875 - prompt_length/max:541.0 - prompt_length/min:189.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:7.173791527748108e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.7882296908646822) - timing_s/agent_loop/generate_sequences/max:np.float64(8.705533424392343) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.197099420365703) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.705533424392343) - timing_s/agent_loop/slowest/tool_calls:np
|
||
step:39 - global_seqlen/min:21926 - global_seqlen/max:31836 - global_seqlen/minmax_diff:9910 - global_seqlen/balanced_min:26048 - global_seqlen/balanced_max:26052 - global_seqlen/mean:26050.5 - actor/entropy:0.23975405097007751 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3535163402557373 - training/rollout_probs_diff_mean:0.0025534790474921465 - training/rollout_probs_diff_std:0.0075296261347830296 - training/rollout_actor_probs_pearson_corr:0.9994619488716125 - rollout_corr/rollout_is_mean:1.0000462532043457 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.9363714456558228 - rollout_corr/rollout_is_min:0.6228123307228088 - rollout_corr/rollout_is_std:0.024104483425617218 - rollout_corr/rollout_is_eff_sample_size:0.9994194303888935 - rollout_corr/rollout_is_seq_mean:1.0000656843185425 - rollout_corr/rollout_is_seq_std:0.001369856996461749 - rollout_corr/rollout_is_seq_max:1.004970669746399 - rollout_corr/rollout_is_seq_min:0.9958161115646362 - rollout_corr/rollout_is_seq_max_deviation:0.004970669746398926 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2866896665655077) - rollout_corr/training_log_ppl:np.float64(0.2466871515935054) - rollout_corr/kl:np.float64(0.0002536588835226894) - rollout_corr/k3_kl:np.float64(0.0003328794842047955) - rollout_corr/rollout_ppl:np.float64(1.2863487033173442) - rollout_corr/rollout_log_ppl:np.float64(0.24643349040707108) - rollout_corr/log_ppl_diff:np.float64(0.00025366118643432856) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010324943286832422) - rollout_corr/log_ppl_diff_max:np.float64(0.004928857088088989) - rollout_corr/log_ppl_diff_min:np.float64(-0.004723459482192993) - rollout_corr/ppl_ratio:np.float64(1.0002546701580286) - rollout_corr/chi2_token:np.float64(0.0008289911784231663) - rollout_corr/chi2_seq:np.float64(0.5572728242259473) - actor/pg_loss:np.float64(7.081078365445137e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(1.9370347672520438e-05) - actor/ppo_kl:np.float64(1.5183129935536499e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.12078280001878738) - perf/mfu/actor:np.float64(0.10567685543937032) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.3458366394043) - actor/lr:np.float64(1e-06) - training/global_step:39 - training/epoch:1 - critic/score/mean:0.7265625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7265625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0008759010815992951 - critic/advantages/max:0.5 - critic/advantages/min:-0.875 - critic/returns/mean:0.0008759010815992951 - critic/returns/max:0.5 - critic/returns/min:-0.875 - response_length/mean:427.015625 - response_length/max:883.0 - response_length/min:164.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:427.015625 - response_length_non_aborted/max:883.0 - response_length_non_aborted/min:164.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:387.0625 - prompt_length/max:505.0 - prompt_length/min:263.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.0001429375261068344 - timing_s/agent_loop/generate_sequences/min:np.float64(2.152031809091568) - timing_s/agent_loop/generate_sequences/max:np.float64(7.988802125677466) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.852606363965606) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.988802125677466) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/prompt_
|
||
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:
|
||
|
||
$$ V = a \times b \times c \times \sqrt{1 + 2\cos\alpha\cos\beta\cos\gamma - \cos^2\alpha - \cos^2\beta - \cos^2\gamma} $$
|
||
|
||
For a cubic crystal structure (which this is, since α = β = γ = 90.00º), 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:
|
||
|
||
$$ 61.1524 \times 17.88 = 1094.31 $$
|
||
|
||
Therefore, the volume of the unit cell is 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-GRPO-Qwen-Qwen3-8B-mbs8-train32-rollout8-lr1e-6-vllm0.8/global_step_40
|
||
step:40 - global_seqlen/min:20189 - global_seqlen/max:28119 - global_seqlen/minmax_diff:7930 - global_seqlen/balanced_min:24920 - global_seqlen/balanced_max:24923 - global_seqlen/mean:24921.875 - actor/entropy:0.2692209780216217 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2310577630996704 - training/rollout_probs_diff_mean:0.002756730420514941 - training/rollout_probs_diff_std:0.007605041842907667 - training/rollout_actor_probs_pearson_corr:0.9995124936103821 - rollout_corr/rollout_is_mean:1.0000061988830566 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4278150796890259 - rollout_corr/rollout_is_min:0.5504452586174011 - rollout_corr/rollout_is_std:0.024521248415112495 - rollout_corr/rollout_is_eff_sample_size:0.9993990696778947 - rollout_corr/rollout_is_seq_mean:1.000065803527832 - rollout_corr/rollout_is_seq_std:0.001380614354275167 - rollout_corr/rollout_is_seq_max:1.0054199695587158 - rollout_corr/rollout_is_seq_min:0.9957116842269897 - rollout_corr/rollout_is_seq_max_deviation:0.00541996955871582 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3192760865204036) - rollout_corr/training_log_ppl:np.float64(0.2728104801790323) - rollout_corr/kl:np.float64(0.0001471567377218097) - rollout_corr/k3_kl:np.float64(0.00035515559358145765) - rollout_corr/rollout_ppl:np.float64(1.319087610580027) - rollout_corr/rollout_log_ppl:np.float64(0.272663321782602) - rollout_corr/log_ppl_diff:np.float64(0.0001471583964303136) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011645762133412063) - rollout_corr/log_ppl_diff_max:np.float64(0.0038172006607055664) - rollout_corr/log_ppl_diff_min:np.float64(-0.004808694124221802) - rollout_corr/ppl_ratio:np.float64(1.00014826701954) - rollout_corr/chi2_token:np.float64(0.0011329331900924444) - rollout_corr/chi2_seq:np.float64(0.39967853645794094) - actor/pg_loss:np.float64(-8.308794349431992e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(5.9222728395980084e-05) - actor/ppo_kl:np.float64(-0.00010214552137752242) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.08158384636044502) - perf/mfu/actor:np.float64(0.1011381841099553) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.36322402954102) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7240691489361702) - val-aux/sciknoweval/reward/std@16:np.float64(0.10122444039363532) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7613085106382979) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.06993133531537328) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6865638297872341) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.09972499691001828) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7238404255319149) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.10065446646187623) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7869787234042553) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.04375591618344764) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6497978723404255) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.09380131256289198) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7325425531914893) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.07347374709542713) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.8037234042553191) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.02647539161149547) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6118829787234042) - val-aux/sciknoweval/reward/worst@8/std:np.float64(0.0869859468809813) - val-aux/sciknoweval/reward/maj@8/mean:np.float64(0.7364999999999999) - val-aux/sciknoweval/reward/maj@8/std:np.float64(0.047862660217948536) - val-aux/sciknoweval/reward/best@16/mean:np.float64(0.8132234042553192) - val-aux/sciknoweval/reward/best@1
|
||
step:41 - global_seqlen/min:22101 - global_seqlen/max:30341 - global_seqlen/minmax_diff:8240 - global_seqlen/balanced_min:26800 - global_seqlen/balanced_max:26806 - global_seqlen/mean:26804.875 - actor/entropy:0.24527297914028168 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23105892539024353 - training/rollout_probs_diff_mean:0.0024506240151822567 - training/rollout_probs_diff_std:0.007172984071075916 - training/rollout_actor_probs_pearson_corr:0.9995378255844116 - rollout_corr/rollout_is_mean:1.0000022649765015 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5441513061523438 - rollout_corr/rollout_is_min:0.5378821492195129 - rollout_corr/rollout_is_std:0.023597173392772675 - rollout_corr/rollout_is_eff_sample_size:0.99944348329181 - rollout_corr/rollout_is_seq_mean:0.9999989867210388 - rollout_corr/rollout_is_seq_std:0.0013331422815099359 - rollout_corr/rollout_is_seq_max:1.0063869953155518 - rollout_corr/rollout_is_seq_min:0.9958586096763611 - rollout_corr/rollout_is_seq_max_deviation:0.006386995315551758 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2987580224871635) - rollout_corr/training_log_ppl:np.float64(0.25461333782732254) - rollout_corr/kl:np.float64(0.00037003947440439333) - rollout_corr/k3_kl:np.float64(0.000317131163605211) - rollout_corr/rollout_ppl:np.float64(1.298254407942295) - rollout_corr/rollout_log_ppl:np.float64(0.25424329534871504) - rollout_corr/log_ppl_diff:np.float64(0.00037004247860750183) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010555475237197243) - rollout_corr/log_ppl_diff_max:np.float64(0.005651324987411499) - rollout_corr/log_ppl_diff_min:np.float64(-0.005803555250167847) - rollout_corr/ppl_ratio:np.float64(1.0003710864111781) - rollout_corr/chi2_token:np.float64(0.0005223318003118038) - rollout_corr/chi2_seq:np.float64(0.19784280378371477) - actor/pg_loss:np.float64(4.0626153349876404e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(6.391483702827827e-05) - actor/ppo_kl:np.float64(6.689386933178554e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.12102004140615463) - perf/mfu/actor:np.float64(0.10706792224524682) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.53229522705078) - actor/lr:np.float64(1e-06) - training/global_step:41 - training/epoch:1 - critic/score/mean:0.56640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.56640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0009183372603729367 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0009183372603729367 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:437.05859375 - response_length/max:997.0 - response_length/min:117.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:437.05859375 - response_length_non_aborted/max:997.0 - response_length_non_aborted/min:117.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:400.59375 - prompt_length/max:499.0 - prompt_length/min:266.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:6.401911377906799e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.5086693558841944) - timing_s/agent_loop/generate_sequences/max:np.float64(8.868929112330079) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.9913268628079095) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.868929112330079) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_lo
|
||
step:42 - global_seqlen/min:25339 - global_seqlen/max:30354 - global_seqlen/minmax_diff:5015 - global_seqlen/balanced_min:27648 - global_seqlen/balanced_max:27657 - global_seqlen/mean:27654.375 - actor/entropy:0.249659925699234 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.30163419246673584 - training/rollout_probs_diff_mean:0.002381665166467428 - training/rollout_probs_diff_std:0.007057615090161562 - training/rollout_actor_probs_pearson_corr:0.9996173977851868 - rollout_corr/rollout_is_mean:1.0001263618469238 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:8.602668458479457e-06 - rollout_corr/rollout_is_max:1.9402073621749878 - rollout_corr/rollout_is_min:0.44588688015937805 - rollout_corr/rollout_is_std:0.023735692724585533 - rollout_corr/rollout_is_eff_sample_size:0.9994370531944818 - rollout_corr/rollout_is_seq_mean:1.0001702308654785 - rollout_corr/rollout_is_seq_std:0.0012351945042610168 - rollout_corr/rollout_is_seq_max:1.0057711601257324 - rollout_corr/rollout_is_seq_min:0.9974271655082703 - rollout_corr/rollout_is_seq_max_deviation:0.005771160125732422 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3183727939613163) - rollout_corr/training_log_ppl:np.float64(0.2636420357521274) - rollout_corr/kl:np.float64(0.0003492014980753799) - rollout_corr/k3_kl:np.float64(0.00035256788812176865) - rollout_corr/rollout_ppl:np.float64(1.3178962673991919) - rollout_corr/rollout_log_ppl:np.float64(0.2632928293242003) - rollout_corr/log_ppl_diff:np.float64(0.00034920642792712897) - rollout_corr/log_ppl_abs_diff:np.float64(0.001064916534232907) - rollout_corr/log_ppl_diff_max:np.float64(0.01382581889629364) - rollout_corr/log_ppl_diff_min:np.float64(-0.004806399345397949) - rollout_corr/ppl_ratio:np.float64(1.0003504659980536) - rollout_corr/chi2_token:np.float64(0.0006320546381175518) - rollout_corr/chi2_seq:np.float64(0.2667905113194138) - actor/pg_loss:np.float64(2.428167499601841e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(1.0820636816788465e-05) - actor/ppo_kl:np.float64(0.00022477524237096702) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.13300273567438126) - perf/mfu/actor:np.float64(0.11066785146384982) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.22729873657227) - actor/lr:np.float64(1e-06) - training/global_step:42 - training/epoch:1 - 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:-0.0015237476909533143 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0015237476909533143 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:454.07421875 - response_length/max:830.0 - response_length/min:170.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:454.07421875 - response_length_non_aborted/max:830.0 - response_length_non_aborted/min:170.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:410.125 - prompt_length/max:510.0 - prompt_length/min:197.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:2.5350600481033325e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.0540550239384174) - timing_s/agent_loop/generate_sequences/max:np.float64(7.967892354354262) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.309633722696162) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.967892354354262) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agen
|
||
step:43 - global_seqlen/min:22391 - global_seqlen/max:31176 - global_seqlen/minmax_diff:8785 - global_seqlen/balanced_min:27125 - global_seqlen/balanced_max:27133 - global_seqlen/mean:27130.5 - actor/entropy:0.2829850912094116 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23087042570114136 - training/rollout_probs_diff_mean:0.0026621695142239332 - training/rollout_probs_diff_std:0.007386308629065752 - training/rollout_actor_probs_pearson_corr:0.9996067881584167 - rollout_corr/rollout_is_mean:1.0000057220458984 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.773489930201322e-05 - rollout_corr/rollout_is_max:1.7430531978607178 - rollout_corr/rollout_is_min:0.22278240323066711 - rollout_corr/rollout_is_std:0.02498597651720047 - rollout_corr/rollout_is_eff_sample_size:0.9993760904567573 - rollout_corr/rollout_is_seq_mean:1.0000249147415161 - rollout_corr/rollout_is_seq_std:0.0012387627502903342 - rollout_corr/rollout_is_seq_max:1.0044147968292236 - rollout_corr/rollout_is_seq_min:0.9966818690299988 - rollout_corr/rollout_is_seq_max_deviation:0.004414796829223633 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.347760524135083) - rollout_corr/training_log_ppl:np.float64(0.28952421667054296) - rollout_corr/kl:np.float64(0.0003012928969590156) - rollout_corr/k3_kl:np.float64(0.0003403808943005515) - rollout_corr/rollout_ppl:np.float64(1.3473055744543672) - rollout_corr/rollout_log_ppl:np.float64(0.28922292120114435) - rollout_corr/log_ppl_diff:np.float64(0.0003012954693986103) - rollout_corr/log_ppl_abs_diff:np.float64(0.0009853317314991727) - rollout_corr/log_ppl_diff_max:np.float64(0.004570573568344116) - rollout_corr/log_ppl_diff_min:np.float64(-0.004001826047897339) - rollout_corr/ppl_ratio:np.float64(1.0003021161537617) - rollout_corr/chi2_token:np.float64(0.0007619182579219341) - rollout_corr/chi2_seq:np.float64(0.20320769771933556) - actor/pg_loss:np.float64(3.108277451246977e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0001063900026565534) - actor/ppo_kl:np.float64(1.231407157575859e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1578393466770649) - perf/mfu/actor:np.float64(0.10808209957965548) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.30550813674927) - actor/lr:np.float64(1e-06) - training/global_step:43 - training/epoch:1 - critic/score/mean:0.53515625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.53515625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0023088620509952307 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0023088620509952307 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:440.515625 - response_length/max:898.0 - response_length/min:174.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:440.515625 - response_length_non_aborted/max:898.0 - response_length_non_aborted/min:174.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:407.3125 - prompt_length/max:494.0 - prompt_length/min:196.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00016414374113082886 - timing_s/agent_loop/generate_sequences/min:np.float64(2.200499527156353) - timing_s/agent_loop/generate_sequences/max:np.float64(8.465374808758497) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.1702591333087184) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.465374808758497) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/a
|
||
step:44 - global_seqlen/min:23687 - global_seqlen/max:28489 - global_seqlen/minmax_diff:4802 - global_seqlen/balanced_min:25919 - global_seqlen/balanced_max:25924 - global_seqlen/mean:25921.0 - actor/entropy:0.23762044310569763 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.17917698621749878 - training/rollout_probs_diff_mean:0.002502892632037401 - training/rollout_probs_diff_std:0.007343742996454239 - training/rollout_actor_probs_pearson_corr:0.9994988441467285 - rollout_corr/rollout_is_mean:0.9999760985374451 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:9.069801308214664e-06 - rollout_corr/rollout_is_max:1.5526788234710693 - rollout_corr/rollout_is_min:0.48514747619628906 - rollout_corr/rollout_is_std:0.02382592298090458 - rollout_corr/rollout_is_eff_sample_size:0.9994326474348206 - rollout_corr/rollout_is_seq_mean:0.9999829530715942 - rollout_corr/rollout_is_seq_std:0.0012827671598643064 - rollout_corr/rollout_is_seq_max:1.005139708518982 - rollout_corr/rollout_is_seq_min:0.9958560466766357 - rollout_corr/rollout_is_seq_max_deviation:0.005139708518981934 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2832285850308836) - rollout_corr/training_log_ppl:np.float64(0.24504046242509503) - rollout_corr/kl:np.float64(0.0004000083458315373) - rollout_corr/k3_kl:np.float64(0.00034762927705855873) - rollout_corr/rollout_ppl:np.float64(1.282694707158953) - rollout_corr/rollout_log_ppl:np.float64(0.24464045157947112) - rollout_corr/log_ppl_diff:np.float64(0.0004000108456239104) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010354671976529062) - rollout_corr/log_ppl_diff_max:np.float64(0.004307284951210022) - rollout_corr/log_ppl_diff_min:np.float64(-0.004696741700172424) - rollout_corr/ppl_ratio:np.float64(1.0004009895492345) - rollout_corr/chi2_token:np.float64(0.0005764113739132881) - rollout_corr/chi2_seq:np.float64(0.057558765169233084) - actor/pg_loss:np.float64(-1.6392907127738e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00010265427090416779) - actor/ppo_kl:np.float64(8.106677549335473e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.12191110476851463) - perf/mfu/actor:np.float64(0.10458949772185872) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.37454605102539) - actor/lr:np.float64(1e-06) - training/global_step:44 - training/epoch:1 - critic/score/mean:0.703125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.703125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.002656318014487624 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.002656318014487624 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:430.6875 - response_length/max:820.0 - response_length/min:155.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:430.6875 - response_length_non_aborted/max:820.0 - response_length_non_aborted/min:155.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:379.34375 - prompt_length/max:524.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:4.637613892555237e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8315623067319393) - timing_s/agent_loop/generate_sequences/max:np.float64(7.903299419209361) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.8979601562678) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.903299419209361) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/sl
|
||
step:45 - global_seqlen/min:20724 - global_seqlen/max:29711 - global_seqlen/minmax_diff:8987 - global_seqlen/balanced_min:25302 - global_seqlen/balanced_max:25306 - global_seqlen/mean:25304.125 - actor/entropy:0.26102858781814575 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.19273829460144043 - training/rollout_probs_diff_mean:0.0026539494283497334 - training/rollout_probs_diff_std:0.007585626561194658 - training/rollout_actor_probs_pearson_corr:0.9995617866516113 - rollout_corr/rollout_is_mean:1.000087022781372 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.9513788223266602 - rollout_corr/rollout_is_min:0.568621039390564 - rollout_corr/rollout_is_std:0.02511446550488472 - rollout_corr/rollout_is_eff_sample_size:0.9993698993447263 - rollout_corr/rollout_is_seq_mean:1.0000715255737305 - rollout_corr/rollout_is_seq_std:0.0013407886726781726 - rollout_corr/rollout_is_seq_max:1.0047191381454468 - rollout_corr/rollout_is_seq_min:0.9964459538459778 - rollout_corr/rollout_is_seq_max_deviation:0.004719138145446777 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3035321370698512) - rollout_corr/training_log_ppl:np.float64(0.2587986676080618) - rollout_corr/kl:np.float64(0.0003129965853041483) - rollout_corr/k3_kl:np.float64(0.0003379961697334011) - rollout_corr/rollout_ppl:np.float64(1.3031289116479456) - rollout_corr/rollout_log_ppl:np.float64(0.258485670099617) - rollout_corr/log_ppl_diff:np.float64(0.0003129975084448233) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011024432169506326) - rollout_corr/log_ppl_diff_max:np.float64(0.003939390182495117) - rollout_corr/log_ppl_diff_min:np.float64(-0.004243284463882446) - rollout_corr/ppl_ratio:np.float64(1.0003140484914184) - rollout_corr/chi2_token:np.float64(0.000735103152692318) - rollout_corr/chi2_seq:np.float64(0.5922304301057011) - actor/pg_loss:np.float64(2.7036177925765514e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00015170505503192544) - actor/ppo_kl:np.float64(5.9133477579464966e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.17508666589856148) - perf/mfu/actor:np.float64(0.10034117675794652) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(111.28366470336914) - actor/lr:np.float64(1e-06) - training/global_step:45 - 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:-0.004022400826215744 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:-0.004022400826215744 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:420.25390625 - response_length/max:906.0 - response_length/min:164.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:420.25390625 - response_length_non_aborted/max:906.0 - response_length_non_aborted/min:164.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:370.5 - prompt_length/max:483.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00010674819350242615 - timing_s/agent_loop/generate_sequences/min:np.float64(2.033995635807514) - timing_s/agent_loop/generate_sequences/max:np.float64(8.038469087332487) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.785468361864332) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.038469087332487) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/pr
|
||
step:46 - global_seqlen/min:23160 - global_seqlen/max:28431 - global_seqlen/minmax_diff:5271 - global_seqlen/balanced_min:25774 - global_seqlen/balanced_max:25777 - global_seqlen/mean:25775.375 - actor/entropy:0.2477809637784958 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2413540482521057 - training/rollout_probs_diff_mean:0.0025389681104570627 - training/rollout_probs_diff_std:0.007506586145609617 - training/rollout_actor_probs_pearson_corr:0.9995504021644592 - rollout_corr/rollout_is_mean:1.0000053644180298 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.605976939201355 - rollout_corr/rollout_is_min:0.5050538778305054 - rollout_corr/rollout_is_std:0.02421550080180168 - rollout_corr/rollout_is_eff_sample_size:0.9994139531569982 - rollout_corr/rollout_is_seq_mean:1.0000325441360474 - rollout_corr/rollout_is_seq_std:0.0013734105741605163 - rollout_corr/rollout_is_seq_max:1.0052671432495117 - rollout_corr/rollout_is_seq_min:0.9960712194442749 - rollout_corr/rollout_is_seq_max_deviation:0.005267143249511719 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.301206385716796) - rollout_corr/training_log_ppl:np.float64(0.25428796322376) - rollout_corr/kl:np.float64(0.0002562967966284191) - rollout_corr/k3_kl:np.float64(0.0003318889672527803) - rollout_corr/rollout_ppl:np.float64(1.3008486884646118) - rollout_corr/rollout_log_ppl:np.float64(0.25403166448813863) - rollout_corr/log_ppl_diff:np.float64(0.0002562987356213853) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010766688355943188) - rollout_corr/log_ppl_diff_max:np.float64(0.0048729777336120605) - rollout_corr/log_ppl_diff_min:np.float64(-0.004036739468574524) - rollout_corr/ppl_ratio:np.float64(1.000257351435721) - rollout_corr/chi2_token:np.float64(0.0008127458859235048) - rollout_corr/chi2_seq:np.float64(0.213467349531129) - actor/pg_loss:np.float64(-2.8856913559138775e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(3.868811063512112e-05) - actor/ppo_kl:np.float64(-1.7918505959002573e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1389861050993204) - perf/mfu/actor:np.float64(0.10119552924817618) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(149.76423168182373) - actor/lr:np.float64(1e-06) - training/global_step:46 - training/epoch:1 - critic/score/mean:0.71484375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.71484375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.006029889453202486 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.006029889453202486 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:424.48046875 - response_length/max:872.0 - response_length/min:160.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:424.48046875 - response_length_non_aborted/max:872.0 - response_length_non_aborted/min:160.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:381.0 - prompt_length/max:481.0 - prompt_length/min:200.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:2.641230821609497e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.0706292800605297) - timing_s/agent_loop/generate_sequences/max:np.float64(8.012809984385967) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.769642198676593) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.012809984385967) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/pr
|
||
step:47 - global_seqlen/min:21281 - global_seqlen/max:28817 - global_seqlen/minmax_diff:7536 - global_seqlen/balanced_min:26442 - global_seqlen/balanced_max:26454 - global_seqlen/mean:26450.75 - actor/entropy:0.27556195855140686 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.19444799423217773 - training/rollout_probs_diff_mean:0.002736711408942938 - training/rollout_probs_diff_std:0.0076127080246806145 - training/rollout_actor_probs_pearson_corr:0.9995213150978088 - rollout_corr/rollout_is_mean:1.0000603199005127 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:8.98327289178269e-06 - rollout_corr/rollout_is_max:1.6245206594467163 - rollout_corr/rollout_is_min:0.21295833587646484 - rollout_corr/rollout_is_std:0.024904737249016762 - rollout_corr/rollout_is_eff_sample_size:0.9993802575945846 - rollout_corr/rollout_is_seq_mean:1.000070571899414 - rollout_corr/rollout_is_seq_std:0.0013441572664305568 - rollout_corr/rollout_is_seq_max:1.0038058757781982 - rollout_corr/rollout_is_seq_min:0.9955684542655945 - rollout_corr/rollout_is_seq_max_deviation:0.004431545734405518 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3388532362878323) - rollout_corr/training_log_ppl:np.float64(0.2835671211942099) - rollout_corr/kl:np.float64(0.00031976593474931925) - rollout_corr/k3_kl:np.float64(0.00036261727734654414) - rollout_corr/rollout_ppl:np.float64(1.338419138919562) - rollout_corr/rollout_log_ppl:np.float64(0.2832473532325821) - rollout_corr/log_ppl_diff:np.float64(0.00031976796162780374) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011114145017927513) - rollout_corr/log_ppl_diff_max:np.float64(0.005098104476928711) - rollout_corr/log_ppl_diff_min:np.float64(-0.00347292423248291) - rollout_corr/ppl_ratio:np.float64(1.0003207672853023) - rollout_corr/chi2_token:np.float64(0.0008113444782793522) - rollout_corr/chi2_seq:np.float64(0.28644665400497615) - actor/pg_loss:np.float64(-4.764122422784567e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0001026247907702782) - actor/ppo_kl:np.float64(6.258977063389182e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1877906322479248) - perf/mfu/actor:np.float64(0.10244830140367661) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(129.80683517456055) - actor/lr:np.float64(1e-06) - training/global_step:47 - training/epoch:1 - 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:-0.009874862618744373 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.009874862618744373 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:434.8359375 - response_length/max:1078.0 - response_length/min:136.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:434.8359375 - response_length_non_aborted/max:1078.0 - response_length_non_aborted/min:136.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:391.75 - prompt_length/max:490.0 - prompt_length/min:196.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:3.805197775363922e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8029491119086742) - timing_s/agent_loop/generate_sequences/max:np.float64(9.529983641579747) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.154630884913786) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(9.529983641579747) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_
|
||
step:48 - global_seqlen/min:22277 - global_seqlen/max:30571 - global_seqlen/minmax_diff:8294 - global_seqlen/balanced_min:26173 - global_seqlen/balanced_max:26178 - global_seqlen/mean:26176.25 - actor/entropy:0.2385566085577011 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24515900015830994 - training/rollout_probs_diff_mean:0.0024435243103653193 - training/rollout_probs_diff_std:0.00727314967662096 - training/rollout_actor_probs_pearson_corr:0.9995198845863342 - rollout_corr/rollout_is_mean:1.0000426769256592 - rollout_corr/rollout_is_ratio_fraction_high:9.144612704403698e-06 - rollout_corr/rollout_is_ratio_fraction_low:9.144612704403698e-06 - rollout_corr/rollout_is_max:2.0436415672302246 - rollout_corr/rollout_is_min:0.34062984585762024 - rollout_corr/rollout_is_std:0.023853426799178123 - rollout_corr/rollout_is_eff_sample_size:0.9994313376218713 - rollout_corr/rollout_is_seq_mean:1.0000296831130981 - rollout_corr/rollout_is_seq_std:0.0014473841292783618 - rollout_corr/rollout_is_seq_max:1.0059645175933838 - rollout_corr/rollout_is_seq_min:0.9947035908699036 - rollout_corr/rollout_is_seq_max_deviation:0.005964517593383789 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2863476877100766) - rollout_corr/training_log_ppl:np.float64(0.24573822057573125) - rollout_corr/kl:np.float64(0.0003334146562288254) - rollout_corr/k3_kl:np.float64(0.0003732351611951401) - rollout_corr/rollout_ppl:np.float64(1.2859108792617917) - rollout_corr/rollout_log_ppl:np.float64(0.24540480367431883) - rollout_corr/log_ppl_diff:np.float64(0.00033341690141242) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011924598220502958) - rollout_corr/log_ppl_diff_max:np.float64(0.0059059858322143555) - rollout_corr/log_ppl_diff_min:np.float64(-0.00526835024356842) - rollout_corr/ppl_ratio:np.float64(1.0003347243182361) - rollout_corr/chi2_token:np.float64(0.0008353295270353556) - rollout_corr/chi2_seq:np.float64(0.3313969324808568) - actor/pg_loss:np.float64(-4.039297346025705e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(5.439815504360013e-05) - actor/ppo_kl:np.float64(6.001183691495271e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.09218540880829096) - perf/mfu/actor:np.float64(0.10079389831904431) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(123.64847230911255) - actor/lr:np.float64(1e-06) - training/global_step:48 - training/epoch:1 - critic/score/mean:0.85546875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.85546875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0028942700009793043 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0028942700009793043 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:427.1640625 - response_length/max:825.0 - response_length/min:155.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:427.1640625 - response_length_non_aborted/max:825.0 - response_length_non_aborted/min:155.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:390.84375 - prompt_length/max:488.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:4.29004430770874e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9806438311934471) - timing_s/agent_loop/generate_sequences/max:np.float64(7.873849323019385) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.021167605707888) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.873849323019385) - timing_s/agent_loop/slowest/tool_calls:np.flo
|
||
step:49 - global_seqlen/min:21747 - global_seqlen/max:29435 - global_seqlen/minmax_diff:7688 - global_seqlen/balanced_min:25439 - global_seqlen/balanced_max:25505 - global_seqlen/mean:25449.75 - actor/entropy:0.254334956407547 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.25539761781692505 - training/rollout_probs_diff_mean:0.002560534281656146 - training/rollout_probs_diff_std:0.007452011574059725 - training/rollout_actor_probs_pearson_corr:0.999518096446991 - rollout_corr/rollout_is_mean:0.9999451041221619 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.916302219091449e-05 - rollout_corr/rollout_is_max:1.7746413946151733 - rollout_corr/rollout_is_min:0.41666513681411743 - rollout_corr/rollout_is_std:0.024758314713835716 - rollout_corr/rollout_is_eff_sample_size:0.9993874013402829 - rollout_corr/rollout_is_seq_mean:0.9999779462814331 - rollout_corr/rollout_is_seq_std:0.0014876748900860548 - rollout_corr/rollout_is_seq_max:1.0042344331741333 - rollout_corr/rollout_is_seq_min:0.994238555431366 - rollout_corr/rollout_is_seq_max_deviation:0.005761444568634033 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3119332161732018) - rollout_corr/training_log_ppl:np.float64(0.264481845399132) - rollout_corr/kl:np.float64(0.0004063297049228254) - rollout_corr/k3_kl:np.float64(0.0003945719078330967) - rollout_corr/rollout_ppl:np.float64(1.311392041388899) - rollout_corr/rollout_log_ppl:np.float64(0.26407551308511756) - rollout_corr/log_ppl_diff:np.float64(0.0004063323140144348) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013702319120056927) - rollout_corr/log_ppl_diff_max:np.float64(0.007741332054138184) - rollout_corr/log_ppl_diff_min:np.float64(-0.004774883389472961) - rollout_corr/ppl_ratio:np.float64(1.0004080620128661) - rollout_corr/chi2_token:np.float64(0.000757753849029541) - rollout_corr/chi2_seq:np.float64(0.3059838619083166) - actor/pg_loss:np.float64(5.814828909933567e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.00010606717933114851) - actor/ppo_kl:np.float64(5.833629540230589e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.16979572549462318) - perf/mfu/actor:np.float64(0.1016400765494086) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(113.97234725952148) - actor/lr:np.float64(1e-06) - training/global_step:49 - training/epoch:1 - 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:-0.00041071255691349506 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00041071255691349506 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:401.8359375 - response_length/max:1363.0 - response_length/min:147.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:401.8359375 - response_length_non_aborted/max:1363.0 - response_length_non_aborted/min:147.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:393.46875 - prompt_length/max:486.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00010422617197036743 - timing_s/agent_loop/generate_sequences/min:np.float64(1.753613343462348) - timing_s/agent_loop/generate_sequences/max:np.float64(10.484201986342669) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.581973708147416) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(10.484201986342669) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_
|
||
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>
|
||
To calculate the volume of a unit cell, we use the formula for the volume of a unit cell in a crystal structure:
|
||
|
||
$$ V = a \times b \times c \times \sqrt{1 + 2\cos\alpha\cos\beta\cos\gamma - \cos^2\alpha - \cos^2\beta - \cos^2\gamma} $$
|
||
|
||
In this case, the crystal structure is described by lattice parameters:
|
||
|
||
- a = 7.82 Å
|
||
- b = 7.82 Å
|
||
- c = 17.88 Å
|
||
- α = 90.00º
|
||
- β = 90.00º
|
||
- γ = 90.00º
|
||
|
||
Since α, β, and γ are all 90 degrees, the crystal is a cubic structure (or at least has orthogonal axes). For a cubic crystal, the volume simplifies to:
|
||
|
||
$$ V = a \times b \times c $$
|
||
|
||
Substituting the given values:
|
||
|
||
$$ V = 7.82 \times 7.82 \times 17.88 $$
|
||
|
||
First, calculate $ 7.82 \times 7.82 $:
|
||
|
||
$$ 7.82 \times 7.82 = 61.1524 $$
|
||
|
||
Now, multiply by 17.88:
|
||
|
||
$$ 61.1524 \times 17.88 = 1094.31 $$
|
||
|
||
Therefore, the volume of the unit cell is 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-GRPO-Qwen-Qwen3-8B-mbs8-train32-rollout8-lr1e-6-vllm0.8/global_step_50
|
||
step:50 - global_seqlen/min:18531 - global_seqlen/max:28752 - global_seqlen/minmax_diff:10221 - global_seqlen/balanced_min:24590 - global_seqlen/balanced_max:24595 - global_seqlen/mean:24592.875 - actor/entropy:0.22492657601833344 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7781625986099243 - training/rollout_probs_diff_mean:0.002371629234403372 - training/rollout_probs_diff_std:0.00784727931022644 - training/rollout_actor_probs_pearson_corr:0.9994186162948608 - rollout_corr/rollout_is_mean:0.9999779462814331 - rollout_corr/rollout_is_ratio_fraction_high:1.0639316315064207e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.0639316315064207e-05 - rollout_corr/rollout_is_max:2.165529489517212 - rollout_corr/rollout_is_min:0.04572895169258118 - rollout_corr/rollout_is_std:0.02394072338938713 - rollout_corr/rollout_is_eff_sample_size:0.999427170058055 - rollout_corr/rollout_is_seq_mean:1.0000346899032593 - rollout_corr/rollout_is_seq_std:0.001483708037994802 - rollout_corr/rollout_is_seq_max:1.006122350692749 - rollout_corr/rollout_is_seq_min:0.9957391619682312 - rollout_corr/rollout_is_seq_max_deviation:0.0061223506927490234 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2762280572205782) - rollout_corr/training_log_ppl:np.float64(0.2360334648838034) - rollout_corr/kl:np.float64(0.0004081390215331382) - rollout_corr/k3_kl:np.float64(0.0004092514230435995) - rollout_corr/rollout_ppl:np.float64(1.2756929220631719) - rollout_corr/rollout_log_ppl:np.float64(0.23562532204960007) - rollout_corr/log_ppl_diff:np.float64(0.00040814283420331776) - rollout_corr/log_ppl_abs_diff:np.float64(0.001212061324622482) - rollout_corr/log_ppl_diff_max:np.float64(0.015186533331871033) - rollout_corr/log_ppl_diff_min:np.float64(-0.003930956125259399) - rollout_corr/ppl_ratio:np.float64(1.0004097421187907) - rollout_corr/chi2_token:np.float64(0.0007466780953109264) - rollout_corr/chi2_seq:np.float64(0.2214487304445356) - actor/pg_loss:np.float64(5.0445785745978355e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(7.287216749318759e-05) - actor/ppo_kl:np.float64(0.00011426370583222933) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1364841517060995) - perf/mfu/actor:np.float64(0.0992731123109947) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(114.06961441040039) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7406914893617021) - val-aux/sciknoweval/reward/std@16:np.float64(0.06782258470826572) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7657872340425533) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.04877188049764316) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.7148404255319148) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.06455783589932332) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7409255319148936) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.06713673383437797) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7843297872340426) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.03278506425382436) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6915957446808512) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.05908327474495294) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7453297872340426) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.049206425903886845) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7974042553191488) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.01955746993586378) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6668829787234042) - val-aux/sciknoweval/reward/worst@8/std:np.float64(0.051961064875554894) - val-aux/sciknoweval/reward/maj@8/mean:np.float64(0.7445212765957446) - val-aux/sciknoweval/reward/maj@8/std:np.float64(0.031088513575260223) - val-aux/sciknoweval/reward/best@16/mean:np.float64(0.804180851063
|
||
step:51 - global_seqlen/min:21651 - global_seqlen/max:30533 - global_seqlen/minmax_diff:8882 - global_seqlen/balanced_min:24618 - global_seqlen/balanced_max:24626 - global_seqlen/mean:24623.375 - actor/entropy:0.2378806620836258 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.31757521629333496 - training/rollout_probs_diff_mean:0.0024352639447897673 - training/rollout_probs_diff_std:0.007294767536222935 - training/rollout_actor_probs_pearson_corr:0.999496579170227 - rollout_corr/rollout_is_mean:1.0000085830688477 - rollout_corr/rollout_is_ratio_fraction_high:1.0320023648091592e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.7408487796783447 - rollout_corr/rollout_is_min:0.5327962040901184 - rollout_corr/rollout_is_std:0.024035144597291946 - rollout_corr/rollout_is_eff_sample_size:0.9994226453138363 - rollout_corr/rollout_is_seq_mean:0.9999585747718811 - rollout_corr/rollout_is_seq_std:0.0015011404175311327 - rollout_corr/rollout_is_seq_max:1.004288673400879 - rollout_corr/rollout_is_seq_min:0.9942871928215027 - rollout_corr/rollout_is_seq_max_deviation:0.0057128071784973145 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3018347015604377) - rollout_corr/training_log_ppl:np.float64(0.2531574094318785) - rollout_corr/kl:np.float64(0.00036208893200040393) - rollout_corr/k3_kl:np.float64(0.00035660695331785064) - rollout_corr/rollout_ppl:np.float64(1.3012964907102287) - rollout_corr/rollout_log_ppl:np.float64(0.25279531987325754) - rollout_corr/log_ppl_diff:np.float64(0.00036208955862093717) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011715776199707761) - rollout_corr/log_ppl_diff_max:np.float64(0.006238698959350586) - rollout_corr/log_ppl_diff_min:np.float64(-0.003474891185760498) - rollout_corr/ppl_ratio:np.float64(1.0003633103333414) - rollout_corr/chi2_token:np.float64(0.0007233740761876106) - rollout_corr/chi2_seq:np.float64(0.3191821831278503) - actor/pg_loss:np.float64(0.00010972644668072462) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(6.342345659504645e-05) - actor/ppo_kl:np.float64(6.792625837803712e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.09654535725712776) - perf/mfu/actor:np.float64(0.10045939727212769) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(107.78156566619873) - actor/lr:np.float64(1e-06) - training/global_step:51 - training/epoch:1 - critic/score/mean:0.67578125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.67578125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.000906872097402811 - critic/advantages/max:0.875 - critic/advantages/min:-0.25 - critic/returns/mean:0.000906872097402811 - critic/returns/max:0.875 - critic/returns/min:-0.25 - response_length/mean:378.51171875 - response_length/max:976.0 - response_length/min:160.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:378.51171875 - response_length_non_aborted/max:976.0 - response_length_non_aborted/min:160.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:390.96875 - prompt_length/max:519.0 - prompt_length/min:196.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00012427382171154022 - timing_s/agent_loop/generate_sequences/min:np.float64(1.5273418612778187) - timing_s/agent_loop/generate_sequences/max:np.float64(8.357269065454602) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.9491922540328233) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.357269065454602) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - ti
|
||
step:52 - global_seqlen/min:23073 - global_seqlen/max:29801 - global_seqlen/minmax_diff:6728 - global_seqlen/balanced_min:26508 - global_seqlen/balanced_max:26513 - global_seqlen/mean:26511.0 - actor/entropy:0.2446550875902176 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2773001194000244 - training/rollout_probs_diff_mean:0.0023589127231389284 - training/rollout_probs_diff_std:0.007042487617582083 - training/rollout_actor_probs_pearson_corr:0.9995072484016418 - rollout_corr/rollout_is_mean:0.9999248385429382 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:9.183073416352272e-06 - rollout_corr/rollout_is_max:1.4406086206436157 - rollout_corr/rollout_is_min:0.44539976119995117 - rollout_corr/rollout_is_std:0.02308904379606247 - rollout_corr/rollout_is_eff_sample_size:0.9994670610232405 - rollout_corr/rollout_is_seq_mean:0.9999663233757019 - rollout_corr/rollout_is_seq_std:0.001290986081585288 - rollout_corr/rollout_is_seq_max:1.004225730895996 - rollout_corr/rollout_is_seq_min:0.995941162109375 - rollout_corr/rollout_is_seq_max_deviation:0.004225730895996094 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2991151995956898) - rollout_corr/training_log_ppl:np.float64(0.2535376882078708) - rollout_corr/kl:np.float64(0.0003585750972416868) - rollout_corr/k3_kl:np.float64(0.00032570855380242847) - rollout_corr/rollout_ppl:np.float64(1.2986278631724417) - rollout_corr/rollout_log_ppl:np.float64(0.2531791118817637) - rollout_corr/log_ppl_diff:np.float64(0.00035857632610714063) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011525552763487212) - rollout_corr/log_ppl_diff_max:np.float64(0.004387974739074707) - rollout_corr/log_ppl_diff_min:np.float64(-0.006045117974281311) - rollout_corr/ppl_ratio:np.float64(1.0003596900496632) - rollout_corr/chi2_token:np.float64(0.0005872410256415606) - rollout_corr/chi2_seq:np.float64(0.2268612238112837) - actor/pg_loss:np.float64(-8.898775558918715e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(9.063289280675235e-05) - actor/ppo_kl:np.float64(3.672884257088782e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1358618326485157) - perf/mfu/actor:np.float64(0.1070155509069357) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(108.36789321899414) - actor/lr:np.float64(1e-06) - training/global_step:52 - 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:-0.0010089902207255363 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0010089902207255363 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:425.375 - response_length/max:767.0 - response_length/min:149.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:425.375 - response_length_non_aborted/max:767.0 - response_length_non_aborted/min:149.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:403.09375 - prompt_length/max:512.0 - prompt_length/min:264.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00014198385179042816 - timing_s/agent_loop/generate_sequences/min:np.float64(1.3870915044099092) - timing_s/agent_loop/generate_sequences/max:np.float64(7.209513144567609) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.068675200593134) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.209513144567609) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/s
|
||
step:53 - global_seqlen/min:21755 - global_seqlen/max:28435 - global_seqlen/minmax_diff:6680 - global_seqlen/balanced_min:24298 - global_seqlen/balanced_max:24302 - global_seqlen/mean:24300.625 - actor/entropy:0.24457094073295593 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5994806289672852 - training/rollout_probs_diff_mean:0.0024542806204408407 - training/rollout_probs_diff_std:0.007331932429224253 - training/rollout_actor_probs_pearson_corr:0.9994651675224304 - rollout_corr/rollout_is_mean:1.000049114227295 - rollout_corr/rollout_is_ratio_fraction_high:2.151717671949882e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.587857246398926 - rollout_corr/rollout_is_min:0.7702428102493286 - rollout_corr/rollout_is_std:0.02378586307168007 - rollout_corr/rollout_is_eff_sample_size:0.9994346716979495 - rollout_corr/rollout_is_seq_mean:1.00004243850708 - rollout_corr/rollout_is_seq_std:0.0014012084575369954 - rollout_corr/rollout_is_seq_max:1.005181908607483 - rollout_corr/rollout_is_seq_min:0.9958944916725159 - rollout_corr/rollout_is_seq_max_deviation:0.00518190860748291 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.289477339014411) - rollout_corr/training_log_ppl:np.float64(0.24894744424091186) - rollout_corr/kl:np.float64(0.0003029550313176088) - rollout_corr/k3_kl:np.float64(0.0003627692941421401) - rollout_corr/rollout_ppl:np.float64(1.289075882639736) - rollout_corr/rollout_log_ppl:np.float64(0.2486444886744721) - rollout_corr/log_ppl_diff:np.float64(0.0003029555664397776) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011978519905824214) - rollout_corr/log_ppl_diff_max:np.float64(0.0045301467180252075) - rollout_corr/log_ppl_diff_min:np.float64(-0.004763618111610413) - rollout_corr/ppl_ratio:np.float64(1.0003041608724743) - rollout_corr/chi2_token:np.float64(0.0008704829961061478) - rollout_corr/chi2_seq:np.float64(0.24797223578207195) - actor/pg_loss:np.float64(-0.00013095582835376263) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(5.983969549561152e-05) - actor/ppo_kl:np.float64(4.8199360510814415e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1310713840648532) - perf/mfu/actor:np.float64(0.09757008713741944) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.20830535888672) - actor/lr:np.float64(1e-06) - training/global_step:53 - training/epoch:2 - critic/score/mean:0.640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0020118560642004013 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0020118560642004013 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:363.08203125 - response_length/max:798.0 - response_length/min:106.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:363.08203125 - response_length_non_aborted/max:798.0 - response_length_non_aborted/min:106.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:396.3125 - prompt_length/max:502.0 - prompt_length/min:262.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00022395513951778412 - timing_s/agent_loop/generate_sequences/min:np.float64(1.381840230897069) - timing_s/agent_loop/generate_sequences/max:np.float64(6.928327467292547) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.167328959978477) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.928327467292547) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent
|
||
step:54 - global_seqlen/min:21771 - global_seqlen/max:29996 - global_seqlen/minmax_diff:8225 - global_seqlen/balanced_min:26633 - global_seqlen/balanced_max:26640 - global_seqlen/mean:26637.625 - actor/entropy:0.22380009293556213 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.22949397563934326 - training/rollout_probs_diff_mean:0.0022567082196474075 - training/rollout_probs_diff_std:0.006998219061642885 - training/rollout_actor_probs_pearson_corr:0.9995344877243042 - rollout_corr/rollout_is_mean:1.0000628232955933 - rollout_corr/rollout_is_ratio_fraction_high:9.23284278542269e-06 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.167020082473755 - rollout_corr/rollout_is_min:0.5657718777656555 - rollout_corr/rollout_is_std:0.02301405742764473 - rollout_corr/rollout_is_eff_sample_size:0.9994706335038355 - rollout_corr/rollout_is_seq_mean:1.0000866651535034 - rollout_corr/rollout_is_seq_std:0.0013085813261568546 - rollout_corr/rollout_is_seq_max:1.0036485195159912 - rollout_corr/rollout_is_seq_min:0.9966868162155151 - rollout_corr/rollout_is_seq_max_deviation:0.003648519515991211 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2876649326644838) - rollout_corr/training_log_ppl:np.float64(0.24327065303805284) - rollout_corr/kl:np.float64(0.00037481868857369705) - rollout_corr/k3_kl:np.float64(0.0003726331062168242) - rollout_corr/rollout_ppl:np.float64(1.2871439228765666) - rollout_corr/rollout_log_ppl:np.float64(0.24289583274367033) - rollout_corr/log_ppl_diff:np.float64(0.0003748202943825163) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011444683550507762) - rollout_corr/log_ppl_diff_max:np.float64(0.006301254034042358) - rollout_corr/log_ppl_diff_min:np.float64(-0.0038700103759765625) - rollout_corr/ppl_ratio:np.float64(1.0003759877290577) - rollout_corr/chi2_token:np.float64(0.0007948835846036673) - rollout_corr/chi2_seq:np.float64(0.26042517158202827) - actor/pg_loss:np.float64(-0.0001082469243556261) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(7.538802492490504e-05) - actor/ppo_kl:np.float64(0.0001657101322258825) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.11135829612612724) - perf/mfu/actor:np.float64(0.1082633724102717) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.19610595703125) - actor/lr:np.float64(1e-06) - training/global_step:54 - training/epoch:2 - critic/score/mean:0.54296875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.54296875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.00029083454865030944 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00029083454865030944 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:423.08203125 - response_length/max:969.0 - response_length/min:81.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:423.08203125 - response_length_non_aborted/max:969.0 - response_length_non_aborted/min:81.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:409.34375 - prompt_length/max:524.0 - prompt_length/min:264.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00015332549810409546 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0272562000900507) - timing_s/agent_loop/generate_sequences/max:np.float64(8.707748917862773) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.7737743248508195) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.707748917862773) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0)
|
||
step:55 - global_seqlen/min:22757 - global_seqlen/max:30719 - global_seqlen/minmax_diff:7962 - global_seqlen/balanced_min:26797 - global_seqlen/balanced_max:26801 - global_seqlen/mean:26799.75 - actor/entropy:0.2322537750005722 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3016374111175537 - training/rollout_probs_diff_mean:0.002216401044279337 - training/rollout_probs_diff_std:0.00684860534965992 - training/rollout_actor_probs_pearson_corr:0.9995991587638855 - rollout_corr/rollout_is_mean:0.9999573826789856 - rollout_corr/rollout_is_ratio_fraction_high:9.197432518703863e-06 - rollout_corr/rollout_is_ratio_fraction_low:1.8394865037407726e-05 - rollout_corr/rollout_is_max:2.096909761428833 - rollout_corr/rollout_is_min:0.3293769061565399 - rollout_corr/rollout_is_std:0.02277715504169464 - rollout_corr/rollout_is_eff_sample_size:0.9994813510988554 - rollout_corr/rollout_is_seq_mean:0.9999474883079529 - rollout_corr/rollout_is_seq_std:0.0012583937495946884 - rollout_corr/rollout_is_seq_max:1.006894826889038 - rollout_corr/rollout_is_seq_min:0.9962959885597229 - rollout_corr/rollout_is_seq_max_deviation:0.006894826889038086 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2898584022186697) - rollout_corr/training_log_ppl:np.float64(0.24666903459728928) - rollout_corr/kl:np.float64(0.0002695444498144184) - rollout_corr/k3_kl:np.float64(0.00033902103465521805) - rollout_corr/rollout_ppl:np.float64(1.2895028642378747) - rollout_corr/rollout_log_ppl:np.float64(0.24639948902040487) - rollout_corr/log_ppl_diff:np.float64(0.0002695455768844113) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011074139329139143) - rollout_corr/log_ppl_diff_max:np.float64(0.004107072949409485) - rollout_corr/log_ppl_diff_min:np.float64(-0.004751652479171753) - rollout_corr/ppl_ratio:np.float64(1.0002705552615225) - rollout_corr/chi2_token:np.float64(0.0008234675042331219) - rollout_corr/chi2_seq:np.float64(0.48434280999936163) - actor/pg_loss:np.float64(4.3725594878196716e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(4.63317041976552e-05) - actor/ppo_kl:np.float64(-7.033058565752981e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1018903236836195) - perf/mfu/actor:np.float64(0.10775877878208463) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.22493743896484) - actor/lr:np.float64(1e-06) - training/global_step:55 - training/epoch:2 - 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:-0.0023579916451126337 - critic/advantages/max:0.25 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0023579916451126337 - critic/returns/max:0.25 - critic/returns/min:-0.875 - response_length/mean:424.7109375 - response_length/max:954.0 - response_length/min:162.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:424.7109375 - response_length_non_aborted/max:954.0 - response_length_non_aborted/min:162.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:412.78125 - prompt_length/max:519.0 - prompt_length/min:271.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00013876333832740784 - timing_s/agent_loop/generate_sequences/min:np.float64(2.095417447388172) - timing_s/agent_loop/generate_sequences/max:np.float64(8.385497575625777) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.71868045329029) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.385497575625777) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timi
|
||
step:56 - global_seqlen/min:18837 - global_seqlen/max:31049 - global_seqlen/minmax_diff:12212 - global_seqlen/balanced_min:25345 - global_seqlen/balanced_max:25347 - global_seqlen/mean:25346.0 - actor/entropy:0.25467386841773987 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24491804838180542 - training/rollout_probs_diff_mean:0.002417678711935878 - training/rollout_probs_diff_std:0.007147489581257105 - training/rollout_actor_probs_pearson_corr:0.9995514154434204 - rollout_corr/rollout_is_mean:0.9999701976776123 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5092898607254028 - rollout_corr/rollout_is_min:0.5352914929389954 - rollout_corr/rollout_is_std:0.023758282884955406 - rollout_corr/rollout_is_eff_sample_size:0.999435862444797 - rollout_corr/rollout_is_seq_mean:0.9999875426292419 - rollout_corr/rollout_is_seq_std:0.0014447674620896578 - rollout_corr/rollout_is_seq_max:1.004019021987915 - rollout_corr/rollout_is_seq_min:0.9958305358886719 - rollout_corr/rollout_is_seq_max_deviation:0.004169464111328125 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3189649819396436) - rollout_corr/training_log_ppl:np.float64(0.26553016485559056) - rollout_corr/kl:np.float64(0.00045735904856680776) - rollout_corr/k3_kl:np.float64(0.0003336688204953475) - rollout_corr/rollout_ppl:np.float64(1.3182676555588841) - rollout_corr/rollout_log_ppl:np.float64(0.2650728039225214) - rollout_corr/log_ppl_diff:np.float64(0.0004573609330691397) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012234431924298406) - rollout_corr/log_ppl_diff_max:np.float64(0.005470991134643555) - rollout_corr/log_ppl_diff_min:np.float64(-0.003992438316345215) - rollout_corr/ppl_ratio:np.float64(1.0004586917348206) - rollout_corr/chi2_token:np.float64(0.0004137852229177952) - rollout_corr/chi2_seq:np.float64(0.20288011711090803) - actor/pg_loss:np.float64(-4.3228152208030224e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(2.432743940516957e-05) - actor/ppo_kl:np.float64(0.00014076509722471542) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.06149144750088453) - perf/mfu/actor:np.float64(0.10336162570898155) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.06003189086914) - actor/lr:np.float64(1e-06) - training/global_step:56 - 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:-0.0024021996650844812 - critic/advantages/max:0.5 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0024021996650844812 - critic/returns/max:0.5 - critic/returns/min:-0.875 - response_length/mean:410.59375 - response_length/max:1051.0 - response_length/min:141.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:410.59375 - response_length_non_aborted/max:1051.0 - response_length_non_aborted/min:141.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:381.46875 - prompt_length/max:480.0 - prompt_length/min:261.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00011040642857551575 - timing_s/agent_loop/generate_sequences/min:np.float64(1.2151698395609856) - timing_s/agent_loop/generate_sequences/max:np.float64(8.617751153185964) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.9353341244204785) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.617751153185964) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slow
|
||
step:57 - global_seqlen/min:20714 - global_seqlen/max:27341 - global_seqlen/minmax_diff:6627 - global_seqlen/balanced_min:24062 - global_seqlen/balanced_max:24063 - global_seqlen/mean:24062.75 - actor/entropy:0.247794970870018 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.30163609981536865 - training/rollout_probs_diff_mean:0.0026478164363652468 - training/rollout_probs_diff_std:0.00758332060649991 - training/rollout_actor_probs_pearson_corr:0.9995037913322449 - rollout_corr/rollout_is_mean:0.9999039769172668 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.2746400645701215e-05 - rollout_corr/rollout_is_max:1.474251627922058 - rollout_corr/rollout_is_min:0.4279352128505707 - rollout_corr/rollout_is_std:0.025090720504522324 - rollout_corr/rollout_is_eff_sample_size:0.9993709708778578 - rollout_corr/rollout_is_seq_mean:0.9998193979263306 - rollout_corr/rollout_is_seq_std:0.00137531163636595 - rollout_corr/rollout_is_seq_max:1.0043578147888184 - rollout_corr/rollout_is_seq_min:0.9959853291511536 - rollout_corr/rollout_is_seq_max_deviation:0.004357814788818359 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2967904824763536) - rollout_corr/training_log_ppl:np.float64(0.2543451767269289) - rollout_corr/kl:np.float64(0.000545830731464747) - rollout_corr/k3_kl:np.float64(0.0003583562258455686) - rollout_corr/rollout_ppl:np.float64(1.296079382300377) - rollout_corr/rollout_log_ppl:np.float64(0.25379934372904245) - rollout_corr/log_ppl_diff:np.float64(0.000545832997886464) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012242147058714181) - rollout_corr/log_ppl_diff_max:np.float64(0.0052535682916641235) - rollout_corr/log_ppl_diff_min:np.float64(-0.00345398485660553) - rollout_corr/ppl_ratio:np.float64(1.0005470658652484) - rollout_corr/chi2_token:np.float64(0.0003389487974345684) - rollout_corr/chi2_seq:np.float64(0.048471454065293074) - actor/pg_loss:np.float64(-4.220055416226387e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(2.6725049792730715e-05) - actor/ppo_kl:np.float64(3.711784023319176e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.08370932471007109) - perf/mfu/actor:np.float64(0.09839206597889999) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(105.99935150146484) - actor/lr:np.float64(1e-06) - training/global_step:57 - training/epoch:2 - critic/score/mean:0.7265625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7265625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0006781270494684577 - critic/advantages/max:0.875 - critic/advantages/min:-0.625 - critic/returns/mean:-0.0006781270494684577 - critic/returns/max:0.875 - critic/returns/min:-0.625 - response_length/mean:343.4609375 - response_length/max:884.0 - response_length/min:165.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:343.4609375 - response_length_non_aborted/max:884.0 - response_length_non_aborted/min:165.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:408.5 - prompt_length/max:499.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:7.432699203491211e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.1311710253357887) - timing_s/agent_loop/generate_sequences/max:np.float64(7.473232988268137) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.930545251489093) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.473232988268137) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent
|
||
step:58 - global_seqlen/min:23061 - global_seqlen/max:28276 - global_seqlen/minmax_diff:5215 - global_seqlen/balanced_min:25176 - global_seqlen/balanced_max:25182 - global_seqlen/mean:25179.875 - actor/entropy:0.21638013422489166 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5083588361740112 - training/rollout_probs_diff_mean:0.002310627605766058 - training/rollout_probs_diff_std:0.007523712236434221 - training/rollout_actor_probs_pearson_corr:0.9994233250617981 - rollout_corr/rollout_is_mean:1.0000600814819336 - rollout_corr/rollout_is_ratio_fraction_high:9.657077498559374e-06 - rollout_corr/rollout_is_ratio_fraction_low:1.931415499711875e-05 - rollout_corr/rollout_is_max:2.903904676437378 - rollout_corr/rollout_is_min:0.3459939956665039 - rollout_corr/rollout_is_std:0.0234451275318861 - rollout_corr/rollout_is_eff_sample_size:0.999450627941437 - rollout_corr/rollout_is_seq_mean:1.0000741481781006 - rollout_corr/rollout_is_seq_std:0.0013978402130305767 - rollout_corr/rollout_is_seq_max:1.0069079399108887 - rollout_corr/rollout_is_seq_min:0.9964627623558044 - rollout_corr/rollout_is_seq_max_deviation:0.006907939910888672 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2778812977485359) - rollout_corr/training_log_ppl:np.float64(0.23755318802432157) - rollout_corr/kl:np.float64(0.0002802361841816037) - rollout_corr/k3_kl:np.float64(0.00033016265430774183) - rollout_corr/rollout_ppl:np.float64(1.2775348122231662) - rollout_corr/rollout_log_ppl:np.float64(0.23727294862328563) - rollout_corr/log_ppl_diff:np.float64(0.00028023940103594214) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010713046794990078) - rollout_corr/log_ppl_diff_max:np.float64(0.005381345748901367) - rollout_corr/log_ppl_diff_min:np.float64(-0.007450282573699951) - rollout_corr/ppl_ratio:np.float64(1.0002813362516463) - rollout_corr/chi2_token:np.float64(0.0007960675284266472) - rollout_corr/chi2_seq:np.float64(0.3917784618679434) - actor/pg_loss:np.float64(-5.9889862313866615e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(8.172775687853573e-05) - actor/ppo_kl:np.float64(4.416867541578995e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.12709747441112995) - perf/mfu/actor:np.float64(0.10145917324682335) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.06658554077148) - actor/lr:np.float64(1e-06) - training/global_step:58 - training/epoch:2 - critic/score/mean:0.796875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.796875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0005021680262871087 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0005021680262871087 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:404.49609375 - response_length/max:832.0 - response_length/min:116.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:404.49609375 - response_length_non_aborted/max:832.0 - response_length_non_aborted/min:116.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:382.375 - prompt_length/max:482.0 - prompt_length/min:263.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00010803155601024628 - timing_s/agent_loop/generate_sequences/min:np.float64(1.5115187149494886) - timing_s/agent_loop/generate_sequences/max:np.float64(7.589536972343922) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.612800014750974) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.589536972343922) - timing_s/agent_loop/slowest/tool_calls:np.floa
|
||
step:59 - global_seqlen/min:20172 - global_seqlen/max:28781 - global_seqlen/minmax_diff:8609 - global_seqlen/balanced_min:23349 - global_seqlen/balanced_max:23358 - global_seqlen/mean:23354.375 - actor/entropy:0.24409091472625732 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2924416661262512 - training/rollout_probs_diff_mean:0.002486287383362651 - training/rollout_probs_diff_std:0.007413405925035477 - training/rollout_actor_probs_pearson_corr:0.9995101094245911 - rollout_corr/rollout_is_mean:0.9999902248382568 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.23826318688225e-05 - rollout_corr/rollout_is_max:1.4768394231796265 - rollout_corr/rollout_is_min:0.24649836122989655 - rollout_corr/rollout_is_std:0.024291684851050377 - rollout_corr/rollout_is_eff_sample_size:0.9994105001503535 - rollout_corr/rollout_is_seq_mean:1.0000391006469727 - rollout_corr/rollout_is_seq_std:0.001515738433226943 - rollout_corr/rollout_is_seq_max:1.005221962928772 - rollout_corr/rollout_is_seq_min:0.9950911402702332 - rollout_corr/rollout_is_seq_max_deviation:0.005221962928771973 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.310080023482442) - rollout_corr/training_log_ppl:np.float64(0.26051520198234357) - rollout_corr/kl:np.float64(0.000524200898032956) - rollout_corr/k3_kl:np.float64(0.0004534090181920192) - rollout_corr/rollout_ppl:np.float64(1.309392552357167) - rollout_corr/rollout_log_ppl:np.float64(0.2599910004355479) - rollout_corr/log_ppl_diff:np.float64(0.0005242015467956662) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013801687164232135) - rollout_corr/log_ppl_diff_max:np.float64(0.006640344858169556) - rollout_corr/log_ppl_diff_min:np.float64(-0.005571231245994568) - rollout_corr/ppl_ratio:np.float64(1.0005259348545223) - rollout_corr/chi2_token:np.float64(0.0007969602011144161) - rollout_corr/chi2_seq:np.float64(0.15531086595728993) - actor/pg_loss:np.float64(-2.4826964363455772e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(8.681123972564819e-05) - actor/ppo_kl:np.float64(0.0002450501577317965) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.13004800118505955) - perf/mfu/actor:np.float64(0.0959988976426908) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.06728744506836) - actor/lr:np.float64(1e-06) - training/global_step:59 - training/epoch:2 - critic/score/mean:0.71484375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.71484375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.001934698666445911 - critic/advantages/max:0.375 - critic/advantages/min:-0.875 - critic/returns/mean:-0.001934698666445911 - critic/returns/max:0.375 - critic/returns/min:-0.875 - response_length/mean:349.04296875 - response_length/max:790.0 - response_length/min:136.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:349.04296875 - response_length_non_aborted/max:790.0 - response_length_non_aborted/min:136.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:380.78125 - prompt_length/max:518.0 - prompt_length/min:197.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.0001479554921388626 - timing_s/agent_loop/generate_sequences/min:np.float64(1.7181454841047525) - timing_s/agent_loop/generate_sequences/max:np.float64(7.045558724552393) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.890985884645488) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.045558724552393) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s
|
||
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 crystal structure is described as having lattice parameters a = 7.82 Å, b = 7.82 Å, c = 17.88 Å, with all angles equal to 90 degrees, which indicates a cubic or orthorhombic structure. Since all angles are 90 degrees, this is an orthorhombic unit cell.
|
||
|
||
The volume of a unit cell in an orthorhombic structure is calculated using the formula:
|
||
|
||
$$ V = a \times b \times c $$
|
||
|
||
Substituting the given values:
|
||
|
||
$$ V = 7.82 \, \text{Å} \times 7.82 \, \text{Å} \times 17.88 \, \text{Å} $$
|
||
|
||
First, calculate the product of a and b:
|
||
|
||
$$ 7.82 \times 7.82 = 61.1524 \, \text{Å}^2 $$
|
||
|
||
Then multiply by c:
|
||
|
||
$$ 61.1524 \times 17.88 = 1094.31 \, \text{Å}^3 $$
|
||
|
||
Thus, the volume of the unit cell is 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-GRPO-Qwen-Qwen3-8B-mbs8-train32-rollout8-lr1e-6-vllm0.8/global_step_60
|
||
step:60 - global_seqlen/min:20680 - global_seqlen/max:28400 - global_seqlen/minmax_diff:7720 - global_seqlen/balanced_min:23980 - global_seqlen/balanced_max:23992 - global_seqlen/mean:23988.5 - actor/entropy:0.2360708862543106 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.48631659150123596 - training/rollout_probs_diff_mean:0.002380544785410166 - training/rollout_probs_diff_std:0.007412472274154425 - training/rollout_actor_probs_pearson_corr:0.9995028972625732 - rollout_corr/rollout_is_mean:0.9999828338623047 - rollout_corr/rollout_is_ratio_fraction_high:2.126483195752371e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:4.63820219039917 - rollout_corr/rollout_is_min:0.5089420080184937 - rollout_corr/rollout_is_std:0.0243162102997303 - rollout_corr/rollout_is_eff_sample_size:0.9994089522585684 - rollout_corr/rollout_is_seq_mean:0.9999392032623291 - rollout_corr/rollout_is_seq_std:0.0015173176070675254 - rollout_corr/rollout_is_seq_max:1.0061588287353516 - rollout_corr/rollout_is_seq_min:0.9955165982246399 - rollout_corr/rollout_is_seq_max_deviation:0.0061588287353515625 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2967511527240276) - rollout_corr/training_log_ppl:np.float64(0.2504257837572368) - rollout_corr/kl:np.float64(0.0005130456188339183) - rollout_corr/k3_kl:np.float64(0.00039365157081761026) - rollout_corr/rollout_ppl:np.float64(1.2960526859387755) - rollout_corr/rollout_log_ppl:np.float64(0.24991273554041982) - rollout_corr/log_ppl_diff:np.float64(0.0005130482168169692) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012746355059789494) - rollout_corr/log_ppl_diff_max:np.float64(0.007074087858200073) - rollout_corr/log_ppl_diff_min:np.float64(-0.004114046692848206) - rollout_corr/ppl_ratio:np.float64(1.000514515209943) - rollout_corr/chi2_token:np.float64(0.0005633197724819183) - rollout_corr/chi2_seq:np.float64(0.19360803021118045) - actor/pg_loss:np.float64(2.6261783204972744e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(8.017660366022028e-05) - actor/ppo_kl:np.float64(0.00012672674524605299) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.10184402484446764) - perf/mfu/actor:np.float64(0.09759851413831924) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.0626106262207) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7533244680851063) - val-aux/sciknoweval/reward/std@16:np.float64(0.05264515376564715) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7701702127659574) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.03719160916663892) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.736276595744681) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.048824922865621616) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7530531914893616) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.05238510513576182) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7838297872340426) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.03035824019330265) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.7190744680851064) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.047430696791599204) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.755372340425532) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.03210443043930431) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7968510638297872) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.02641430219640358) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6982340425531915) - val-aux/sciknoweval/reward/worst@8/std:np.float64(0.04790582542211792) - val-aux/sciknoweval/reward/maj@8/mean:np.float64(0.7554361702127659) - val-aux/sciknoweval/reward/maj@8/std:np.float64(0.01487437652497796) - val-aux/sciknoweval/reward/best@16/mean:np.float64(0.808627659574468) - val-aux/sciknowe
|
||
step:61 - global_seqlen/min:20712 - global_seqlen/max:26801 - global_seqlen/minmax_diff:6089 - global_seqlen/balanced_min:23429 - global_seqlen/balanced_max:23433 - global_seqlen/mean:23431.0 - actor/entropy:0.28184887766838074 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18471911549568176 - training/rollout_probs_diff_mean:0.0026123705320060253 - training/rollout_probs_diff_std:0.007209790870547295 - training/rollout_actor_probs_pearson_corr:0.9995990991592407 - rollout_corr/rollout_is_mean:0.9999793767929077 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4511632919311523 - rollout_corr/rollout_is_min:0.6871916055679321 - rollout_corr/rollout_is_std:0.024666661396622658 - rollout_corr/rollout_is_eff_sample_size:0.9993919257653825 - rollout_corr/rollout_is_seq_mean:0.9999731183052063 - rollout_corr/rollout_is_seq_std:0.0016377248102799058 - rollout_corr/rollout_is_seq_max:1.0047904253005981 - rollout_corr/rollout_is_seq_min:0.9949299097061157 - rollout_corr/rollout_is_seq_max_deviation:0.005070090293884277 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.352380177937448) - rollout_corr/training_log_ppl:np.float64(0.2933591184410034) - rollout_corr/kl:np.float64(0.0003880567859333439) - rollout_corr/k3_kl:np.float64(0.0003783676123703117) - rollout_corr/rollout_ppl:np.float64(1.351829611696303) - rollout_corr/rollout_log_ppl:np.float64(0.2929710615426302) - rollout_corr/log_ppl_diff:np.float64(0.00038805689837317914) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013752124359598383) - rollout_corr/log_ppl_diff_max:np.float64(0.0052146315574646) - rollout_corr/log_ppl_diff_min:np.float64(-0.004297047853469849) - rollout_corr/ppl_ratio:np.float64(1.0003896711859852) - rollout_corr/chi2_token:np.float64(0.0007408889941871166) - rollout_corr/chi2_seq:np.float64(0.2107625617645681) - actor/pg_loss:np.float64(-2.980930730700493e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(2.4723101887502708e-05) - actor/ppo_kl:np.float64(4.149326706581746e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.05792093137279153) - perf/mfu/actor:np.float64(0.09517745982226297) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(107.07573699951172) - actor/lr:np.float64(1e-06) - training/global_step:61 - training/epoch:2 - critic/score/mean:0.8046875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.8046875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00027224450604990125 - critic/advantages/max:0.125 - critic/advantages/min:-0.875 - critic/returns/mean:0.00027224450604990125 - critic/returns/max:0.125 - critic/returns/min:-0.875 - response_length/mean:319.25 - response_length/max:737.0 - response_length/min:124.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:319.25 - response_length_non_aborted/max:737.0 - response_length_non_aborted/min:124.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:412.96875 - prompt_length/max:509.0 - prompt_length/min:269.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00014871545135974884 - timing_s/agent_loop/generate_sequences/min:np.float64(1.4992734659463167) - timing_s/agent_loop/generate_sequences/max:np.float64(6.306750409305096) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.624103704605659) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.306750409305096) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/prompt_l
|
||
step:62 - global_seqlen/min:18620 - global_seqlen/max:28360 - global_seqlen/minmax_diff:9740 - global_seqlen/balanced_min:23900 - global_seqlen/balanced_max:23903 - global_seqlen/mean:23901.875 - actor/entropy:0.22449631989002228 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.34021374583244324 - training/rollout_probs_diff_mean:0.0022299615666270256 - training/rollout_probs_diff_std:0.007056043017655611 - training/rollout_actor_probs_pearson_corr:0.9995299577713013 - rollout_corr/rollout_is_mean:0.999995768070221 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.1769656086689793e-05 - rollout_corr/rollout_is_max:1.5585025548934937 - rollout_corr/rollout_is_min:0.09886863827705383 - rollout_corr/rollout_is_std:0.022548340260982513 - rollout_corr/rollout_is_eff_sample_size:0.9994918307473496 - rollout_corr/rollout_is_seq_mean:1.0000267028808594 - rollout_corr/rollout_is_seq_std:0.00136755988933146 - rollout_corr/rollout_is_seq_max:1.003589153289795 - rollout_corr/rollout_is_seq_min:0.9956575632095337 - rollout_corr/rollout_is_seq_max_deviation:0.004342436790466309 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2885832339525223) - rollout_corr/training_log_ppl:np.float64(0.24286671256413683) - rollout_corr/kl:np.float64(0.0003957929533555671) - rollout_corr/k3_kl:np.float64(0.000319797927829768) - rollout_corr/rollout_ppl:np.float64(1.2880218625068665) - rollout_corr/rollout_log_ppl:np.float64(0.24247091660799924) - rollout_corr/log_ppl_diff:np.float64(0.0003957959561375901) - rollout_corr/log_ppl_abs_diff:np.float64(0.001179586790385656) - rollout_corr/log_ppl_diff_max:np.float64(0.005161285400390625) - rollout_corr/log_ppl_diff_min:np.float64(-0.004351675510406494) - rollout_corr/ppl_ratio:np.float64(1.0003969802055508) - rollout_corr/chi2_token:np.float64(0.00047266576439142227) - rollout_corr/chi2_seq:np.float64(0.14369945717044175) - actor/pg_loss:np.float64(3.2330979593098164e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(6.0179311731189955e-05) - actor/ppo_kl:np.float64(0.00012892496328453795) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.08937415108084679) - perf/mfu/actor:np.float64(0.09680483357257347) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.24963760375977) - actor/lr:np.float64(1e-06) - training/global_step:62 - 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:1.4966638445912395e-05 - critic/advantages/max:0.125 - critic/advantages/min:-0.875 - critic/returns/mean:1.4966638445912395e-05 - critic/returns/max:0.125 - critic/returns/min:-0.875 - response_length/mean:358.87109375 - response_length/max:988.0 - response_length/min:122.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:358.87109375 - response_length_non_aborted/max:988.0 - response_length_non_aborted/min:122.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:388.0625 - prompt_length/max:541.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00012834742665290833 - timing_s/agent_loop/generate_sequences/min:np.float64(1.3620898555964231) - timing_s/agent_loop/generate_sequences/max:np.float64(8.082440830767155) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.747333021972736) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.082440830767155) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - t
|
||
step:63 - global_seqlen/min:19818 - global_seqlen/max:26499 - global_seqlen/minmax_diff:6681 - global_seqlen/balanced_min:23404 - global_seqlen/balanced_max:23410 - global_seqlen/mean:23408.125 - actor/entropy:0.27887508273124695 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5032555460929871 - training/rollout_probs_diff_mean:0.002610958879813552 - training/rollout_probs_diff_std:0.007603233214467764 - training/rollout_actor_probs_pearson_corr:0.9995499849319458 - rollout_corr/rollout_is_mean:0.9998920559883118 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:3.616157118813135e-05 - rollout_corr/rollout_is_max:1.424423098564148 - rollout_corr/rollout_is_min:0.4286360740661621 - rollout_corr/rollout_is_std:0.02481122314929962 - rollout_corr/rollout_is_eff_sample_size:0.9993846628923608 - rollout_corr/rollout_is_seq_mean:0.9998520016670227 - rollout_corr/rollout_is_seq_std:0.0016024266369640827 - rollout_corr/rollout_is_seq_max:1.0046424865722656 - rollout_corr/rollout_is_seq_min:0.994684636592865 - rollout_corr/rollout_is_seq_max_deviation:0.00531536340713501 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.360725431703031) - rollout_corr/training_log_ppl:np.float64(0.2997116361220833) - rollout_corr/kl:np.float64(0.0005157101149642074) - rollout_corr/k3_kl:np.float64(0.00036520069393475296) - rollout_corr/rollout_ppl:np.float64(1.3599829538725317) - rollout_corr/rollout_log_ppl:np.float64(0.2991959253558889) - rollout_corr/log_ppl_diff:np.float64(0.0005157107661943883) - rollout_corr/log_ppl_abs_diff:np.float64(0.001327492733253166) - rollout_corr/log_ppl_diff_max:np.float64(0.0076059699058532715) - rollout_corr/log_ppl_diff_min:np.float64(-0.0042442381381988525) - rollout_corr/ppl_ratio:np.float64(1.0005172591190785) - rollout_corr/chi2_token:np.float64(0.00040490622632205486) - rollout_corr/chi2_seq:np.float64(0.13411166868172586) - actor/pg_loss:np.float64(2.5239423848688602e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(2.4135695639415644e-05) - actor/ppo_kl:np.float64(3.2647710623834314e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.0763076157309115) - perf/mfu/actor:np.float64(0.09596526124371177) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.12528991699219) - actor/lr:np.float64(1e-06) - training/global_step:63 - training/epoch:2 - critic/score/mean:0.71484375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.71484375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0038225792814046144 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0038225792814046144 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:324.06640625 - response_length/max:997.0 - response_length/min:100.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:324.06640625 - response_length_non_aborted/max:997.0 - response_length_non_aborted/min:100.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:407.4375 - prompt_length/max:483.0 - prompt_length/min:264.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:8.467212319374084e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.324038576334715) - timing_s/agent_loop/generate_sequences/max:np.float64(7.843218516558409) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.6979008383350447) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.843218516558409) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timi
|
||
step:64 - global_seqlen/min:20169 - global_seqlen/max:25066 - global_seqlen/minmax_diff:4897 - global_seqlen/balanced_min:23043 - global_seqlen/balanced_max:23048 - global_seqlen/mean:23046.25 - actor/entropy:0.2526596784591675 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.611835241317749 - training/rollout_probs_diff_mean:0.00250222091563046 - training/rollout_probs_diff_std:0.007975784130394459 - training/rollout_actor_probs_pearson_corr:0.9994510412216187 - rollout_corr/rollout_is_mean:0.9999549388885498 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:4.7495785111095756e-05 - rollout_corr/rollout_is_max:1.5585436820983887 - rollout_corr/rollout_is_min:0.24953652918338776 - rollout_corr/rollout_is_std:0.024059930816292763 - rollout_corr/rollout_is_eff_sample_size:0.9994213355271035 - rollout_corr/rollout_is_seq_mean:0.999975323677063 - rollout_corr/rollout_is_seq_std:0.001513405004516244 - rollout_corr/rollout_is_seq_max:1.004136085510254 - rollout_corr/rollout_is_seq_min:0.9950628876686096 - rollout_corr/rollout_is_seq_max_deviation:0.004937112331390381 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3078968808986247) - rollout_corr/training_log_ppl:np.float64(0.2614987124252366) - rollout_corr/kl:np.float64(0.0003369308145106942) - rollout_corr/k3_kl:np.float64(0.00036924221276990465) - rollout_corr/rollout_ppl:np.float64(1.3074609604664147) - rollout_corr/rollout_log_ppl:np.float64(0.2611617771326564) - rollout_corr/log_ppl_diff:np.float64(0.00033693529258016497) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012274637265363708) - rollout_corr/log_ppl_diff_max:np.float64(0.009813323616981506) - rollout_corr/log_ppl_diff_min:np.float64(-0.003359168767929077) - rollout_corr/ppl_ratio:np.float64(1.0003384032752365) - rollout_corr/chi2_token:np.float64(0.000732076819986105) - rollout_corr/chi2_seq:np.float64(0.19428053102456033) - actor/pg_loss:np.float64(-5.030236206948757e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(2.072281131404452e-05) - actor/ppo_kl:np.float64(-8.341730296290706e-06) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.0369609696790576) - perf/mfu/actor:np.float64(0.09428030451113695) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.12153482437134) - actor/lr:np.float64(1e-06) - training/global_step:64 - training/epoch:2 - critic/score/mean:0.80859375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.80859375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0006679094512946904 - critic/advantages/max:0.125 - critic/advantages/min:-0.875 - critic/returns/mean:0.0006679094512946904 - critic/returns/max:0.125 - critic/returns/min:-0.875 - response_length/mean:328.9765625 - response_length/max:976.0 - response_length/min:105.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:328.9765625 - response_length_non_aborted/max:976.0 - response_length_non_aborted/min:105.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:391.21875 - prompt_length/max:473.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00012168474495410919 - timing_s/agent_loop/generate_sequences/min:np.float64(1.3025468364357948) - timing_s/agent_loop/generate_sequences/max:np.float64(7.783195244148374) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.66632833393669) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.783195244148374) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/a
|
||
step:65 - global_seqlen/min:20002 - global_seqlen/max:28030 - global_seqlen/minmax_diff:8028 - global_seqlen/balanced_min:23359 - global_seqlen/balanced_max:23360 - global_seqlen/mean:23359.25 - actor/entropy:0.26026061177253723 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.15484067797660828 - training/rollout_probs_diff_mean:0.00245410343632102 - training/rollout_probs_diff_std:0.007108656223863363 - training/rollout_actor_probs_pearson_corr:0.9995998740196228 - rollout_corr/rollout_is_mean:0.9998462200164795 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.1785504284489434e-05 - rollout_corr/rollout_is_max:1.645729422569275 - rollout_corr/rollout_is_min:0.2046152651309967 - rollout_corr/rollout_is_std:0.024017779156565666 - rollout_corr/rollout_is_eff_sample_size:0.9994233597444102 - rollout_corr/rollout_is_seq_mean:0.9998414516448975 - rollout_corr/rollout_is_seq_std:0.0014760512858629227 - rollout_corr/rollout_is_seq_max:1.0046924352645874 - rollout_corr/rollout_is_seq_min:0.9952191114425659 - rollout_corr/rollout_is_seq_max_deviation:0.004780888557434082 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3243399313651025) - rollout_corr/training_log_ppl:np.float64(0.2717995819548378) - rollout_corr/kl:np.float64(0.0004931192334041157) - rollout_corr/k3_kl:np.float64(0.00032269406943896684) - rollout_corr/rollout_ppl:np.float64(1.3236418650485575) - rollout_corr/rollout_log_ppl:np.float64(0.2713064616546035) - rollout_corr/log_ppl_diff:np.float64(0.0004931203002342954) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012263796961633489) - rollout_corr/log_ppl_diff_max:np.float64(0.005062103271484375) - rollout_corr/log_ppl_diff_min:np.float64(-0.004357486963272095) - rollout_corr/ppl_ratio:np.float64(1.00049435836263) - rollout_corr/chi2_token:np.float64(0.0002919696271419525) - rollout_corr/chi2_seq:np.float64(0.05615922715514898) - actor/pg_loss:np.float64(6.77348580211401e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(3.6549815831676824e-05) - actor/ppo_kl:np.float64(2.9900374386215844e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.07062362832948565) - perf/mfu/actor:np.float64(0.09549129968903941) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.11310195922852) - actor/lr:np.float64(1e-06) - training/global_step:65 - 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:0.0005215085693635046 - critic/advantages/max:0.75 - critic/advantages/min:-0.625 - critic/returns/mean:0.0005215085693635046 - critic/returns/max:0.75 - critic/returns/min:-0.625 - response_length/mean:331.4453125 - response_length/max:856.0 - response_length/min:107.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:331.4453125 - response_length_non_aborted/max:856.0 - response_length_non_aborted/min:107.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:398.53125 - prompt_length/max:490.0 - prompt_length/min:189.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00010526925325393677 - timing_s/agent_loop/generate_sequences/min:np.float64(1.3068315219134092) - timing_s/agent_loop/generate_sequences/max:np.float64(7.289796233177185) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.759431266633328) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.289796233177185) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/
|
||
step:66 - global_seqlen/min:16888 - global_seqlen/max:25871 - global_seqlen/minmax_diff:8983 - global_seqlen/balanced_min:22834 - global_seqlen/balanced_max:22836 - global_seqlen/mean:22834.875 - actor/entropy:0.21960677206516266 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18740245699882507 - training/rollout_probs_diff_mean:0.002241658978164196 - training/rollout_probs_diff_std:0.006920246407389641 - training/rollout_actor_probs_pearson_corr:0.9995320439338684 - rollout_corr/rollout_is_mean:0.999957263469696 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4334044456481934 - rollout_corr/rollout_is_min:0.6489461064338684 - rollout_corr/rollout_is_std:0.02259587123990059 - rollout_corr/rollout_is_eff_sample_size:0.9994895680773683 - rollout_corr/rollout_is_seq_mean:0.9999371767044067 - rollout_corr/rollout_is_seq_std:0.0015744587872177362 - rollout_corr/rollout_is_seq_max:1.0062541961669922 - rollout_corr/rollout_is_seq_min:0.9941355586051941 - rollout_corr/rollout_is_seq_max_deviation:0.0062541961669921875 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2800852260552347) - rollout_corr/training_log_ppl:np.float64(0.2388803932844894) - rollout_corr/kl:np.float64(0.0003673631687366452) - rollout_corr/k3_kl:np.float64(0.00030391120974115893) - rollout_corr/rollout_ppl:np.float64(1.2795846569351852) - rollout_corr/rollout_log_ppl:np.float64(0.23851302760886028) - rollout_corr/log_ppl_diff:np.float64(0.0003673656756291166) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012513460969785228) - rollout_corr/log_ppl_diff_max:np.float64(0.0068174004554748535) - rollout_corr/log_ppl_diff_min:np.float64(-0.004504576325416565) - rollout_corr/ppl_ratio:np.float64(1.0003687869757414) - rollout_corr/chi2_token:np.float64(0.0004690245259553194) - rollout_corr/chi2_seq:np.float64(0.12948566675186157) - actor/pg_loss:np.float64(-8.635222911834717e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(1.3391224911174504e-05) - actor/ppo_kl:np.float64(1.975016734201418e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.06018312368541956) - perf/mfu/actor:np.float64(0.09330624794717815) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.09537124633789) - actor/lr:np.float64(1e-06) - training/global_step:66 - 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:-0.0019717176910489798 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0019717176910489798 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:325.40234375 - response_length/max:822.0 - response_length/min:106.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:325.40234375 - response_length_non_aborted/max:822.0 - response_length_non_aborted/min:106.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:388.1875 - prompt_length/max:475.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:7.301941514015198e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.2752054762095213) - timing_s/agent_loop/generate_sequences/max:np.float64(6.971778251230717) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.6238729834512924) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.971778251230717) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_l
|
||
step:67 - global_seqlen/min:21129 - global_seqlen/max:23608 - global_seqlen/minmax_diff:2479 - global_seqlen/balanced_min:22432 - global_seqlen/balanced_max:22436 - global_seqlen/mean:22434.375 - actor/entropy:0.2391292154788971 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9956695437431335 - training/rollout_probs_diff_mean:0.002481213305145502 - training/rollout_probs_diff_std:0.008158995769917965 - training/rollout_actor_probs_pearson_corr:0.9994234442710876 - rollout_corr/rollout_is_mean:1.0000455379486084 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.474420762155205e-05 - rollout_corr/rollout_is_max:1.775370717048645 - rollout_corr/rollout_is_min:0.00026000928482972085 - rollout_corr/rollout_is_std:0.024126727133989334 - rollout_corr/rollout_is_eff_sample_size:0.9994183587518422 - rollout_corr/rollout_is_seq_mean:1.0000216960906982 - rollout_corr/rollout_is_seq_std:0.0014827295672148466 - rollout_corr/rollout_is_seq_max:1.0066418647766113 - rollout_corr/rollout_is_seq_min:0.9953941106796265 - rollout_corr/rollout_is_seq_max_deviation:0.006641864776611328 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2991883950307965) - rollout_corr/training_log_ppl:np.float64(0.25347993172181305) - rollout_corr/kl:np.float64(0.0003144908936771884) - rollout_corr/k3_kl:np.float64(0.0003958175733913549) - rollout_corr/rollout_ppl:np.float64(1.2987574469298124) - rollout_corr/rollout_log_ppl:np.float64(0.25316543960070703) - rollout_corr/log_ppl_diff:np.float64(0.00031449212110601366) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012145455984864384) - rollout_corr/log_ppl_diff_max:np.float64(0.007181018590927124) - rollout_corr/log_ppl_diff_min:np.float64(-0.005029022693634033) - rollout_corr/ppl_ratio:np.float64(1.0003158359322697) - rollout_corr/chi2_token:np.float64(0.0009363754652440548) - rollout_corr/chi2_seq:np.float64(0.2373002462554723) - actor/pg_loss:np.float64(1.0611605830490589e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(4.2922226384689566e-05) - actor/ppo_kl:np.float64(-7.573282338313447e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.060480975080281496) - perf/mfu/actor:np.float64(0.09189412437163944) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.12161254882812) - actor/lr:np.float64(1e-06) - training/global_step:67 - training/epoch:2 - 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:-0.0017661177553236485 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0017661177553236485 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:315.73046875 - response_length/max:867.0 - response_length/min:97.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:315.73046875 - response_length_non_aborted/max:867.0 - response_length_non_aborted/min:97.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:385.34375 - prompt_length/max:472.0 - prompt_length/min:266.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00014377757906913757 - timing_s/agent_loop/generate_sequences/min:np.float64(1.256573947146535) - timing_s/agent_loop/generate_sequences/max:np.float64(7.064186401665211) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.4151445413008332) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.064186401665211) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - ti
|
||
step:68 - global_seqlen/min:17478 - global_seqlen/max:24987 - global_seqlen/minmax_diff:7509 - global_seqlen/balanced_min:21376 - global_seqlen/balanced_max:21379 - global_seqlen/mean:21377.75 - actor/entropy:0.2649036645889282 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9373443126678467 - training/rollout_probs_diff_mean:0.002694986527785659 - training/rollout_probs_diff_std:0.008782610297203064 - training/rollout_actor_probs_pearson_corr:0.999350905418396 - rollout_corr/rollout_is_mean:1.0000685453414917 - rollout_corr/rollout_is_ratio_fraction_high:2.6007126507465728e-05 - rollout_corr/rollout_is_ratio_fraction_low:3.901069067069329e-05 - rollout_corr/rollout_is_max:6.416873931884766 - rollout_corr/rollout_is_min:0.0026309515815228224 - rollout_corr/rollout_is_std:0.025890419259667397 - rollout_corr/rollout_is_eff_sample_size:0.9993300161348888 - rollout_corr/rollout_is_seq_mean:0.9999976754188538 - rollout_corr/rollout_is_seq_std:0.0016944073140621185 - rollout_corr/rollout_is_seq_max:1.0089281797409058 - rollout_corr/rollout_is_seq_min:0.9944455027580261 - rollout_corr/rollout_is_seq_max_deviation:0.008928179740905762 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3318745321594179) - rollout_corr/training_log_ppl:np.float64(0.2706391986721428) - rollout_corr/kl:np.float64(0.0005419229773551137) - rollout_corr/k3_kl:np.float64(0.0004925900274201922) - rollout_corr/rollout_ppl:np.float64(1.3310793451964855) - rollout_corr/rollout_log_ppl:np.float64(0.2700972726306645) - rollout_corr/log_ppl_diff:np.float64(0.0005419260414782912) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014799507334828377) - rollout_corr/log_ppl_diff_max:np.float64(0.026928484439849854) - rollout_corr/log_ppl_diff_min:np.float64(-0.0034056007862091064) - rollout_corr/ppl_ratio:np.float64(1.0005449776072055) - rollout_corr/chi2_token:np.float64(0.000718752620741725) - rollout_corr/chi2_seq:np.float64(0.20082624815404415) - actor/pg_loss:np.float64(-4.2798928916454315e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(4.8068650357890874e-05) - actor/ppo_kl:np.float64(8.310840217617965e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1402107272297144) - perf/mfu/actor:np.float64(0.08789061359051692) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.2520523071289) - actor/lr:np.float64(1e-06) - training/global_step:68 - training/epoch:2 - 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:-0.0004600010288413614 - critic/advantages/max:0.625 - critic/advantages/min:-0.75 - critic/returns/mean:-0.0004600010288413614 - critic/returns/max:0.625 - critic/returns/min:-0.75 - response_length/mean:300.3984375 - response_length/max:758.0 - response_length/min:98.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:300.3984375 - response_length_non_aborted/max:758.0 - response_length_non_aborted/min:98.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:367.65625 - prompt_length/max:491.0 - prompt_length/min:197.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00020078755915164948 - timing_s/agent_loop/generate_sequences/min:np.float64(1.3047517966479063) - timing_s/agent_loop/generate_sequences/max:np.float64(6.5038600247353315) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.4670009663968813) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.5038600247353315) - timing_s/agent_loop/slowest/tool_calls:np.flo
|
||
step:69 - global_seqlen/min:18174 - global_seqlen/max:25281 - global_seqlen/minmax_diff:7107 - global_seqlen/balanced_min:22244 - global_seqlen/balanced_max:22250 - global_seqlen/mean:22247.875 - actor/entropy:0.2577400505542755 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1846359670162201 - training/rollout_probs_diff_mean:0.0026801007334142923 - training/rollout_probs_diff_std:0.007650184910744429 - training/rollout_actor_probs_pearson_corr:0.9994962215423584 - rollout_corr/rollout_is_mean:1.000064492225647 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5282295942306519 - rollout_corr/rollout_is_min:0.6526354551315308 - rollout_corr/rollout_is_std:0.025199759751558304 - rollout_corr/rollout_is_eff_sample_size:0.9993654941771045 - rollout_corr/rollout_is_seq_mean:1.0000888109207153 - rollout_corr/rollout_is_seq_std:0.0015610881382599473 - rollout_corr/rollout_is_seq_max:1.0052450895309448 - rollout_corr/rollout_is_seq_min:0.9963499903678894 - rollout_corr/rollout_is_seq_max_deviation:0.005245089530944824 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3210419910028577) - rollout_corr/training_log_ppl:np.float64(0.26073044213990215) - rollout_corr/kl:np.float64(0.0002998370159144059) - rollout_corr/k3_kl:np.float64(0.0003779013446205681) - rollout_corr/rollout_ppl:np.float64(1.3205989100970328) - rollout_corr/rollout_log_ppl:np.float64(0.2604306025314145) - rollout_corr/log_ppl_diff:np.float64(0.0002998396084876731) - rollout_corr/log_ppl_abs_diff:np.float64(0.001366822500131093) - rollout_corr/log_ppl_diff_max:np.float64(0.0053320229053497314) - rollout_corr/log_ppl_diff_min:np.float64(-0.005263388156890869) - rollout_corr/ppl_ratio:np.float64(1.0003015059046447) - rollout_corr/chi2_token:np.float64(0.0009072565007954836) - rollout_corr/chi2_seq:np.float64(0.4102727370336652) - actor/pg_loss:np.float64(-5.396897904574871e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(1.595677440491272e-05) - actor/ppo_kl:np.float64(5.623625808581778e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.050544146448373795) - perf/mfu/actor:np.float64(0.0909921613438466) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.04339218139648) - actor/lr:np.float64(1e-06) - training/global_step:69 - 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:0.0013345195911824703 - critic/advantages/max:0.375 - critic/advantages/min:-0.875 - critic/returns/mean:0.0013345195911824703 - critic/returns/max:0.375 - critic/returns/min:-0.875 - response_length/mean:332.58984375 - response_length/max:779.0 - response_length/min:93.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:332.58984375 - response_length_non_aborted/max:779.0 - response_length_non_aborted/min:93.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:362.65625 - prompt_length/max:461.0 - prompt_length/min:190.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.0001451447606086731 - timing_s/agent_loop/generate_sequences/min:np.float64(1.2613888587802649) - timing_s/agent_loop/generate_sequences/max:np.float64(6.885465707629919) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.741194956128311) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.885465707629919) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowe
|
||
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 for a unit cell with given lattice parameters. For a unit cell with orthorhombic symmetry (where all angles are 90 degrees), the formula is:
|
||
|
||
$$
|
||
V = a \times b \times c
|
||
$$
|
||
|
||
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 $:
|
||
|
||
$$
|
||
7.82 \times 7.82 = 61.1524
|
||
$$
|
||
|
||
Now, multiply by $ 17.88 $:
|
||
|
||
$$
|
||
61.1524 \times 17.88 = 1094.31
|
||
$$
|
||
|
||
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-GRPO-Qwen-Qwen3-8B-mbs8-train32-rollout8-lr1e-6-vllm0.8/global_step_70
|
||
step:70 - global_seqlen/min:20129 - global_seqlen/max:27403 - global_seqlen/minmax_diff:7274 - global_seqlen/balanced_min:22945 - global_seqlen/balanced_max:22953 - global_seqlen/mean:22948.75 - actor/entropy:0.2474832534790039 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.697953999042511 - training/rollout_probs_diff_mean:0.0025176035705953836 - training/rollout_probs_diff_std:0.007814755663275719 - training/rollout_actor_probs_pearson_corr:0.9994720220565796 - rollout_corr/rollout_is_mean:0.9999285340309143 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.189371778309578e-05 - rollout_corr/rollout_is_max:1.6485027074813843 - rollout_corr/rollout_is_min:0.024857966229319572 - rollout_corr/rollout_is_std:0.024460405111312866 - rollout_corr/rollout_is_eff_sample_size:0.9994020463382516 - rollout_corr/rollout_is_seq_mean:0.9999470710754395 - rollout_corr/rollout_is_seq_std:0.0016255476512014866 - rollout_corr/rollout_is_seq_max:1.0055850744247437 - rollout_corr/rollout_is_seq_min:0.994884729385376 - rollout_corr/rollout_is_seq_max_deviation:0.005585074424743652 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.314159161876887) - rollout_corr/training_log_ppl:np.float64(0.2612044901034096) - rollout_corr/kl:np.float64(0.0005048151205322426) - rollout_corr/k3_kl:np.float64(0.0004144159162251526) - rollout_corr/rollout_ppl:np.float64(1.3134744726121426) - rollout_corr/rollout_log_ppl:np.float64(0.260699674137868) - rollout_corr/log_ppl_diff:np.float64(0.0005048159655416384) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012870829232269898) - rollout_corr/log_ppl_diff_max:np.float64(0.009302586317062378) - rollout_corr/log_ppl_diff_min:np.float64(-0.0068359375) - rollout_corr/ppl_ratio:np.float64(1.0005064497236162) - rollout_corr/chi2_token:np.float64(0.0006057738792151213) - rollout_corr/chi2_seq:np.float64(0.13440353889018297) - actor/pg_loss:np.float64(-6.187940016388893e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(5.0769394874805585e-05) - actor/ppo_kl:np.float64(0.00011248510861205929) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.11688512889668345) - perf/mfu/actor:np.float64(0.09366797054721682) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.0029411315918) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7593085106382979) - val-aux/sciknoweval/reward/std@16:np.float64(0.03784596039508453) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7730106382978723) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.03245061258021307) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.746063829787234) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.02930060432284286) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7596914893617022) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.037449532559174645) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7857340425531916) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.029635888676231206) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.7364255319148935) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.023699060838306855) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7574255319148936) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.02639621551830921) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7981914893617023) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.02430626418681097) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.7282872340425532) - val-aux/sciknoweval/reward/worst@8/std:np.float64(0.022082307000321155) - val-aux/sciknoweval/reward/maj@8/mean:np.float64(0.7558297872340426) - val-aux/sciknoweval/reward/maj@8/std:np.float64(0.015649633880830637) - val-aux/sciknoweval/reward/best@16/mean:np.float64(0.8088191489361701) - val-aux/sciknowev
|
||
step:71 - global_seqlen/min:20204 - global_seqlen/max:27625 - global_seqlen/minmax_diff:7421 - global_seqlen/balanced_min:23791 - global_seqlen/balanced_max:23796 - global_seqlen/mean:23793.75 - actor/entropy:0.23197931051254272 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9734354019165039 - training/rollout_probs_diff_mean:0.0025075948797166348 - training/rollout_probs_diff_std:0.008153514936566353 - training/rollout_actor_probs_pearson_corr:0.9994100332260132 - rollout_corr/rollout_is_mean:0.9999135136604309 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.2147904019220732e-05 - rollout_corr/rollout_is_max:1.6415942907333374 - rollout_corr/rollout_is_min:0.018141763284802437 - rollout_corr/rollout_is_std:0.02477516233921051 - rollout_corr/rollout_is_eff_sample_size:0.9993864488349556 - rollout_corr/rollout_is_seq_mean:0.9999602437019348 - rollout_corr/rollout_is_seq_std:0.0016766986809670925 - rollout_corr/rollout_is_seq_max:1.0064085721969604 - rollout_corr/rollout_is_seq_min:0.9950252175331116 - rollout_corr/rollout_is_seq_max_deviation:0.006408572196960449 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.294114553835243) - rollout_corr/training_log_ppl:np.float64(0.25148121974780224) - rollout_corr/kl:np.float64(0.0005309915984341718) - rollout_corr/k3_kl:np.float64(0.00038931291395272183) - rollout_corr/rollout_ppl:np.float64(1.2933996450155973) - rollout_corr/rollout_log_ppl:np.float64(0.2509502254106337) - rollout_corr/log_ppl_diff:np.float64(0.0005309943371685222) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013700544222956523) - rollout_corr/log_ppl_diff_max:np.float64(0.01174214482307434) - rollout_corr/log_ppl_diff_min:np.float64(-0.004778727889060974) - rollout_corr/ppl_ratio:np.float64(1.0005327749531716) - rollout_corr/chi2_token:np.float64(0.0004457335453480482) - rollout_corr/chi2_seq:np.float64(0.12818227359093726) - actor/pg_loss:np.float64(-2.503779251128435e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(2.104266059177462e-05) - actor/ppo_kl:np.float64(0.00013277838810599185) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.0483645498752594) - perf/mfu/actor:np.float64(0.09704824978962062) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.64053344726562) - actor/lr:np.float64(1e-06) - training/global_step:71 - training/epoch:2 - 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:0.0008568470366299152 - critic/advantages/max:0.125 - critic/advantages/min:-0.875 - critic/returns/mean:0.0008568470366299152 - critic/returns/max:0.125 - critic/returns/min:-0.875 - response_length/mean:352.7421875 - response_length/max:988.0 - response_length/min:146.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:352.7421875 - response_length_non_aborted/max:988.0 - response_length_non_aborted/min:146.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:390.8125 - prompt_length/max:482.0 - prompt_length/min:263.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:5.001761019229889e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8390790149569511) - timing_s/agent_loop/generate_sequences/max:np.float64(7.671897960826755) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.904139525635401) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.671897960826755) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_
|
||
step:72 - global_seqlen/min:21608 - global_seqlen/max:27136 - global_seqlen/minmax_diff:5528 - global_seqlen/balanced_min:24254 - global_seqlen/balanced_max:24258 - global_seqlen/mean:24256.25 - actor/entropy:0.19864249229431152 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1831313669681549 - training/rollout_probs_diff_mean:0.0021919270511716604 - training/rollout_probs_diff_std:0.006964217871427536 - training/rollout_actor_probs_pearson_corr:0.9995092749595642 - rollout_corr/rollout_is_mean:0.999876856803894 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5391958951950073 - rollout_corr/rollout_is_min:0.6549360752105713 - rollout_corr/rollout_is_std:0.022048480808734894 - rollout_corr/rollout_is_eff_sample_size:0.9995141007300881 - rollout_corr/rollout_is_seq_mean:0.9998931884765625 - rollout_corr/rollout_is_seq_std:0.0013704376760870218 - rollout_corr/rollout_is_seq_max:1.003908634185791 - rollout_corr/rollout_is_seq_min:0.9955668449401855 - rollout_corr/rollout_is_seq_max_deviation:0.004433155059814453 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2588270916603506) - rollout_corr/training_log_ppl:np.float64(0.2237605569389416) - rollout_corr/kl:np.float64(0.000427495997047056) - rollout_corr/k3_kl:np.float64(0.0003172674925409069) - rollout_corr/rollout_ppl:np.float64(1.2582736886106431) - rollout_corr/rollout_log_ppl:np.float64(0.22333305978099816) - rollout_corr/log_ppl_diff:np.float64(0.000427497157943435) - rollout_corr/log_ppl_abs_diff:np.float64(0.001277009301702492) - rollout_corr/log_ppl_diff_max:np.float64(0.0061948299407958984) - rollout_corr/log_ppl_diff_min:np.float64(-0.004580110311508179) - rollout_corr/ppl_ratio:np.float64(1.0004288996569812) - rollout_corr/chi2_token:np.float64(0.0004238623660057783) - rollout_corr/chi2_seq:np.float64(0.11839799722656608) - actor/pg_loss:np.float64(-8.923967834562063e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(2.7999737540085334e-05) - actor/ppo_kl:np.float64(4.3629116550647495e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.06856918195262551) - perf/mfu/actor:np.float64(0.09946419479837837) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.11327743530273) - actor/lr:np.float64(1e-06) - training/global_step:72 - training/epoch:2 - critic/score/mean:0.796875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.796875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:1.2195121598779224e-05 - critic/advantages/max:0.125 - critic/advantages/min:-0.875 - critic/returns/mean:1.2195121598779224e-05 - critic/returns/max:0.125 - critic/returns/min:-0.875 - response_length/mean:360.3515625 - response_length/max:797.0 - response_length/min:162.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:360.3515625 - response_length_non_aborted/max:797.0 - response_length_non_aborted/min:162.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:397.65625 - prompt_length/max:512.0 - prompt_length/min:264.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:9.410083293914795e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.1268845815211535) - timing_s/agent_loop/generate_sequences/max:np.float64(7.048265373334289) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.9732546996310703) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.048265373334289) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowes
|
||
step:73 - global_seqlen/min:19053 - global_seqlen/max:28796 - global_seqlen/minmax_diff:9743 - global_seqlen/balanced_min:23456 - global_seqlen/balanced_max:23464 - global_seqlen/mean:23461.25 - actor/entropy:0.22118522226810455 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2151082158088684 - training/rollout_probs_diff_mean:0.0023402885999530554 - training/rollout_probs_diff_std:0.007146602030843496 - training/rollout_actor_probs_pearson_corr:0.9995121359825134 - rollout_corr/rollout_is_mean:1.00008225440979 - rollout_corr/rollout_is_ratio_fraction_high:1.1420479495427571e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.1420479495427571e-05 - rollout_corr/rollout_is_max:2.6541292667388916 - rollout_corr/rollout_is_min:0.45500603318214417 - rollout_corr/rollout_is_std:0.024027705192565918 - rollout_corr/rollout_is_eff_sample_size:0.9994232406725769 - rollout_corr/rollout_is_seq_mean:1.000099778175354 - rollout_corr/rollout_is_seq_std:0.0016823429614305496 - rollout_corr/rollout_is_seq_max:1.0069019794464111 - rollout_corr/rollout_is_seq_min:0.995582640171051 - rollout_corr/rollout_is_seq_max_deviation:0.006901979446411133 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2888292078860104) - rollout_corr/training_log_ppl:np.float64(0.23963962492416613) - rollout_corr/kl:np.float64(0.00026983858866547195) - rollout_corr/k3_kl:np.float64(0.0003792109853435477) - rollout_corr/rollout_ppl:np.float64(1.2885228227823973) - rollout_corr/rollout_log_ppl:np.float64(0.23936978423444089) - rollout_corr/log_ppl_diff:np.float64(0.00026984068972524256) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012609190453076735) - rollout_corr/log_ppl_diff_max:np.float64(0.007149204611778259) - rollout_corr/log_ppl_diff_min:np.float64(-0.006377771496772766) - rollout_corr/ppl_ratio:np.float64(1.0002713806461543) - rollout_corr/chi2_token:np.float64(0.0011249079834669828) - rollout_corr/chi2_seq:np.float64(0.2845069239847362) - actor/pg_loss:np.float64(4.49218787252903e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(9.43538634601282e-06) - actor/ppo_kl:np.float64(4.757973179181363e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1339050605893135) - perf/mfu/actor:np.float64(0.09533829770092217) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.18564987182617) - actor/lr:np.float64(1e-06) - training/global_step:73 - training/epoch:2 - critic/score/mean:0.71484375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.71484375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0012262739473953843 - critic/advantages/max:0.875 - critic/advantages/min:-0.5 - critic/returns/mean:0.0012262739473953843 - critic/returns/max:0.875 - critic/returns/min:-0.5 - response_length/mean:342.0390625 - response_length/max:847.0 - response_length/min:126.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:342.0390625 - response_length_non_aborted/max:847.0 - response_length_non_aborted/min:126.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:391.125 - prompt_length/max:494.0 - prompt_length/min:196.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:8.323974907398224e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.573115924373269) - timing_s/agent_loop/generate_sequences/max:np.float64(7.276758732274175) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.74214253291575) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.276758732274175) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0)
|
||
step:74 - global_seqlen/min:18126 - global_seqlen/max:26046 - global_seqlen/minmax_diff:7920 - global_seqlen/balanced_min:22541 - global_seqlen/balanced_max:22548 - global_seqlen/mean:22545.0 - actor/entropy:0.260564386844635 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.16881024837493896 - training/rollout_probs_diff_mean:0.0027874174993485212 - training/rollout_probs_diff_std:0.0077921925112605095 - training/rollout_actor_probs_pearson_corr:0.9994821548461914 - rollout_corr/rollout_is_mean:1.000143051147461 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.586216926574707 - rollout_corr/rollout_is_min:0.6137421727180481 - rollout_corr/rollout_is_std:0.025568412616848946 - rollout_corr/rollout_is_eff_sample_size:0.9993470405186317 - rollout_corr/rollout_is_seq_mean:1.0001143217086792 - rollout_corr/rollout_is_seq_std:0.0015101114986464381 - rollout_corr/rollout_is_seq_max:1.0060126781463623 - rollout_corr/rollout_is_seq_min:0.9956173896789551 - rollout_corr/rollout_is_seq_max_deviation:0.006012678146362305 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3129614721983671) - rollout_corr/training_log_ppl:np.float64(0.26557272001809906) - rollout_corr/kl:np.float64(0.0003072068811249551) - rollout_corr/k3_kl:np.float64(0.0003651777730624417) - rollout_corr/rollout_ppl:np.float64(1.312554101459682) - rollout_corr/rollout_log_ppl:np.float64(0.26526551083952654) - rollout_corr/log_ppl_diff:np.float64(0.0003072091785725206) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012191955756861717) - rollout_corr/log_ppl_diff_max:np.float64(0.004254922270774841) - rollout_corr/log_ppl_diff_min:np.float64(-0.005225449800491333) - rollout_corr/ppl_ratio:np.float64(1.0003084915224463) - rollout_corr/chi2_token:np.float64(0.0008592354133725166) - rollout_corr/chi2_seq:np.float64(0.40940430806949735) - actor/pg_loss:np.float64(6.841961294412613e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(5.0106103572034044e-05) - actor/ppo_kl:np.float64(8.371459145450899e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.05365279596298933) - perf/mfu/actor:np.float64(0.09285714441128551) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.1694221496582) - actor/lr:np.float64(1e-06) - training/global_step:74 - 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:-0.0009363640565425158 - critic/advantages/max:0.25 - critic/advantages/min:-0.75 - critic/returns/mean:-0.0009363640565425158 - critic/returns/max:0.25 - critic/returns/min:-0.75 - response_length/mean:318.09375 - response_length/max:961.0 - response_length/min:166.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:318.09375 - response_length_non_aborted/max:961.0 - response_length_non_aborted/min:166.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:386.4375 - prompt_length/max:481.0 - prompt_length/min:264.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00012472458183765411 - timing_s/agent_loop/generate_sequences/min:np.float64(1.4892469886690378) - timing_s/agent_loop/generate_sequences/max:np.float64(7.678874081000686) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.469133668273571) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.678874081000686) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/prompt_
|
||
step:75 - global_seqlen/min:19411 - global_seqlen/max:26389 - global_seqlen/minmax_diff:6978 - global_seqlen/balanced_min:22835 - global_seqlen/balanced_max:22839 - global_seqlen/mean:22837.125 - actor/entropy:0.26598554849624634 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12735021114349365 - training/rollout_probs_diff_mean:0.0026302863843739033 - training/rollout_probs_diff_std:0.007375993765890598 - training/rollout_actor_probs_pearson_corr:0.9995732307434082 - rollout_corr/rollout_is_mean:0.9998831152915955 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4389069080352783 - rollout_corr/rollout_is_min:0.6956850290298462 - rollout_corr/rollout_is_std:0.024746274575591087 - rollout_corr/rollout_is_eff_sample_size:0.9993877585302487 - rollout_corr/rollout_is_seq_mean:0.999903678894043 - rollout_corr/rollout_is_seq_std:0.001642070128582418 - rollout_corr/rollout_is_seq_max:1.006669282913208 - rollout_corr/rollout_is_seq_min:0.9947586059570312 - rollout_corr/rollout_is_seq_max_deviation:0.006669282913208008 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3507126765325665) - rollout_corr/training_log_ppl:np.float64(0.28722253389423713) - rollout_corr/kl:np.float64(0.0003447701374001966) - rollout_corr/k3_kl:np.float64(0.0003724003217939753) - rollout_corr/rollout_ppl:np.float64(1.350269827991724) - rollout_corr/rollout_log_ppl:np.float64(0.286877762133372) - rollout_corr/log_ppl_diff:np.float64(0.00034477176086511463) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012942667381139472) - rollout_corr/log_ppl_diff_max:np.float64(0.005844295024871826) - rollout_corr/log_ppl_diff_min:np.float64(-0.006119906902313232) - rollout_corr/ppl_ratio:np.float64(1.0003462715540081) - rollout_corr/chi2_token:np.float64(0.0008182954043149948) - rollout_corr/chi2_seq:np.float64(0.32993280817754567) - actor/pg_loss:np.float64(2.505222801119089e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(6.734913768013939e-06) - actor/ppo_kl:np.float64(-8.327829606002979e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.06800705939531326) - perf/mfu/actor:np.float64(0.09338293257404955) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.09074020385742) - actor/lr:np.float64(1e-06) - training/global_step:75 - training/epoch:2 - critic/score/mean:0.6796875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6796875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0049235341139137745 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0049235341139137745 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:319.53515625 - response_length/max:740.0 - response_length/min:165.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:319.53515625 - response_length_non_aborted/max:740.0 - response_length_non_aborted/min:165.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:394.125 - prompt_length/max:493.0 - prompt_length/min:196.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:4.5044347643852234e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9954011496156454) - timing_s/agent_loop/generate_sequences/max:np.float64(6.424747133627534) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.6563248890815885) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.424747133627534) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slo
|
||
step:76 - global_seqlen/min:18401 - global_seqlen/max:23716 - global_seqlen/minmax_diff:5315 - global_seqlen/balanced_min:21350 - global_seqlen/balanced_max:21353 - global_seqlen/mean:21352.0 - actor/entropy:0.29671216011047363 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2932341992855072 - training/rollout_probs_diff_mean:0.002921811304986477 - training/rollout_probs_diff_std:0.007850022055208683 - training/rollout_actor_probs_pearson_corr:0.9994902014732361 - rollout_corr/rollout_is_mean:0.9999136924743652 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.6809651899384335e-05 - rollout_corr/rollout_is_max:1.6220922470092773 - rollout_corr/rollout_is_min:0.3131377696990967 - rollout_corr/rollout_is_std:0.026378488168120384 - rollout_corr/rollout_is_eff_sample_size:0.9993046592121337 - rollout_corr/rollout_is_seq_mean:0.9999048113822937 - rollout_corr/rollout_is_seq_std:0.001711147604510188 - rollout_corr/rollout_is_seq_max:1.0068550109863281 - rollout_corr/rollout_is_seq_min:0.9951537847518921 - rollout_corr/rollout_is_seq_max_deviation:0.006855010986328125 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3681150269694626) - rollout_corr/training_log_ppl:np.float64(0.3019279351574369) - rollout_corr/kl:np.float64(0.00037963077198455153) - rollout_corr/k3_kl:np.float64(0.00038022535707682437) - rollout_corr/rollout_ppl:np.float64(1.3675709716044366) - rollout_corr/rollout_log_ppl:np.float64(0.30154830423998646) - rollout_corr/log_ppl_diff:np.float64(0.0003796309174504131) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012849238410126418) - rollout_corr/log_ppl_diff_max:np.float64(0.005592137575149536) - rollout_corr/log_ppl_diff_min:np.float64(-0.0062049925327301025) - rollout_corr/ppl_ratio:np.float64(1.0003810485359281) - rollout_corr/chi2_token:np.float64(0.0007722426671534777) - rollout_corr/chi2_seq:np.float64(0.23392329807393253) - actor/pg_loss:np.float64(-9.966082870960236e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(5.643598342430778e-05) - actor/ppo_kl:np.float64(-7.630967955130075e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1309410072863102) - perf/mfu/actor:np.float64(0.08796069759198259) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.3106803894043) - actor/lr:np.float64(1e-06) - training/global_step:76 - training/epoch:2 - 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:-0.0013857238227501512 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0013857238227501512 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:291.40625 - response_length/max:715.0 - response_length/min:142.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:291.40625 - response_length_non_aborted/max:715.0 - response_length_non_aborted/min:142.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:375.84375 - prompt_length/max:471.0 - prompt_length/min:196.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:2.397596836090088e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8652194179594517) - timing_s/agent_loop/generate_sequences/max:np.float64(6.0747572015970945) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.322189557933598) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.0747572015970945) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/ag
|
||
step:77 - global_seqlen/min:20873 - global_seqlen/max:27379 - global_seqlen/minmax_diff:6506 - global_seqlen/balanced_min:24520 - global_seqlen/balanced_max:24522 - global_seqlen/mean:24521.25 - actor/entropy:0.21010300517082214 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9525610208511353 - training/rollout_probs_diff_mean:0.002213509287685156 - training/rollout_probs_diff_std:0.007591753266751766 - training/rollout_actor_probs_pearson_corr:0.9994261264801025 - rollout_corr/rollout_is_mean:0.9999380707740784 - rollout_corr/rollout_is_ratio_fraction_high:1.0898707841988653e-05 - rollout_corr/rollout_is_ratio_fraction_low:3.269612352596596e-05 - rollout_corr/rollout_is_max:2.0322699546813965 - rollout_corr/rollout_is_min:1.2050280020048376e-05 - rollout_corr/rollout_is_std:0.02297777123749256 - rollout_corr/rollout_is_eff_sample_size:0.9994720625032244 - rollout_corr/rollout_is_seq_mean:0.9999042749404907 - rollout_corr/rollout_is_seq_std:0.0015725604025647044 - rollout_corr/rollout_is_seq_max:1.0046145915985107 - rollout_corr/rollout_is_seq_min:0.9941759705543518 - rollout_corr/rollout_is_seq_max_deviation:0.005824029445648193 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2686810540035367) - rollout_corr/training_log_ppl:np.float64(0.22998299021855928) - rollout_corr/kl:np.float64(0.0006073243849300525) - rollout_corr/k3_kl:np.float64(0.0005009522885188744) - rollout_corr/rollout_ppl:np.float64(1.2678813147358596) - rollout_corr/rollout_log_ppl:np.float64(0.22937566417385824) - rollout_corr/log_ppl_diff:np.float64(0.0006073260447010398) - rollout_corr/log_ppl_abs_diff:np.float64(0.001405421266099438) - rollout_corr/log_ppl_diff_max:np.float64(0.05526180565357208) - rollout_corr/log_ppl_diff_min:np.float64(-0.004332512617111206) - rollout_corr/ppl_ratio:np.float64(1.0006146889645606) - rollout_corr/chi2_token:np.float64(0.00042232475243508816) - rollout_corr/chi2_seq:np.float64(0.13180363178253174) - 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(2.2369566373470207e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.0) - perf/mfu/actor:np.float64(0.09757031732597408) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.31251525878906) - actor/lr:np.float64(1e-06) - training/global_step:77 - training/epoch:2 - critic/score/mean:0.53125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.53125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0 - 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:358.4140625 - response_length/max:697.0 - response_length/min:156.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:358.4140625 - response_length_non_aborted/max:697.0 - response_length_non_aborted/min:156.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:407.875 - prompt_length/max:510.0 - prompt_length/min:264.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.000134320929646492 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9717217683792114) - timing_s/agent_loop/generate_sequences/max:np.float64(6.515862211585045) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.02899411416729) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.515862211585045) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/prompt_length:420 - timing_s/agent_loop/slowest/response_length:
|
||
step:78 - global_seqlen/min:21151 - global_seqlen/max:25607 - global_seqlen/minmax_diff:4456 - global_seqlen/balanced_min:23008 - global_seqlen/balanced_max:23012 - global_seqlen/mean:23011.25 - actor/entropy:0.2615104913711548 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4400331377983093 - training/rollout_probs_diff_mean:0.0026174753438681364 - training/rollout_probs_diff_std:0.007456088438630104 - training/rollout_actor_probs_pearson_corr:0.9995262026786804 - rollout_corr/rollout_is_mean:0.9998167753219604 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.2719733604171779e-05 - rollout_corr/rollout_is_max:1.454869270324707 - rollout_corr/rollout_is_min:0.46178218722343445 - rollout_corr/rollout_is_std:0.024490846320986748 - rollout_corr/rollout_is_eff_sample_size:0.9994004984726531 - rollout_corr/rollout_is_seq_mean:0.9998263716697693 - rollout_corr/rollout_is_seq_std:0.0016342015005648136 - rollout_corr/rollout_is_seq_max:1.005255103111267 - rollout_corr/rollout_is_seq_min:0.9942693710327148 - rollout_corr/rollout_is_seq_max_deviation:0.005730628967285156 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.330007707234472) - rollout_corr/training_log_ppl:np.float64(0.27612566971220076) - rollout_corr/kl:np.float64(0.0005081333625520656) - rollout_corr/k3_kl:np.float64(0.00034416606136744576) - rollout_corr/rollout_ppl:np.float64(1.3292845152318478) - rollout_corr/rollout_log_ppl:np.float64(0.2756175332906423) - rollout_corr/log_ppl_diff:np.float64(0.0005081364215584472) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013684313307749107) - rollout_corr/log_ppl_diff_max:np.float64(0.006606578826904297) - rollout_corr/log_ppl_diff_min:np.float64(-0.004864111542701721) - rollout_corr/ppl_ratio:np.float64(1.00050969584845) - rollout_corr/chi2_token:np.float64(0.00036964891478419304) - rollout_corr/chi2_seq:np.float64(0.0903538609854877) - actor/pg_loss:np.float64(2.2566295228898525e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0) - actor/ppo_kl:np.float64(1.0493426182200949e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.020936077926307917) - perf/mfu/actor:np.float64(0.09409445481639155) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.23200035095215) - actor/lr:np.float64(1e-06) - training/global_step:78 - 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:0.0004102113889530301 - critic/advantages/max:0.125 - critic/advantages/min:-0.875 - critic/returns/mean:0.0004102113889530301 - critic/returns/max:0.125 - critic/returns/min:-0.875 - response_length/mean:307.1015625 - response_length/max:779.0 - response_length/min:123.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:307.1015625 - response_length_non_aborted/max:779.0 - response_length_non_aborted/min:123.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:412.0 - prompt_length/max:493.0 - prompt_length/min:261.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00012887082993984222 - timing_s/agent_loop/generate_sequences/min:np.float64(1.6045669373124838) - timing_s/agent_loop/generate_sequences/max:np.float64(6.628957685083151) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.5532616262498777) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.628957685083151) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowes
|
||
step:79 - global_seqlen/min:18926 - global_seqlen/max:25995 - global_seqlen/minmax_diff:7069 - global_seqlen/balanced_min:23504 - global_seqlen/balanced_max:23505 - global_seqlen/mean:23504.5 - actor/entropy:0.24806106090545654 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4102373421192169 - training/rollout_probs_diff_mean:0.0023650529328733683 - training/rollout_probs_diff_std:0.0071376836858689785 - training/rollout_actor_probs_pearson_corr:0.9995306730270386 - rollout_corr/rollout_is_mean:1.0001239776611328 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.1717834240698721e-05 - rollout_corr/rollout_is_max:1.7178680896759033 - rollout_corr/rollout_is_min:0.4388447403907776 - rollout_corr/rollout_is_std:0.023599699139595032 - rollout_corr/rollout_is_eff_sample_size:0.99944348329181 - rollout_corr/rollout_is_seq_mean:1.0001518726348877 - rollout_corr/rollout_is_seq_std:0.0014191392110660672 - rollout_corr/rollout_is_seq_max:1.004914402961731 - rollout_corr/rollout_is_seq_min:0.9959833025932312 - rollout_corr/rollout_is_seq_max_deviation:0.004914402961730957 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3128615743480623) - rollout_corr/training_log_ppl:np.float64(0.26136835565557703) - rollout_corr/kl:np.float64(0.00019384605705319302) - rollout_corr/k3_kl:np.float64(0.0003144730202855328) - rollout_corr/rollout_ppl:np.float64(1.3125933022238314) - rollout_corr/rollout_log_ppl:np.float64(0.26117450813762844) - rollout_corr/log_ppl_diff:np.float64(0.00019384751794859767) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011084461002610624) - rollout_corr/log_ppl_diff_max:np.float64(0.0037963688373565674) - rollout_corr/log_ppl_diff_min:np.float64(-0.004113912582397461) - rollout_corr/ppl_ratio:np.float64(1.0001948736608028) - rollout_corr/chi2_token:np.float64(0.0008782581426203251) - rollout_corr/chi2_seq:np.float64(0.2646336832549423) - actor/pg_loss:np.float64(1.6814330592751503e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(8.797860573395155e-06) - actor/ppo_kl:np.float64(5.1398302620242475e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.03347182273864746) - perf/mfu/actor:np.float64(0.09539341168894122) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.25070190429688) - actor/lr:np.float64(1e-06) - training/global_step:79 - 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:-0.003666217438876629 - critic/advantages/max:0.125 - critic/advantages/min:-0.875 - critic/returns/mean:-0.003666217438876629 - critic/returns/max:0.125 - critic/returns/min:-0.875 - response_length/mean:333.359375 - response_length/max:757.0 - response_length/min:161.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:333.359375 - response_length_non_aborted/max:757.0 - response_length_non_aborted/min:161.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:401.15625 - prompt_length/max:494.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00027751922607421875 - timing_s/agent_loop/generate_sequences/min:np.float64(2.0520722176879644) - timing_s/agent_loop/generate_sequences/max:np.float64(6.681050945073366) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.8504629519884475) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.681050945073366) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timin
|
||
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 of a unit cell in a crystal structure can be calculated using the formula for the volume of a parallelepiped defined by the lattice vectors:
|
||
|
||
$$ V = a \cdot b \cdot c \cdot \sqrt{1 + 2\cos\alpha\cos\beta\cos\gamma - \cos^2\alpha - \cos^2\beta - \cos^2\gamma} $$
|
||
|
||
In this case, the crystal structure is a unit cell with all angles equal to 90°, which means it is a **primitive orthorhombic** unit cell. For such a structure, the formula simplifies to:
|
||
|
||
$$ V = a \cdot b \cdot c $$
|
||
|
||
Given:
|
||
- a = 7.82 Å
|
||
- b = 7.82 Å
|
||
- c = 17.88 Å
|
||
|
||
Now, calculate the volume:
|
||
|
||
$$ V = 7.82 \cdot 7.82 \cdot 17.88 $$
|
||
|
||
First, calculate $ 7.82 \cdot 7.82 $:
|
||
|
||
$$ 7.82 \cdot 7.82 = 61.1524 $$
|
||
|
||
Then multiply by 17.88:
|
||
|
||
$$ 61.1524 \cdot 17.88 = 1094.31 $$
|
||
|
||
Therefore, the volume of the unit cell is 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-GRPO-Qwen-Qwen3-8B-mbs8-train32-rollout8-lr1e-6-vllm0.8/global_step_80
|
||
step:80 - global_seqlen/min:20780 - global_seqlen/max:26076 - global_seqlen/minmax_diff:5296 - global_seqlen/balanced_min:23379 - global_seqlen/balanced_max:23381 - global_seqlen/mean:23380.0 - actor/entropy:0.2778765857219696 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23994797468185425 - training/rollout_probs_diff_mean:0.0025989830028265715 - training/rollout_probs_diff_std:0.007374095730483532 - training/rollout_actor_probs_pearson_corr:0.9996104836463928 - rollout_corr/rollout_is_mean:0.9999150037765503 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.1238986189709976e-05 - rollout_corr/rollout_is_max:1.4688119888305664 - rollout_corr/rollout_is_min:0.30741897225379944 - rollout_corr/rollout_is_std:0.02457466721534729 - rollout_corr/rollout_is_eff_sample_size:0.9993962121006339 - rollout_corr/rollout_is_seq_mean:0.999893307685852 - rollout_corr/rollout_is_seq_std:0.0016186292050406337 - rollout_corr/rollout_is_seq_max:1.0051453113555908 - rollout_corr/rollout_is_seq_min:0.9950850605964661 - rollout_corr/rollout_is_seq_max_deviation:0.00514531135559082 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3502345071174204) - rollout_corr/training_log_ppl:np.float64(0.289737257262459) - rollout_corr/kl:np.float64(0.0005005977039438392) - rollout_corr/k3_kl:np.float64(0.000352018927735287) - rollout_corr/rollout_ppl:np.float64(1.3495450923219323) - rollout_corr/rollout_log_ppl:np.float64(0.2892366573069012) - rollout_corr/log_ppl_diff:np.float64(0.0005005999555578455) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013773568934993818) - rollout_corr/log_ppl_diff_max:np.float64(0.006760120391845703) - rollout_corr/log_ppl_diff_min:np.float64(-0.004993259906768799) - rollout_corr/ppl_ratio:np.float64(1.0005022925324738) - rollout_corr/chi2_token:np.float64(0.00039854785427451134) - rollout_corr/chi2_seq:np.float64(0.2169972979463637) - actor/pg_loss:np.float64(-0.00011445709969848394) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(8.960122886492172e-05) - actor/ppo_kl:np.float64(6.888215898825933e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.12062663584947586) - perf/mfu/actor:np.float64(0.09535256392760892) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.24382019042969) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7506648936170213) - val-aux/sciknoweval/reward/std@16:np.float64(0.03880700651700434) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7653510638297872) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.029800870503189608) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.7373936170212767) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.0355050684444622) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7515851063829787) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.03928199705006994) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7755851063829787) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.02222459000259867) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.7243510638297872) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.03196328070235044) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7520319148936169) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.027285394753994465) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7842234042553193) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.016187705596573548) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.7106489361702127) - val-aux/sciknoweval/reward/worst@8/std:np.float64(0.029416911005616678) - val-aux/sciknoweval/reward/maj@8/mean:np.float64(0.7507765957446809) - val-aux/sciknoweval/reward/maj@8/std:np.float64(0.015715671528587116) - val-aux/sciknoweval/reward/best@16/mean:np.float64(0.7905000000000002) - val-aux/sc
|
||
step:81 - global_seqlen/min:18017 - global_seqlen/max:28164 - global_seqlen/minmax_diff:10147 - global_seqlen/balanced_min:21751 - global_seqlen/balanced_max:21756 - global_seqlen/mean:21754.5 - actor/entropy:0.2879306674003601 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4863179922103882 - training/rollout_probs_diff_mean:0.0025793167296797037 - training/rollout_probs_diff_std:0.007397323846817017 - training/rollout_actor_probs_pearson_corr:0.9995822906494141 - rollout_corr/rollout_is_mean:1.0000742673873901 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.5967281544581056e-05 - rollout_corr/rollout_is_max:1.549249529838562 - rollout_corr/rollout_is_min:0.29609444737434387 - rollout_corr/rollout_is_std:0.024538258090615273 - rollout_corr/rollout_is_eff_sample_size:0.999398593413883 - rollout_corr/rollout_is_seq_mean:1.0001031160354614 - rollout_corr/rollout_is_seq_std:0.0016027017263695598 - rollout_corr/rollout_is_seq_max:1.0047954320907593 - rollout_corr/rollout_is_seq_min:0.9945899248123169 - rollout_corr/rollout_is_seq_max_deviation:0.0054100751876831055 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3673228165134788) - rollout_corr/training_log_ppl:np.float64(0.2993711391609395) - rollout_corr/kl:np.float64(0.00029862392293544815) - rollout_corr/k3_kl:np.float64(0.00041313828008071596) - rollout_corr/rollout_ppl:np.float64(1.366905136499554) - rollout_corr/rollout_log_ppl:np.float64(0.2990725145646138) - rollout_corr/log_ppl_diff:np.float64(0.0002986245963256806) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013556781050283462) - rollout_corr/log_ppl_diff_max:np.float64(0.006214797496795654) - rollout_corr/log_ppl_diff_min:np.float64(-0.004116475582122803) - rollout_corr/ppl_ratio:np.float64(1.0003002183511853) - rollout_corr/chi2_token:np.float64(0.001083672046661377) - rollout_corr/chi2_seq:np.float64(0.25717995106242597) - actor/pg_loss:np.float64(4.586763679981232e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0) - actor/ppo_kl:np.float64(7.988616901499768e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1057282155379653) - perf/mfu/actor:np.float64(0.08956160130432747) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(107.28913879394531) - actor/lr:np.float64(1e-06) - training/global_step:81 - training/epoch:3 - critic/score/mean:0.703125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.703125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0001314593682764098 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0001314593682764098 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:300.859375 - response_length/max:690.0 - response_length/min:113.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:300.859375 - response_length_non_aborted/max:690.0 - response_length_non_aborted/min:113.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:378.96875 - prompt_length/max:524.0 - prompt_length/min:189.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:2.6717782020568848e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.362369280308485) - timing_s/agent_loop/generate_sequences/max:np.float64(6.274985628202558) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.3638904910403653) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.274985628202558) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/pro
|
||
step:82 - global_seqlen/min:20421 - global_seqlen/max:24727 - global_seqlen/minmax_diff:4306 - global_seqlen/balanced_min:22230 - global_seqlen/balanced_max:22232 - global_seqlen/mean:22231.125 - actor/entropy:0.32268545031547546 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.15518474578857422 - training/rollout_probs_diff_mean:0.0027826493605971336 - training/rollout_probs_diff_std:0.0072498926892876625 - training/rollout_actor_probs_pearson_corr:0.9995722770690918 - rollout_corr/rollout_is_mean:1.000037431716919 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.266447998204967e-05 - rollout_corr/rollout_is_max:1.586849331855774 - rollout_corr/rollout_is_min:0.3824073374271393 - rollout_corr/rollout_is_std:0.025383582338690758 - rollout_corr/rollout_is_eff_sample_size:0.9993560886789191 - rollout_corr/rollout_is_seq_mean:1.0000476837158203 - rollout_corr/rollout_is_seq_std:0.001457554753869772 - rollout_corr/rollout_is_seq_max:1.0035213232040405 - rollout_corr/rollout_is_seq_min:0.9956403374671936 - rollout_corr/rollout_is_seq_max_deviation:0.0043596625328063965 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4052514918148518) - rollout_corr/training_log_ppl:np.float64(0.32896969285502564) - rollout_corr/kl:np.float64(0.00020728115616858167) - rollout_corr/k3_kl:np.float64(0.00039224870215548435) - rollout_corr/rollout_ppl:np.float64(1.4049864276312292) - rollout_corr/rollout_log_ppl:np.float64(0.32876241035410203) - rollout_corr/log_ppl_diff:np.float64(0.00020728250092361122) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013390206877375022) - rollout_corr/log_ppl_diff_max:np.float64(0.005917102098464966) - rollout_corr/log_ppl_diff_min:np.float64(-0.004551202058792114) - rollout_corr/ppl_ratio:np.float64(1.0002087897155434) - rollout_corr/chi2_token:np.float64(0.0011829480063170195) - rollout_corr/chi2_seq:np.float64(0.323864143807441) - actor/pg_loss:np.float64(-2.7368776500225067e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(3.27955094689969e-05) - actor/ppo_kl:np.float64(-7.605274061872613e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.06758737610653043) - perf/mfu/actor:np.float64(0.09014706940472565) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.4574966430664) - actor/lr:np.float64(1e-06) - training/global_step:82 - training/epoch:3 - 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:0.002290687756612897 - critic/advantages/max:0.5 - critic/advantages/min:-0.875 - critic/returns/mean:0.002290687756612897 - critic/returns/max:0.5 - critic/returns/min:-0.875 - response_length/mean:308.44140625 - response_length/max:903.0 - response_length/min:146.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:308.44140625 - response_length_non_aborted/max:903.0 - response_length_non_aborted/min:146.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:386.28125 - prompt_length/max:510.0 - prompt_length/min:266.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00011214055120944977 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8820667397230864) - timing_s/agent_loop/generate_sequences/max:np.float64(7.370335368439555) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.4890292429554393) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.370335368439555) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timi
|
||
step:83 - global_seqlen/min:20784 - global_seqlen/max:24821 - global_seqlen/minmax_diff:4037 - global_seqlen/balanced_min:22452 - global_seqlen/balanced_max:22456 - global_seqlen/mean:22454.375 - actor/entropy:0.28845345973968506 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24496477842330933 - training/rollout_probs_diff_mean:0.0025984023232012987 - training/rollout_probs_diff_std:0.007140541914850473 - training/rollout_actor_probs_pearson_corr:0.9996364712715149 - rollout_corr/rollout_is_mean:0.9999492764472961 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6114643812179565 - rollout_corr/rollout_is_min:0.5105829834938049 - rollout_corr/rollout_is_std:0.02469322830438614 - rollout_corr/rollout_is_eff_sample_size:0.9993904969951359 - rollout_corr/rollout_is_seq_mean:1.0000005960464478 - rollout_corr/rollout_is_seq_std:0.001637228881008923 - rollout_corr/rollout_is_seq_max:1.0061671733856201 - rollout_corr/rollout_is_seq_min:0.9956061244010925 - rollout_corr/rollout_is_seq_max_deviation:0.006167173385620117 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3608696786686778) - rollout_corr/training_log_ppl:np.float64(0.298661266482668) - rollout_corr/kl:np.float64(0.0002497441816160517) - rollout_corr/k3_kl:np.float64(0.0003628389737002635) - rollout_corr/rollout_ppl:np.float64(1.360557473730296) - rollout_corr/rollout_log_ppl:np.float64(0.2984115183062386) - rollout_corr/log_ppl_diff:np.float64(0.0002497481764294207) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013755984837189317) - rollout_corr/log_ppl_diff_max:np.float64(0.004447370767593384) - rollout_corr/log_ppl_diff_min:np.float64(-0.006148874759674072) - rollout_corr/ppl_ratio:np.float64(1.000251316698268) - rollout_corr/chi2_token:np.float64(0.0009627032559365034) - rollout_corr/chi2_seq:np.float64(0.50378093495965) - actor/pg_loss:np.float64(-1.185922883450985e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(4.797973315362469e-05) - actor/ppo_kl:np.float64(-6.918077795292632e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.048877181485295296) - perf/mfu/actor:np.float64(0.09183496883024235) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.3936767578125) - actor/lr:np.float64(1e-06) - training/global_step:83 - 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:-0.0006229627761058509 - critic/advantages/max:0.75 - critic/advantages/min:-0.375 - critic/returns/mean:-0.0006229627761058509 - critic/returns/max:0.75 - critic/returns/min:-0.375 - response_length/mean:310.38671875 - response_length/max:659.0 - response_length/min:166.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:310.38671875 - response_length_non_aborted/max:659.0 - response_length_non_aborted/min:166.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:391.3125 - prompt_length/max:492.0 - prompt_length/min:267.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.000145619735121727 - timing_s/agent_loop/generate_sequences/min:np.float64(2.076954545453191) - timing_s/agent_loop/generate_sequences/max:np.float64(6.009804593399167) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.534746044242638) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.009804593399167) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/pr
|
||
step:84 - global_seqlen/min:18888 - global_seqlen/max:27836 - global_seqlen/minmax_diff:8948 - global_seqlen/balanced_min:23252 - global_seqlen/balanced_max:23255 - global_seqlen/mean:23253.375 - actor/entropy:0.26086071133613586 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.22949278354644775 - training/rollout_probs_diff_mean:0.0024000119883567095 - training/rollout_probs_diff_std:0.006870268378406763 - training/rollout_actor_probs_pearson_corr:0.9995623230934143 - rollout_corr/rollout_is_mean:1.0000234842300415 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.447716474533081 - rollout_corr/rollout_is_min:0.6345890760421753 - rollout_corr/rollout_is_std:0.023518739268183708 - rollout_corr/rollout_is_eff_sample_size:0.9994470556038549 - rollout_corr/rollout_is_seq_mean:1.0000110864639282 - rollout_corr/rollout_is_seq_std:0.0014413440367206931 - rollout_corr/rollout_is_seq_max:1.0039033889770508 - rollout_corr/rollout_is_seq_min:0.9952189922332764 - rollout_corr/rollout_is_seq_max_deviation:0.004781007766723633 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3491347096860409) - rollout_corr/training_log_ppl:np.float64(0.28971572226146236) - rollout_corr/kl:np.float64(0.00028144666232599036) - rollout_corr/k3_kl:np.float64(0.0003370657866810234) - rollout_corr/rollout_ppl:np.float64(1.348737888969481) - rollout_corr/rollout_log_ppl:np.float64(0.2894342727086041) - rollout_corr/log_ppl_diff:np.float64(0.00028144955285824835) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012165008520241827) - rollout_corr/log_ppl_diff_max:np.float64(0.00624963641166687) - rollout_corr/log_ppl_diff_min:np.float64(-0.004302710294723511) - rollout_corr/ppl_ratio:np.float64(1.000282758846879) - rollout_corr/chi2_token:np.float64(0.0007954011671245098) - rollout_corr/chi2_seq:np.float64(0.2558585754595697) - actor/pg_loss:np.float64(2.7131754904985428e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(3.780518636631314e-05) - actor/ppo_kl:np.float64(-1.1325415965046659e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.03505889978259802) - perf/mfu/actor:np.float64(0.09341460669569832) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.45610427856445) - actor/lr:np.float64(1e-06) - training/global_step:84 - training/epoch:3 - critic/score/mean:0.83203125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.83203125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0018751678289845586 - critic/advantages/max:0.125 - critic/advantages/min:-0.875 - critic/returns/mean:0.0018751678289845586 - critic/returns/max:0.125 - critic/returns/min:-0.875 - response_length/mean:334.60546875 - response_length/max:836.0 - response_length/min:145.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:334.60546875 - response_length_non_aborted/max:836.0 - response_length_non_aborted/min:145.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:392.0625 - prompt_length/max:519.0 - prompt_length/min:264.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00011912919580936432 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8652551472187042) - timing_s/agent_loop/generate_sequences/max:np.float64(7.026093482971191) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.726336318446556) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.026093482971191) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/
|
||
step:85 - global_seqlen/min:19239 - global_seqlen/max:26701 - global_seqlen/minmax_diff:7462 - global_seqlen/balanced_min:22624 - global_seqlen/balanced_max:22640 - global_seqlen/mean:22635.75 - actor/entropy:0.3317025303840637 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6983705163002014 - training/rollout_probs_diff_mean:0.00267635565251112 - training/rollout_probs_diff_std:0.007437863387167454 - training/rollout_actor_probs_pearson_corr:0.999566912651062 - rollout_corr/rollout_is_mean:0.9997298717498779 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.4775165002210997e-05 - rollout_corr/rollout_is_max:1.4265085458755493 - rollout_corr/rollout_is_min:0.20711393654346466 - rollout_corr/rollout_is_std:0.024886781349778175 - rollout_corr/rollout_is_eff_sample_size:0.9993807338411205 - rollout_corr/rollout_is_seq_mean:0.9997522830963135 - rollout_corr/rollout_is_seq_std:0.001566800638101995 - rollout_corr/rollout_is_seq_max:1.006334662437439 - rollout_corr/rollout_is_seq_min:0.995550274848938 - rollout_corr/rollout_is_seq_max_deviation:0.006334662437438965 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4340935898944736) - rollout_corr/training_log_ppl:np.float64(0.34704944033001084) - rollout_corr/kl:np.float64(0.0006850926394577073) - rollout_corr/k3_kl:np.float64(0.00040617002476039943) - rollout_corr/rollout_ppl:np.float64(1.433068967424333) - rollout_corr/rollout_log_ppl:np.float64(0.346364343771711) - rollout_corr/log_ppl_diff:np.float64(0.000685096558299847) - rollout_corr/log_ppl_abs_diff:np.float64(0.001369192890706472) - rollout_corr/log_ppl_diff_max:np.float64(0.005961418151855469) - rollout_corr/log_ppl_diff_min:np.float64(-0.005674391984939575) - rollout_corr/ppl_ratio:np.float64(1.000686737941578) - rollout_corr/chi2_token:np.float64(0.00025137816555798054) - rollout_corr/chi2_seq:np.float64(0.04498665127903223) - actor/pg_loss:np.float64(8.018629159778357e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(6.963335999898845e-05) - actor/ppo_kl:np.float64(0.00010742824364484704) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.09896238800138235) - perf/mfu/actor:np.float64(0.09221646713692358) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.29238510131836) - actor/lr:np.float64(1e-06) - training/global_step:85 - training/epoch:3 - 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:-0.0025595843326300383 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0025595843326300383 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:315.3359375 - response_length/max:920.0 - response_length/min:62.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:315.3359375 - response_length_non_aborted/max:920.0 - response_length_non_aborted/min:62.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:392.03125 - prompt_length/max:541.0 - prompt_length/min:264.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00010436028242111206 - timing_s/agent_loop/generate_sequences/min:np.float64(0.8224622216075659) - timing_s/agent_loop/generate_sequences/max:np.float64(7.511046390980482) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.6148865458744694) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.511046390980482) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/ag
|
||
step:86 - global_seqlen/min:19108 - global_seqlen/max:24507 - global_seqlen/minmax_diff:5399 - global_seqlen/balanced_min:21176 - global_seqlen/balanced_max:21184 - global_seqlen/mean:21181.75 - actor/entropy:0.3438378870487213 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.822213351726532 - training/rollout_probs_diff_mean:0.0030232411809265614 - training/rollout_probs_diff_std:0.008595702238380909 - training/rollout_actor_probs_pearson_corr:0.9994766116142273 - rollout_corr/rollout_is_mean:1.0000633001327515 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:4.151559551246464e-05 - rollout_corr/rollout_is_max:1.739801049232483 - rollout_corr/rollout_is_min:2.8528873372124508e-05 - rollout_corr/rollout_is_std:0.026786576956510544 - rollout_corr/rollout_is_eff_sample_size:0.999282993753028 - rollout_corr/rollout_is_seq_mean:1.000069499015808 - rollout_corr/rollout_is_seq_std:0.0017593996599316597 - rollout_corr/rollout_is_seq_max:1.005941390991211 - rollout_corr/rollout_is_seq_min:0.9954469799995422 - rollout_corr/rollout_is_seq_max_deviation:0.0059413909912109375 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.419941368047148) - rollout_corr/training_log_ppl:np.float64(0.34356227729585953) - rollout_corr/kl:np.float64(0.0003319210262313632) - rollout_corr/k3_kl:np.float64(0.0004817469163640453) - rollout_corr/rollout_ppl:np.float64(1.4194591976702213) - rollout_corr/rollout_log_ppl:np.float64(0.3432303518638946) - rollout_corr/log_ppl_diff:np.float64(0.00033192543196491897) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014975782542023808) - rollout_corr/log_ppl_diff_max:np.float64(0.02202536165714264) - rollout_corr/log_ppl_diff_min:np.float64(-0.005789533257484436) - rollout_corr/ppl_ratio:np.float64(1.0003345324657857) - rollout_corr/chi2_token:np.float64(0.0011146750766783953) - rollout_corr/chi2_seq:np.float64(0.34562928532250226) - actor/pg_loss:np.float64(2.0878855139017105e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(9.801963642530609e-06) - actor/ppo_kl:np.float64(-2.804078054907677e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.03001499082893133) - perf/mfu/actor:np.float64(0.08669465010272584) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.32245254516602) - actor/lr:np.float64(1e-06) - training/global_step:86 - training/epoch:3 - critic/score/mean:0.83203125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.83203125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0003598018374759704 - critic/advantages/max:0.375 - critic/advantages/min:-0.625 - critic/returns/mean:-0.0003598018374759704 - critic/returns/max:0.375 - critic/returns/min:-0.625 - response_length/mean:282.2734375 - response_length/max:971.0 - response_length/min:133.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:282.2734375 - response_length_non_aborted/max:971.0 - response_length_non_aborted/min:133.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:379.65625 - prompt_length/max:483.0 - prompt_length/min:258.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00011705607175827026 - timing_s/agent_loop/generate_sequences/min:np.float64(1.650878168642521) - timing_s/agent_loop/generate_sequences/max:np.float64(7.470661237835884) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.279068757787172) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.470661237835884) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timi
|
||
step:87 - global_seqlen/min:19166 - global_seqlen/max:25802 - global_seqlen/minmax_diff:6636 - global_seqlen/balanced_min:22712 - global_seqlen/balanced_max:22718 - global_seqlen/mean:22715.375 - actor/entropy:0.2796725332736969 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.17467910051345825 - training/rollout_probs_diff_mean:0.002430505584925413 - training/rollout_probs_diff_std:0.006751000415533781 - training/rollout_actor_probs_pearson_corr:0.9996549487113953 - rollout_corr/rollout_is_mean:0.9999507665634155 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.643362045288086 - rollout_corr/rollout_is_min:0.6558393836021423 - rollout_corr/rollout_is_std:0.023985495790839195 - rollout_corr/rollout_is_eff_sample_size:0.9994250267530548 - rollout_corr/rollout_is_seq_mean:0.9999579787254333 - rollout_corr/rollout_is_seq_std:0.0015496320556849241 - rollout_corr/rollout_is_seq_max:1.0060547590255737 - rollout_corr/rollout_is_seq_min:0.9959386587142944 - rollout_corr/rollout_is_seq_max_deviation:0.0060547590255737305 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3713900046423078) - rollout_corr/training_log_ppl:np.float64(0.307053470693063) - rollout_corr/kl:np.float64(0.0003921358480982462) - rollout_corr/k3_kl:np.float64(0.00034856318789877605) - rollout_corr/rollout_ppl:np.float64(1.3708278317935765) - rollout_corr/rollout_log_ppl:np.float64(0.30666133074555546) - rollout_corr/log_ppl_diff:np.float64(0.00039213994750753045) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012159571633674204) - rollout_corr/log_ppl_diff_max:np.float64(0.0056915283203125) - rollout_corr/log_ppl_diff_min:np.float64(-0.003270745277404785) - rollout_corr/ppl_ratio:np.float64(1.0003934551496059) - rollout_corr/chi2_token:np.float64(0.0006380670238286257) - rollout_corr/chi2_seq:np.float64(0.17924676439724863) - actor/pg_loss:np.float64(4.446483217179775e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0) - actor/ppo_kl:np.float64(3.656892177161808e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.04396366095170379) - perf/mfu/actor:np.float64(0.09381662705297268) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.2486572265625) - actor/lr:np.float64(1e-06) - training/global_step:87 - training/epoch:3 - 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:0.003371447790414095 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.003371447790414095 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:297.04296875 - response_length/max:791.0 - response_length/min:86.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:297.04296875 - response_length_non_aborted/max:791.0 - response_length_non_aborted/min:86.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:412.8125 - prompt_length/max:518.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00015601515769958496 - timing_s/agent_loop/generate_sequences/min:np.float64(1.120266143232584) - timing_s/agent_loop/generate_sequences/max:np.float64(6.718985510990024) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.512847683778091) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.718985510990024) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/prompt_length:430 - timin
|
||
step:88 - global_seqlen/min:19663 - global_seqlen/max:25654 - global_seqlen/minmax_diff:5991 - global_seqlen/balanced_min:22652 - global_seqlen/balanced_max:22666 - global_seqlen/mean:22661.25 - actor/entropy:0.2953377664089203 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3797217309474945 - training/rollout_probs_diff_mean:0.0025792568922042847 - training/rollout_probs_diff_std:0.00730118015781045 - training/rollout_actor_probs_pearson_corr:0.999626874923706 - rollout_corr/rollout_is_mean:0.9999725818634033 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.8165900707244873 - rollout_corr/rollout_is_min:0.5354034304618835 - rollout_corr/rollout_is_std:0.025083592161536217 - rollout_corr/rollout_is_eff_sample_size:0.9993710899372364 - rollout_corr/rollout_is_seq_mean:1.0000321865081787 - rollout_corr/rollout_is_seq_std:0.0015428932383656502 - rollout_corr/rollout_is_seq_max:1.00552499294281 - rollout_corr/rollout_is_seq_min:0.9950153231620789 - rollout_corr/rollout_is_seq_max_deviation:0.005524992942810059 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3939012219198048) - rollout_corr/training_log_ppl:np.float64(0.31474355689715594) - rollout_corr/kl:np.float64(0.0004259274350104647) - rollout_corr/k3_kl:np.float64(0.00037728191566088753) - rollout_corr/rollout_ppl:np.float64(1.3932252889499068) - rollout_corr/rollout_log_ppl:np.float64(0.314317626907723) - rollout_corr/log_ppl_diff:np.float64(0.00042592998943291605) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014040337118785828) - rollout_corr/log_ppl_diff_max:np.float64(0.005751430988311768) - rollout_corr/log_ppl_diff_min:np.float64(-0.005627959966659546) - rollout_corr/ppl_ratio:np.float64(1.000427548540756) - rollout_corr/chi2_token:np.float64(0.0006589947734028101) - rollout_corr/chi2_seq:np.float64(0.39225143822841346) - actor/pg_loss:np.float64(-3.2695126719772816e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(8.326313172801747e-05) - actor/ppo_kl:np.float64(0.00012912661738972986) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.07530395220965147) - perf/mfu/actor:np.float64(0.09325487071830077) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.34903717041016) - actor/lr:np.float64(1e-06) - training/global_step:88 - training/epoch:3 - critic/score/mean:0.59765625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.59765625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0029977471567690372 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:0.0029977471567690372 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:322.5078125 - response_length/max:1015.0 - response_length/min:73.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:322.5078125 - response_length_non_aborted/max:1015.0 - response_length_non_aborted/min:73.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:385.65625 - prompt_length/max:490.0 - prompt_length/min:196.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00010713748633861542 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9955077525228262) - timing_s/agent_loop/generate_sequences/max:np.float64(7.942003663629293) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.6675338528002612) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.942003663629293) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slow
|
||
step:89 - global_seqlen/min:21430 - global_seqlen/max:28221 - global_seqlen/minmax_diff:6791 - global_seqlen/balanced_min:24733 - global_seqlen/balanced_max:24735 - global_seqlen/mean:24734.125 - actor/entropy:0.3050276041030884 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7560598850250244 - training/rollout_probs_diff_mean:0.002423952566459775 - training/rollout_probs_diff_std:0.007534638978540897 - training/rollout_actor_probs_pearson_corr:0.9996098875999451 - rollout_corr/rollout_is_mean:0.9999791383743286 - rollout_corr/rollout_is_ratio_fraction_high:1.1024750165233854e-05 - rollout_corr/rollout_is_ratio_fraction_low:3.3074251405196264e-05 - rollout_corr/rollout_is_max:2.385235548019409 - rollout_corr/rollout_is_min:0.14922362565994263 - rollout_corr/rollout_is_std:0.024784782901406288 - rollout_corr/rollout_is_eff_sample_size:0.9993859725829729 - rollout_corr/rollout_is_seq_mean:0.9999644160270691 - rollout_corr/rollout_is_seq_std:0.0017730066319927573 - rollout_corr/rollout_is_seq_max:1.0053385496139526 - rollout_corr/rollout_is_seq_min:0.9937400221824646 - rollout_corr/rollout_is_seq_max_deviation:0.0062599778175354 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4235105477273464) - rollout_corr/training_log_ppl:np.float64(0.3350572675553849) - rollout_corr/kl:np.float64(0.0004977377876294042) - rollout_corr/k3_kl:np.float64(0.00043506097114232034) - rollout_corr/rollout_ppl:np.float64(1.4227662342600524) - rollout_corr/rollout_log_ppl:np.float64(0.33455952878284734) - rollout_corr/log_ppl_diff:np.float64(0.0004977387725375593) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014161956496536732) - rollout_corr/log_ppl_diff_max:np.float64(0.006734251976013184) - rollout_corr/log_ppl_diff_min:np.float64(-0.004191935062408447) - rollout_corr/ppl_ratio:np.float64(1.000499605666846) - rollout_corr/chi2_token:np.float64(0.0007642176933586597) - rollout_corr/chi2_seq:np.float64(0.3206390021368861) - actor/pg_loss:np.float64(5.431764293462038e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(9.499900716036791e-05) - actor/ppo_kl:np.float64(0.00011755074425323642) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.21098365262150764) - perf/mfu/actor:np.float64(0.09995132241417987) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.35758590698242) - actor/lr:np.float64(1e-06) - training/global_step:89 - 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:0.0021939254365861416 - critic/advantages/max:0.375 - critic/advantages/min:-0.875 - critic/returns/mean:0.0021939254365861416 - critic/returns/max:0.375 - critic/returns/min:-0.875 - response_length/mean:354.31640625 - response_length/max:789.0 - response_length/min:141.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:354.31640625 - response_length_non_aborted/max:789.0 - response_length_non_aborted/min:141.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:418.625 - prompt_length/max:512.0 - prompt_length/min:264.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:9.185262024402618e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.6536885611712933) - timing_s/agent_loop/generate_sequences/max:np.float64(6.9988903403282166) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.955004553521576) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(6.9988903403282166) - timing_s/agent_loop/slowest/tool_calls:np.float64(
|
||
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 given crystal structure is a unit cell with lattice parameters a = 7.82 Å, b = 7.82 Å, c = 17.88 Å, and all angles α, β, γ are 90°, which indicates that the unit cell is orthorhombic.
|
||
|
||
For an orthorhombic unit cell, the volume (V) of the unit cell is calculated using the formula:
|
||
|
||
V = a × b × c
|
||
|
||
Substituting the given values:
|
||
|
||
V = 7.82 Å × 7.82 Å × 17.88 Å
|
||
|
||
First, calculate 7.82 × 7.82:
|
||
|
||
7.82 × 7.82 = 61.1524
|
||
|
||
Then multiply that 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-GRPO-Qwen-Qwen3-8B-mbs8-train32-rollout8-lr1e-6-vllm0.8/global_step_90
|
||
step:90 - global_seqlen/min:20047 - global_seqlen/max:28634 - global_seqlen/minmax_diff:8587 - global_seqlen/balanced_min:23235 - global_seqlen/balanced_max:23239 - global_seqlen/mean:23238.25 - actor/entropy:0.34761369228363037 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2881855070590973 - training/rollout_probs_diff_mean:0.002914301585406065 - training/rollout_probs_diff_std:0.00754848076030612 - training/rollout_actor_probs_pearson_corr:0.9996027946472168 - rollout_corr/rollout_is_mean:0.9999606609344482 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.3682652681600302e-05 - rollout_corr/rollout_is_max:1.6418794393539429 - rollout_corr/rollout_is_min:0.3688293397426605 - rollout_corr/rollout_is_std:0.026443934068083763 - rollout_corr/rollout_is_eff_sample_size:0.9993012069606744 - rollout_corr/rollout_is_seq_mean:0.9999961256980896 - rollout_corr/rollout_is_seq_std:0.0017832440789788961 - rollout_corr/rollout_is_seq_max:1.0080827474594116 - rollout_corr/rollout_is_seq_min:0.9929141402244568 - rollout_corr/rollout_is_seq_max_deviation:0.008082747459411621 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4519235398620367) - rollout_corr/training_log_ppl:np.float64(0.36163410761218984) - rollout_corr/kl:np.float64(0.0004622778555010143) - rollout_corr/k3_kl:np.float64(0.0004612364294018789) - rollout_corr/rollout_ppl:np.float64(1.451245247386396) - rollout_corr/rollout_log_ppl:np.float64(0.36117182987800334) - rollout_corr/log_ppl_diff:np.float64(0.0004622777341865003) - rollout_corr/log_ppl_abs_diff:np.float64(0.001413370599038899) - rollout_corr/log_ppl_diff_max:np.float64(0.008033037185668945) - rollout_corr/log_ppl_diff_min:np.float64(-0.007330834865570068) - rollout_corr/ppl_ratio:np.float64(1.0004640952683985) - rollout_corr/chi2_token:np.float64(0.0009107908699661493) - rollout_corr/chi2_seq:np.float64(0.3217647692654282) - actor/pg_loss:np.float64(1.382431946694851e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(1.2271325431356672e-05) - actor/ppo_kl:np.float64(8.642425431104606e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.01626203558407724) - perf/mfu/actor:np.float64(0.09550263987107718) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.47405242919922) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7593085106382979) - val-aux/sciknoweval/reward/std@16:np.float64(0.05324482498876583) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7813297872340427) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.033221983145840114) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.7378085106382979) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.056841760464986006) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7595957446808509) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.0533915093622308) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7919361702127661) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.015034088527283876) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.7136276595744682) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.053166060035626864) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.766095744680851) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.0424948683594337) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7966170212765958) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.0047140238339681995) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6905531914893618) - val-aux/sciknoweval/reward/worst@8/std:np.float64(0.04196233640540949) - val-aux/sciknoweval/reward/maj@8/mean:np.float64(0.7699468085106383) - val-aux/sciknoweval/reward/maj@8/std:np.float64(0.028759654583830934) - val-aux/sciknoweval/reward/best@16/mean:np.float64(0.7978191489361702) - val-aux/s
|
||
step:91 - global_seqlen/min:21635 - global_seqlen/max:25466 - global_seqlen/minmax_diff:3831 - global_seqlen/balanced_min:23610 - global_seqlen/balanced_max:23624 - global_seqlen/mean:23619.75 - actor/entropy:0.35342922806739807 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.21307647228240967 - training/rollout_probs_diff_mean:0.0028058807365596294 - training/rollout_probs_diff_std:0.007240186911076307 - training/rollout_actor_probs_pearson_corr:0.9996463656425476 - rollout_corr/rollout_is_mean:0.999942421913147 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6107667684555054 - rollout_corr/rollout_is_min:0.6440470814704895 - rollout_corr/rollout_is_std:0.025918031111359596 - rollout_corr/rollout_is_eff_sample_size:0.9993284684924016 - rollout_corr/rollout_is_seq_mean:0.999967098236084 - rollout_corr/rollout_is_seq_std:0.0016982742818072438 - rollout_corr/rollout_is_seq_max:1.0109471082687378 - rollout_corr/rollout_is_seq_min:0.9943731427192688 - rollout_corr/rollout_is_seq_max_deviation:0.010947108268737793 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5175115521997213) - rollout_corr/training_log_ppl:np.float64(0.39346077786467504) - rollout_corr/kl:np.float64(0.0004233603360290772) - rollout_corr/k3_kl:np.float64(0.00041688081708457503) - rollout_corr/rollout_ppl:np.float64(1.5167877241037786) - rollout_corr/rollout_log_ppl:np.float64(0.3930374163610395) - rollout_corr/log_ppl_diff:np.float64(0.00042336150363553315) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012970607058377936) - rollout_corr/log_ppl_diff_max:np.float64(0.007829904556274414) - rollout_corr/log_ppl_diff_min:np.float64(-0.010272711515426636) - rollout_corr/ppl_ratio:np.float64(1.0004250255879015) - rollout_corr/chi2_token:np.float64(0.0008352859877049923) - rollout_corr/chi2_seq:np.float64(0.39926669048145413) - actor/pg_loss:np.float64(-0.00021755159832537174) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(4.048991831950843e-05) - actor/ppo_kl:np.float64(3.1574491124786164e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.13898340100422502) - perf/mfu/actor:np.float64(0.09595114108973381) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.60107803344727) - actor/lr:np.float64(1e-06) - training/global_step:91 - training/epoch:3 - critic/score/mean:0.84765625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.84765625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0018889433704316616 - critic/advantages/max:0.375 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0018889433704316616 - critic/returns/max:0.375 - critic/returns/min:-0.875 - response_length/mean:382.0546875 - response_length/max:873.0 - response_length/min:67.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:382.0546875 - response_length_non_aborted/max:873.0 - response_length_non_aborted/min:67.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:356.0625 - prompt_length/max:477.0 - prompt_length/min:196.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:6.923265755176544e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(0.8506027832627296) - timing_s/agent_loop/generate_sequences/max:np.float64(7.396001921966672) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.8466681276404415) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.396001921966672) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/
|
||
step:92 - global_seqlen/min:22074 - global_seqlen/max:29852 - global_seqlen/minmax_diff:7778 - global_seqlen/balanced_min:24847 - global_seqlen/balanced_max:24849 - global_seqlen/mean:24848.0 - actor/entropy:0.3595642149448395 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.22151517868041992 - training/rollout_probs_diff_mean:0.0029321240726858377 - training/rollout_probs_diff_std:0.00750844506546855 - training/rollout_actor_probs_pearson_corr:0.9995761513710022 - rollout_corr/rollout_is_mean:1.0001015663146973 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5213674306869507 - rollout_corr/rollout_is_min:0.6559000015258789 - rollout_corr/rollout_is_std:0.026828821748495102 - rollout_corr/rollout_is_eff_sample_size:0.9992808510663643 - rollout_corr/rollout_is_seq_mean:1.0001308917999268 - rollout_corr/rollout_is_seq_std:0.0014905876014381647 - rollout_corr/rollout_is_seq_max:1.0047581195831299 - rollout_corr/rollout_is_seq_min:0.9957191944122314 - rollout_corr/rollout_is_seq_max_deviation:0.004758119583129883 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4703190927393734) - rollout_corr/training_log_ppl:np.float64(0.3702059087372618) - rollout_corr/kl:np.float64(0.00043767166106079003) - rollout_corr/k3_kl:np.float64(0.0004534227082331199) - rollout_corr/rollout_ppl:np.float64(1.4696018886752427) - rollout_corr/rollout_log_ppl:np.float64(0.369768237680546) - rollout_corr/log_ppl_diff:np.float64(0.00043767105671577156) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013431093539111316) - rollout_corr/log_ppl_diff_max:np.float64(0.00604403018951416) - rollout_corr/log_ppl_diff_min:np.float64(-0.004392862319946289) - rollout_corr/ppl_ratio:np.float64(1.0004392443224788) - rollout_corr/chi2_token:np.float64(0.0009216954931616783) - rollout_corr/chi2_seq:np.float64(0.4805670231580734) - actor/pg_loss:np.float64(4.529592115432024e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(7.733584970992524e-05) - actor/ppo_kl:np.float64(0.000197046403414447) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.11538068484514952) - perf/mfu/actor:np.float64(0.100455938692909) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.47601699829102) - actor/lr:np.float64(1e-06) - training/global_step:92 - training/epoch:3 - 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:-0.006862935610115528 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.006862935610115528 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:382.0625 - response_length/max:908.0 - response_length/min:177.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:382.0625 - response_length_non_aborted/max:908.0 - response_length_non_aborted/min:177.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:394.4375 - prompt_length/max:476.0 - prompt_length/min:264.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00011311471462249756 - timing_s/agent_loop/generate_sequences/min:np.float64(2.233831040561199) - timing_s/agent_loop/generate_sequences/max:np.float64(7.89228375069797) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.414751171891112) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.89228375069797) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/prompt_lengt
|
||
step:93 - global_seqlen/min:20407 - global_seqlen/max:28006 - global_seqlen/minmax_diff:7599 - global_seqlen/balanced_min:24042 - global_seqlen/balanced_max:24044 - global_seqlen/mean:24043.25 - actor/entropy:0.3005387485027313 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4241411089897156 - training/rollout_probs_diff_mean:0.002642803592607379 - training/rollout_probs_diff_std:0.007475018501281738 - training/rollout_actor_probs_pearson_corr:0.9995802044868469 - rollout_corr/rollout_is_mean:1.000084638595581 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.0518122508074157e-05 - rollout_corr/rollout_is_max:1.5421065092086792 - rollout_corr/rollout_is_min:0.4320620894432068 - rollout_corr/rollout_is_std:0.025235213339328766 - rollout_corr/rollout_is_eff_sample_size:0.9993635892517825 - rollout_corr/rollout_is_seq_mean:1.0000985860824585 - rollout_corr/rollout_is_seq_std:0.0016015670262277126 - rollout_corr/rollout_is_seq_max:1.0049409866333008 - rollout_corr/rollout_is_seq_min:0.9953435659408569 - rollout_corr/rollout_is_seq_max_deviation:0.004940986633300781 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4219031422398984) - rollout_corr/training_log_ppl:np.float64(0.3349203703983221) - rollout_corr/kl:np.float64(0.00030562687818047607) - rollout_corr/k3_kl:np.float64(0.00046132832230227905) - rollout_corr/rollout_ppl:np.float64(1.421450737863779) - rollout_corr/rollout_log_ppl:np.float64(0.334614738705568) - rollout_corr/log_ppl_diff:np.float64(0.00030563169275410473) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013734388921875507) - rollout_corr/log_ppl_diff_max:np.float64(0.007729679346084595) - rollout_corr/log_ppl_diff_min:np.float64(-0.004610300064086914) - rollout_corr/ppl_ratio:np.float64(1.0003073420375586) - rollout_corr/chi2_token:np.float64(0.001192569499835372) - rollout_corr/chi2_seq:np.float64(0.4643619789276272) - actor/pg_loss:np.float64(7.481081411242485e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(4.265500319888815e-05) - actor/ppo_kl:np.float64(4.5113914802996646e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.12982227839529514) - perf/mfu/actor:np.float64(0.0976179469248955) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.53211975097656) - actor/lr:np.float64(1e-06) - training/global_step:93 - training/epoch:3 - 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:0.003967961762100458 - critic/advantages/max:0.375 - critic/advantages/min:-0.875 - critic/returns/mean:0.003967961762100458 - critic/returns/max:0.375 - critic/returns/min:-0.875 - response_length/mean:371.3828125 - response_length/max:898.0 - response_length/min:165.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:371.3828125 - response_length_non_aborted/max:898.0 - response_length_non_aborted/min:165.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:379.96875 - prompt_length/max:492.0 - prompt_length/min:203.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:8.372776210308075e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.136307967826724) - timing_s/agent_loop/generate_sequences/max:np.float64(7.731545675545931) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.174539480387466) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.731545675545931) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_
|
||
step:94 - global_seqlen/min:21534 - global_seqlen/max:29660 - global_seqlen/minmax_diff:8126 - global_seqlen/balanced_min:25407 - global_seqlen/balanced_max:25422 - global_seqlen/mean:25417.125 - actor/entropy:0.2767510414123535 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24135532975196838 - training/rollout_probs_diff_mean:0.0025500222109258175 - training/rollout_probs_diff_std:0.0073505425825715065 - training/rollout_actor_probs_pearson_corr:0.9995822310447693 - rollout_corr/rollout_is_mean:0.9999530911445618 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:9.667346603237092e-06 - rollout_corr/rollout_is_max:1.9516299962997437 - rollout_corr/rollout_is_min:0.2820626199245453 - rollout_corr/rollout_is_std:0.024969272315502167 - rollout_corr/rollout_is_eff_sample_size:0.999376804820774 - rollout_corr/rollout_is_seq_mean:0.9999873042106628 - rollout_corr/rollout_is_seq_std:0.0015164752257987857 - rollout_corr/rollout_is_seq_max:1.0061312913894653 - rollout_corr/rollout_is_seq_min:0.9966298937797546 - rollout_corr/rollout_is_seq_max_deviation:0.006131291389465332 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3853449211455882) - rollout_corr/training_log_ppl:np.float64(0.3129881990898866) - rollout_corr/kl:np.float64(0.0004005604543886321) - rollout_corr/k3_kl:np.float64(0.0004372991470091847) - rollout_corr/rollout_ppl:np.float64(1.3847369984723628) - rollout_corr/rollout_log_ppl:np.float64(0.3125876357080415) - rollout_corr/log_ppl_diff:np.float64(0.0004005633818451315) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013818044390063733) - rollout_corr/log_ppl_diff_max:np.float64(0.005264222621917725) - rollout_corr/log_ppl_diff_min:np.float64(-0.005869626998901367) - rollout_corr/ppl_ratio:np.float64(1.000402232632041) - rollout_corr/chi2_token:np.float64(0.0009447787888348103) - rollout_corr/chi2_seq:np.float64(0.45626901066862047) - actor/pg_loss:np.float64(2.013996709138155e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0001884714770312712) - actor/ppo_kl:np.float64(3.779682675641993e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.13156491424888372) - perf/mfu/actor:np.float64(0.10275564150104508) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.4635009765625) - actor/lr:np.float64(1e-06) - training/global_step:94 - training/epoch:3 - 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:0.000761303526815027 - critic/advantages/max:0.875 - critic/advantages/min:-0.5 - critic/returns/mean:0.000761303526815027 - critic/returns/max:0.875 - critic/returns/min:-0.5 - response_length/mean:404.06640625 - response_length/max:928.0 - response_length/min:70.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:404.06640625 - response_length_non_aborted/max:928.0 - response_length_non_aborted/min:70.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:390.21875 - prompt_length/max:483.0 - prompt_length/min:200.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:0.00010003708302974701 - timing_s/agent_loop/generate_sequences/min:np.float64(0.8698799815028906) - timing_s/agent_loop/generate_sequences/max:np.float64(8.220686331391335) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.46872838906711) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.220686331391335) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent
|
||
step:95 - global_seqlen/min:22838 - global_seqlen/max:27064 - global_seqlen/minmax_diff:4226 - global_seqlen/balanced_min:25386 - global_seqlen/balanced_max:25389 - global_seqlen/mean:25387.875 - actor/entropy:0.2577309012413025 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.893662691116333 - training/rollout_probs_diff_mean:0.0024719759821891785 - training/rollout_probs_diff_std:0.007829725742340088 - training/rollout_actor_probs_pearson_corr:0.9994516968727112 - rollout_corr/rollout_is_mean:1.0001028776168823 - rollout_corr/rollout_is_ratio_fraction_high:9.833324838837143e-06 - rollout_corr/rollout_is_ratio_fraction_low:3.933329935534857e-05 - rollout_corr/rollout_is_max:2.013819932937622 - rollout_corr/rollout_is_min:0.012144971638917923 - rollout_corr/rollout_is_std:0.024124257266521454 - rollout_corr/rollout_is_eff_sample_size:0.9994184778225121 - rollout_corr/rollout_is_seq_mean:1.0001139640808105 - rollout_corr/rollout_is_seq_std:0.0013081853976473212 - rollout_corr/rollout_is_seq_max:1.0043644905090332 - rollout_corr/rollout_is_seq_min:0.9966682195663452 - rollout_corr/rollout_is_seq_max_deviation:0.004364490509033203 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3321691052988172) - rollout_corr/training_log_ppl:np.float64(0.2772668176476145) - rollout_corr/kl:np.float64(0.0003558635719524972) - rollout_corr/k3_kl:np.float64(0.00039730372719759544) - rollout_corr/rollout_ppl:np.float64(1.3316714563407004) - rollout_corr/rollout_log_ppl:np.float64(0.27691095299087465) - rollout_corr/log_ppl_diff:np.float64(0.0003558646567398682) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012297478533582762) - rollout_corr/log_ppl_diff_max:np.float64(0.007804989814758301) - rollout_corr/log_ppl_diff_min:np.float64(-0.0037243664264678955) - rollout_corr/ppl_ratio:np.float64(1.000357209937647) - rollout_corr/chi2_token:np.float64(0.0008305651135742664) - rollout_corr/chi2_seq:np.float64(0.3522763052023947) - actor/pg_loss:np.float64(-4.14624810218811e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(6.430264829759835e-05) - actor/ppo_kl:np.float64(0.0001356113971553441) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.09277421608567238) - perf/mfu/actor:np.float64(0.10337933781473602) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.35813522338867) - actor/lr:np.float64(1e-06) - training/global_step:95 - 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:0.0005334579036571085 - critic/advantages/max:0.5 - critic/advantages/min:-0.875 - critic/returns/mean:0.0005334579036571085 - critic/returns/max:0.5 - critic/returns/min:-0.875 - response_length/mean:397.24609375 - response_length/max:889.0 - response_length/min:170.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:397.24609375 - response_length_non_aborted/max:889.0 - response_length_non_aborted/min:170.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:396.125 - prompt_length/max:493.0 - prompt_length/min:260.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:9.716302156448364e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.2121583987027407) - timing_s/agent_loop/generate_sequences/max:np.float64(7.992377787828445) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.505745570415456) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.992377787828445) - timing_s/agent_loop/slowest/tool_calls:np.float64(0
|
||
step:96 - global_seqlen/min:21482 - global_seqlen/max:29626 - global_seqlen/minmax_diff:8144 - global_seqlen/balanced_min:26147 - global_seqlen/balanced_max:26151 - global_seqlen/mean:26149.375 - actor/entropy:0.27476727962493896 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2821924090385437 - training/rollout_probs_diff_mean:0.002515679458156228 - training/rollout_probs_diff_std:0.007363538723438978 - training/rollout_actor_probs_pearson_corr:0.9994983077049255 - rollout_corr/rollout_is_mean:1.0000098943710327 - rollout_corr/rollout_is_ratio_fraction_high:1.907468686113134e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:3.167215347290039 - rollout_corr/rollout_is_min:0.569692075252533 - rollout_corr/rollout_is_std:0.024926267564296722 - rollout_corr/rollout_is_eff_sample_size:0.9993789479189513 - rollout_corr/rollout_is_seq_mean:1.0000368356704712 - rollout_corr/rollout_is_seq_std:0.0015217246254906058 - rollout_corr/rollout_is_seq_max:1.0109775066375732 - rollout_corr/rollout_is_seq_min:0.9955570101737976 - rollout_corr/rollout_is_seq_max_deviation:0.010977506637573242 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3714127410203218) - rollout_corr/training_log_ppl:np.float64(0.30417355547251645) - rollout_corr/kl:np.float64(0.0004562080985355266) - rollout_corr/k3_kl:np.float64(0.0004123068538604002) - rollout_corr/rollout_ppl:np.float64(1.370782433077693) - rollout_corr/rollout_log_ppl:np.float64(0.30371734405343886) - rollout_corr/log_ppl_diff:np.float64(0.0004562114190775901) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011979831324424595) - rollout_corr/log_ppl_diff_max:np.float64(0.0051811933517456055) - rollout_corr/log_ppl_diff_min:np.float64(-0.006937205791473389) - rollout_corr/ppl_ratio:np.float64(1.000457545509562) - rollout_corr/chi2_token:np.float64(0.0007856255397200584) - rollout_corr/chi2_seq:np.float64(0.5816815719008446) - actor/pg_loss:np.float64(-5.977042019367218e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(2.639629201439675e-05) - actor/ppo_kl:np.float64(0.0001432839844417444) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.04847895633429289) - perf/mfu/actor:np.float64(0.10562419744407212) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.48655700683594) - actor/lr:np.float64(1e-06) - training/global_step:96 - training/epoch:3 - 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:0.0001442523207515478 - critic/advantages/max:0.375 - critic/advantages/min:-0.625 - critic/returns/mean:0.0001442523207515478 - critic/returns/max:0.375 - critic/returns/min:-0.625 - response_length/mean:409.57421875 - response_length/max:843.0 - response_length/min:170.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:409.57421875 - response_length_non_aborted/max:843.0 - response_length_non_aborted/min:170.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:407.59375 - prompt_length/max:505.0 - prompt_length/min:263.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:7.192417979240417e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.067355601117015) - timing_s/agent_loop/generate_sequences/max:np.float64(7.962549405172467) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.60729234876635) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.962549405172467) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s
|
||
step:97 - global_seqlen/min:20859 - global_seqlen/max:29785 - global_seqlen/minmax_diff:8926 - global_seqlen/balanced_min:24566 - global_seqlen/balanced_max:24575 - global_seqlen/mean:24572.5 - actor/entropy:0.28956183791160583 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3534809648990631 - training/rollout_probs_diff_mean:0.002682067221030593 - training/rollout_probs_diff_std:0.007446713745594025 - training/rollout_actor_probs_pearson_corr:0.9995591044425964 - rollout_corr/rollout_is_mean:0.9999961256980896 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.936397671699524 - rollout_corr/rollout_is_min:0.5016483664512634 - rollout_corr/rollout_is_std:0.025280050933361053 - rollout_corr/rollout_is_eff_sample_size:0.9993613271623948 - rollout_corr/rollout_is_seq_mean:1.0000274181365967 - rollout_corr/rollout_is_seq_std:0.0013985984260216355 - rollout_corr/rollout_is_seq_max:1.0056803226470947 - rollout_corr/rollout_is_seq_min:0.9960852861404419 - rollout_corr/rollout_is_seq_max_deviation:0.0056803226470947266 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3826120416633785) - rollout_corr/training_log_ppl:np.float64(0.3127178924914915) - rollout_corr/kl:np.float64(0.0003792919155714003) - rollout_corr/k3_kl:np.float64(0.00042534956372719535) - rollout_corr/rollout_ppl:np.float64(1.382072074804455) - rollout_corr/rollout_log_ppl:np.float64(0.3123385964281624) - rollout_corr/log_ppl_diff:np.float64(0.00037929606332909316) - rollout_corr/log_ppl_abs_diff:np.float64(0.00126680287939962) - rollout_corr/log_ppl_diff_max:np.float64(0.006162881851196289) - rollout_corr/log_ppl_diff_min:np.float64(-0.004757493734359741) - rollout_corr/ppl_ratio:np.float64(1.0003806476015598) - rollout_corr/chi2_token:np.float64(0.0009697633795440197) - rollout_corr/chi2_seq:np.float64(0.3747521855402738) - actor/pg_loss:np.float64(-1.547869760543108e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(4.352384894446004e-05) - actor/ppo_kl:np.float64(6.558593235084231e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.08563856221735477) - perf/mfu/actor:np.float64(0.09803966351839774) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.51029205322266) - actor/lr:np.float64(1e-06) - training/global_step:97 - training/epoch:3 - 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:-0.003381997812539339 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:-0.003381997812539339 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:401.078125 - response_length/max:956.0 - response_length/min:159.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:401.078125 - response_length_non_aborted/max:956.0 - response_length_non_aborted/min:159.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:366.8125 - prompt_length/max:490.0 - prompt_length/min:262.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:2.434663474559784e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.1238491851836443) - timing_s/agent_loop/generate_sequences/max:np.float64(8.404852511361241) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.601780556979065) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.404852511361241) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/pro
|
||
step:98 - global_seqlen/min:21789 - global_seqlen/max:28252 - global_seqlen/minmax_diff:6463 - global_seqlen/balanced_min:23858 - global_seqlen/balanced_max:23860 - global_seqlen/mean:23859.5 - actor/entropy:0.36053964495658875 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18463629484176636 - training/rollout_probs_diff_mean:0.003115718485787511 - training/rollout_probs_diff_std:0.007648924365639687 - training/rollout_actor_probs_pearson_corr:0.9995946288108826 - rollout_corr/rollout_is_mean:0.9998584389686584 - rollout_corr/rollout_is_ratio_fraction_high:1.1389002793293912e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.1389002793293912e-05 - rollout_corr/rollout_is_max:2.498016595840454 - rollout_corr/rollout_is_min:0.2521287500858307 - rollout_corr/rollout_is_std:0.027102908119559288 - rollout_corr/rollout_is_eff_sample_size:0.9992657334826961 - rollout_corr/rollout_is_seq_mean:0.9998365044593811 - rollout_corr/rollout_is_seq_std:0.0016116970218718052 - rollout_corr/rollout_is_seq_max:1.0038925409317017 - rollout_corr/rollout_is_seq_min:0.9945869445800781 - rollout_corr/rollout_is_seq_max_deviation:0.005413055419921875 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.469639522023499) - rollout_corr/training_log_ppl:np.float64(0.3758351509895874) - rollout_corr/kl:np.float64(0.0006500774032822676) - rollout_corr/k3_kl:np.float64(0.0005059725597362785) - rollout_corr/rollout_ppl:np.float64(1.4686928312294185) - rollout_corr/rollout_log_ppl:np.float64(0.3751850715925684) - rollout_corr/log_ppl_diff:np.float64(0.0006500793970189989) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014755238080397248) - rollout_corr/log_ppl_diff_max:np.float64(0.006156846880912781) - rollout_corr/log_ppl_diff_min:np.float64(-0.004006743431091309) - rollout_corr/ppl_ratio:np.float64(1.0006518627051264) - rollout_corr/chi2_token:np.float64(0.002016712911427021) - rollout_corr/chi2_seq:np.float64(0.6680954364128411) - actor/pg_loss:np.float64(-2.2069783881306648e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0) - actor/ppo_kl:np.float64(9.759571342371487e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.060099998489022255) - perf/mfu/actor:np.float64(0.09587603943024309) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.60213470458984) - actor/lr:np.float64(1e-06) - training/global_step:98 - training/epoch:3 - critic/score/mean:0.9296875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.9296875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.001405118266120553 - critic/advantages/max:0.125 - critic/advantages/min:-0.875 - critic/returns/mean:0.001405118266120553 - critic/returns/max:0.125 - critic/returns/min:-0.875 - response_length/mean:342.984375 - response_length/max:814.0 - response_length/min:92.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:342.984375 - response_length_non_aborted/max:814.0 - response_length_non_aborted/min:92.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:402.625 - prompt_length/max:486.0 - prompt_length/min:263.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:9.733065962791443e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1825728360563517) - timing_s/agent_loop/generate_sequences/max:np.float64(7.279220985248685) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.037709981312219) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(7.279220985248685) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_lo
|
||
step:99 - global_seqlen/min:20296 - global_seqlen/max:26640 - global_seqlen/minmax_diff:6344 - global_seqlen/balanced_min:24543 - global_seqlen/balanced_max:24549 - global_seqlen/mean:24547.25 - actor/entropy:0.37918776273727417 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24402102828025818 - training/rollout_probs_diff_mean:0.002885072724893689 - training/rollout_probs_diff_std:0.007192612625658512 - training/rollout_actor_probs_pearson_corr:0.9996479749679565 - rollout_corr/rollout_is_mean:0.9999138116836548 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.599069595336914 - rollout_corr/rollout_is_min:0.7051597237586975 - rollout_corr/rollout_is_std:0.025952504947781563 - rollout_corr/rollout_is_eff_sample_size:0.9993268018058533 - rollout_corr/rollout_is_seq_mean:0.9999309182167053 - rollout_corr/rollout_is_seq_std:0.0014725772198289633 - rollout_corr/rollout_is_seq_max:1.0040785074234009 - rollout_corr/rollout_is_seq_min:0.9944436550140381 - rollout_corr/rollout_is_seq_max_deviation:0.005556344985961914 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.520531804766506) - rollout_corr/training_log_ppl:np.float64(0.40119856897217687) - rollout_corr/kl:np.float64(0.00043854576770385734) - rollout_corr/k3_kl:np.float64(0.00043854525202391414) - rollout_corr/rollout_ppl:np.float64(1.5198842682875693) - rollout_corr/rollout_log_ppl:np.float64(0.40076001826673746) - rollout_corr/log_ppl_diff:np.float64(0.0004385507054394111) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013136765192029998) - rollout_corr/log_ppl_diff_max:np.float64(0.005993157625198364) - rollout_corr/log_ppl_diff_min:np.float64(-0.004958093166351318) - rollout_corr/ppl_ratio:np.float64(1.0004400596953928) - rollout_corr/chi2_token:np.float64(0.0009022736921906471) - rollout_corr/chi2_seq:np.float64(0.3876523340586573) - actor/pg_loss:np.float64(2.1395273506641388e-05) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(0.0) - actor/ppo_kl:np.float64(1.1412551813805294e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.07385709136724472) - perf/mfu/actor:np.float64(0.09943231597237594) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.54281997680664) - actor/lr:np.float64(1e-06) - training/global_step:99 - training/epoch:3 - critic/score/mean:0.76171875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.76171875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0004624001740012318 - critic/advantages/max:0.625 - critic/advantages/min:-0.375 - critic/returns/mean:0.0004624001740012318 - critic/returns/max:0.625 - critic/returns/min:-0.375 - response_length/mean:372.7578125 - response_length/max:945.0 - response_length/min:182.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:372.7578125 - response_length_non_aborted/max:945.0 - response_length_non_aborted/min:182.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:394.34375 - prompt_length/max:480.0 - prompt_length/min:197.0 - prompt_length/clip_ratio:0.0 - num_turns/min:np.int32(2) - num_turns/max:np.int32(2) - num_turns/mean:np.float64(2.0) - timing_s/start_profile:2.4596229195594788e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.3601644691079855) - timing_s/agent_loop/generate_sequences/max:np.float64(8.149387737736106) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.401201009575743) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool_calls/max:np.float64(0.0) - timing_s/agent_loop/tool_calls/mean:np.float64(0.0) - timing_s/agent_loop/slowest/generate_sequences:np.float64(8.149387737736106) - timing_s/agent_loop/slowest/tool_calls:np.float64(0.0) - timing_s/agent_loop/slowest/prompt_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': 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 given crystal structure has lattice parameters a = 7.82 Å, b = 7.82 Å, c = 17.88 Å, and all angles α, β, γ are 90.00º. This indicates that the crystal structure is orthorhombic, as all angles are 90 degrees and the lattice parameters are not equal.
|
||
|
||
The volume of the unit cell in a crystal structure is calculated using the formula:
|
||
|
||
$$ V = a \times b \times c $$
|
||
|
||
Plugging in the values:
|
||
$$ V = 7.82 \times 7.82 \times 17.88 $$
|
||
|
||
First, calculate $ 7.82 \times 7.82 $:
|
||
$$ 7.82 \times 7.82 = 61.1524 $$
|
||
|
||
Now multiply that result by 17.88:
|
||
$$ 61.1524 \times 17.88 = 1094.31 $$
|
||
|
||
Therefore, the volume of the unit cell is approximately 1094.31 ų, which corresponds to option B.
|
||
</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-GRPO-Qwen-Qwen3-8B-mbs8-train32-rollout8-lr1e-6-vllm0.8/global_step_100
|
||
step:100 - global_seqlen/min:20619 - global_seqlen/max:28642 - global_seqlen/minmax_diff:8023 - global_seqlen/balanced_min:24226 - global_seqlen/balanced_max:24234 - global_seqlen/mean:24231.0 - actor/entropy:0.41687414050102234 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2016180157661438 - training/rollout_probs_diff_mean:0.0030894633382558823 - training/rollout_probs_diff_std:0.007387364748865366 - training/rollout_actor_probs_pearson_corr:0.9996216893196106 - rollout_corr/rollout_is_mean:0.9999920725822449 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6438628435134888 - rollout_corr/rollout_is_min:0.5598989725112915 - rollout_corr/rollout_is_std:0.02741776406764984 - rollout_corr/rollout_is_eff_sample_size:0.9992487118690302 - rollout_corr/rollout_is_seq_mean:0.9999960064888 - rollout_corr/rollout_is_seq_std:0.0014865802368149161 - rollout_corr/rollout_is_seq_max:1.0055478811264038 - rollout_corr/rollout_is_seq_min:0.9939270615577698 - rollout_corr/rollout_is_seq_max_deviation:0.006072938442230225 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5640644407831132) - rollout_corr/training_log_ppl:np.float64(0.4244483573420439) - rollout_corr/kl:np.float64(0.0003323329324800284) - rollout_corr/k3_kl:np.float64(0.0004623864396933186) - rollout_corr/rollout_ppl:np.float64(1.5636177295818925) - rollout_corr/rollout_log_ppl:np.float64(0.4241160247474909) - rollout_corr/log_ppl_diff:np.float64(0.0003323325945530087) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012461927544791251) - rollout_corr/log_ppl_diff_max:np.float64(0.006919294595718384) - rollout_corr/log_ppl_diff_min:np.float64(-0.004684805870056152) - rollout_corr/ppl_ratio:np.float64(1.00033376715146) - rollout_corr/chi2_token:np.float64(0.0012154490686953068) - rollout_corr/chi2_seq:np.float64(0.4026260869577527) - actor/pg_loss:np.float64(-1.95368193089962e-06) - actor/kl_loss:np.float64(0.0) - actor/pg_clipfrac:np.float64(2.2877348328620428e-05) - actor/ppo_kl:np.float64(-5.772552391647423e-05) - actor/pg_clipfrac_lower:np.float64(0.0) - actor/grad_norm:np.float64(0.1109579335898161) - perf/mfu/actor:np.float64(0.09790439064664765) - perf/max_memory_allocated_gb:np.float64(128.6631588935852) - perf/max_memory_reserved_gb:np.float64(136.9375) - perf/cpu_memory_used_gb:np.float64(106.49691772460938) - actor/lr:np.float64(1e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7799202127659575) - val-aux/sciknoweval/reward/std@16:np.float64(0.052101428059437495) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.799936170212766) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.03891400102821863) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.7602021276595745) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.0484960635134995) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7798829787234042) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.052006938967530975) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.8139574468085107) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.027455777877525876) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.7414893617021275) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.04241546758668117) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7823297872340426) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.038877017813791076) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.8243297872340425) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.019003297583575607) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.7248191489361702) - val-aux/sciknoweval/reward/worst@8/std:np.float64(0.035434813866661946) - val-aux/sciknoweval/reward/maj@8/mean:np.float64(0.7831382978723406) - val-aux/sciknoweval/reward/maj@8/std:np.float64(0.02545128949215922) - val-aux/sciknoweval/reward/best@16/mean:np.float64(0.8322765957446808) - val-aux/sciknoweval/reward/best@
|
||
("Final validation metrics: {'val-aux/sciknoweval/reward/mean@16': "
|
||
"np.float64(0.7799202127659575), 'val-aux/sciknoweval/reward/std@16': "
|
||
"np.float64(0.052101428059437495), 'val-aux/sciknoweval/reward/best@2/mean': "
|
||
"np.float64(0.799936170212766), 'val-aux/sciknoweval/reward/best@2/std': "
|
||
"np.float64(0.03891400102821863), 'val-aux/sciknoweval/reward/worst@2/mean': "
|
||
"np.float64(0.7602021276595745), 'val-aux/sciknoweval/reward/worst@2/std': "
|
||
"np.float64(0.0484960635134995), 'val-aux/sciknoweval/reward/maj@2/mean': "
|
||
"np.float64(0.7798829787234042), 'val-aux/sciknoweval/reward/maj@2/std': "
|
||
"np.float64(0.052006938967530975), 'val-aux/sciknoweval/reward/best@4/mean': "
|
||
"np.float64(0.8139574468085107), 'val-aux/sciknoweval/reward/best@4/std': "
|
||
"np.float64(0.027455777877525876), 'val-aux/sciknoweval/reward/worst@4/mean': "
|
||
"np.float64(0.7414893617021275), 'val-aux/sciknoweval/reward/worst@4/std': "
|
||
"np.float64(0.04241546758668117), 'val-aux/sciknoweval/reward/maj@4/mean': "
|
||
"np.float64(0.7823297872340426), 'val-aux/sciknoweval/reward/maj@4/std': "
|
||
"np.float64(0.038877017813791076), 'val-aux/sciknoweval/reward/best@8/mean': "
|
||
"np.float64(0.8243297872340425), 'val-aux/sciknoweval/reward/best@8/std': "
|
||
"np.float64(0.019003297583575607), 'val-aux/sciknoweval/reward/worst@8/mean': "
|
||
"np.float64(0.7248191489361702), 'val-aux/sciknoweval/reward/worst@8/std': "
|
||
"np.float64(0.035434813866661946), 'val-aux/sciknoweval/reward/maj@8/mean': "
|
||
"np.float64(0.7831382978723406), 'val-aux/sciknoweval/reward/maj@8/std': "
|
||
"np.float64(0.02545128949215922), 'val-aux/sciknoweval/reward/best@16/mean': "
|
||
"np.float64(0.8322765957446808), 'val-aux/sciknoweval/reward/best@16/std': "
|
||
'np.float64(0.013309078916541482), '
|
||
"'val-aux/sciknoweval/reward/worst@16/mean': np.float64(0.7087659574468085), "
|
||
"'val-aux/sciknoweval/reward/worst@16/std': np.float64(0.027047818566610645), "
|
||
"'val-aux/sciknoweval/reward/maj@16/mean': np.float64(0.7843510638297873), "
|
||
"'val-aux/sciknoweval/reward/maj@16/std': np.float64(0.01931287084263904), "
|
||
"'val-aux/sciknoweval/score/mean@16': np.float64(0.7799202127659575), "
|
||
"'val-aux/sciknoweval/score/std@16': np.float64(0.052101428059437495), "
|
||
"'val-aux/sciknoweval/score/best@2/mean': np.float64(0.799936170212766), "
|
||
"'val-aux/sciknoweval/score/best@2/std': np.float64(0.03891400102821863), "
|
||
"'val-aux/sciknoweval/score/worst@2/mean': np.float64(0.7602021276595745), "
|
||
"'val-aux/sciknoweval/score/worst@2/std': np.float64(0.0484960635134995), "
|
||
"'val-aux/sciknoweval/score/maj@2/mean': np.float64(0.7798829787234042), "
|
||
"'val-aux/sciknoweval/score/maj@2/std': np.float64(0.052006938967530975), "
|
||
"'val-aux/sciknoweval/score/best@4/mean': np.float64(0.8139574468085107), "
|
||
"'val-aux/sciknoweval/score/best@4/std': np.float64(0.027455777877525876), "
|
||
"'val-aux/sciknoweval/score/worst@4/mean': np.float64(0.7414893617021275), "
|
||
"'val-aux/sciknoweval/score/worst@4/std': np.float64(0.04241546758668117), "
|
||
"'val-aux/sciknoweval/score/maj@4/mean': np.float64(0.7823297872340426), "
|
||
"'val-aux/sciknoweval/score/maj@4/std': np.float64(0.038877017813791076), "
|
||
"'val-aux/sciknoweval/score/best@8/mean': np.float64(0.8243297872340425), "
|
||
"'val-aux/sciknoweval/score/best@8/std': np.float64(0.019003297583575607), "
|
||
"'val-aux/sciknoweval/score/worst@8/mean': np.float64(0.7248191489361702), "
|
||
"'val-aux/sciknoweval/score/worst@8/std': np.float64(0.035434813866661946), "
|
||
"'val-aux/sciknoweval/score/maj@8/mean': np.float64(0.7831382978723406), "
|
||
"'val-aux/sciknoweval/score/maj@8/std': np.float64(0.02545128949215922), "
|
||
"'val-aux/sciknoweval/score/best@16/mean': np.float64(0.8322765957446808), "
|
||
"'val-aux/sciknoweval/score/best@16/std': np.float64(0.013309078916541482), "
|
||
"'val-aux/sciknoweval/score/worst@16/mean': np.float64(0.7087659574468085), "
|
||
"'val-aux/sciknoweval/score/worst@16/std': np.float64(0.027047818566610645), "
|
||
"'val-aux/sciknoweval/score/maj@16/mean': np.float64(0.7843510638297873), "
|
||
"'val-aux/sciknoweval/score/maj@16/std': np.float64(0.01931287084263904), "
|
||
"'val-core/sciknoweval/acc/mean@16': np.float64(0.7799202127659575), "
|
||
"'val-aux/sciknoweval/acc/std@16': np.float64(0.052101428059437495), "
|
||
"'val-aux/sciknoweval/acc/best@2/mean': np.float64(0.799936170212766), "
|
||
"'val-aux/sciknoweval/acc/best@2/std': np.float64(0.03891400102821863), "
|
||
"'val-aux/sciknoweval/acc/worst@2/mean': np.float64(0.7602021276595745), "
|
||
"'val-aux/sciknoweval/acc/worst@2/std': np.float64(0.0484960635134995), "
|
||
"'val-aux/sciknoweval/acc/maj@2/mean': np.float64(0.7798829787234042), "
|
||
"'val-aux/sciknoweval/acc/maj@2/std': np.float64(0.052006938967530975), "
|
||
"'val-aux/sciknoweval/acc/best@4/mean': np.float64(0.8139574468085107), "
|
||
"'val-aux/sciknoweval/acc/best@4/std': np.float64(0.027455777877525876), "
|
||
"'val-aux/sciknoweval/acc/worst@4/mean': np.float64(0.7414893617021275), "
|
||
"'val-aux/sciknoweval/acc/worst@4/std': np.float64(0.04241546758668117), "
|
||
"'val-aux/sciknoweval/acc/maj@4/mean': np.float64(0.7823297872340426), "
|
||
"'val-aux/sciknoweval/acc/maj@4/std': np.float64(0.038877017813791076), "
|
||
"'val-aux/sciknoweval/acc/best@8/mean': np.float64(0.8243297872340425), "
|
||
"'val-aux/sciknoweval/acc/best@8/std': np.float64(0.019003297583575607), "
|
||
"'val-aux/sciknoweval/acc/worst@8/mean': np.float64(0.7248191489361702), "
|
||
"'val-aux/sciknoweval/acc/worst@8/std': np.float64(0.035434813866661946), "
|
||
"'val-aux/sciknoweval/acc/maj@8/mean': np.float64(0.7831382978723406), "
|
||
"'val-aux/sciknoweval/acc/maj@8/std': np.float64(0.02545128949215922), "
|
||
"'val-core/sciknoweval/acc/best@16/mean': np.float64(0.8322765957446808), "
|
||
"'val-core/sciknoweval/acc/best@16/std': np.float64(0.013309078916541482), "
|
||
"'val-aux/sciknoweval/acc/worst@16/mean': np.float64(0.7087659574468085), "
|
||
"'val-aux/sciknoweval/acc/worst@16/std': np.float64(0.027047818566610645), "
|
||
"'val-core/sciknoweval/acc/maj@16/mean': np.float64(0.7843510638297873), "
|
||
"'val-core/sciknoweval/acc/maj@16/std': np.float64(0.01931287084263904), "
|
||
"'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)}")
|