{"val-aux/num_turns/mean":2,"val-aux/sciknoweval/reward/maj@16/mean":0.7719148936170213,"val-aux/sciknoweval/score/maj@16/std":0.018618881588821194,"response_length_non_aborted/max":1110,"val-core/sciknoweval/acc/best@16/std":0.00848440146040565,"actor/entropy":0.05979691818356514,"rollout_corr/chi2_token":0.0007009264081716537,"val-aux/sciknoweval/incorrect_format/worst@4/mean":0.9895957446808511,"timing_s/agent_loop/tool_calls/max":0,"val-aux/sciknoweval/incorrect_format/maj@8/std":0.0012050425965809124,"perf/max_memory_reserved_gb":129.208984375,"val-aux/sciknoweval/incorrect_format/maj@16/mean":1,"critic/advantages/min":-0.875,"val-aux/sciknoweval/incorrect_format/worst@2/std":0.011671809930422904,"val-core/sciknoweval/acc/mean@16":0.7626329787234043,"val-aux/sciknoweval/reward/std@16":0.06729335213589911,"global_seqlen/minmax_diff":12650,"rollout_corr/rollout_is_eff_sample_size":0.9990782896419121,"val-aux/sciknoweval/reward/best@8/std":0.014096974337829643,"val-aux/sciknoweval/reward/maj@4/std":0.04686638157004486,"val-aux/sciknoweval/acc/maj@2/std":0.06820989645423031,"val-aux/sciknoweval/incorrect_format/best@8/mean":1,"val-aux/sciknoweval/reward/best@16/std":0.00848440146040565,"val-aux/sciknoweval/reward/best@2/mean":0.7876382978723404,"val-aux/sciknoweval/incorrect_format/maj@2/mean":0.996872340425532,"val-aux/sciknoweval/reward/best@2/std":0.041836077372629385,"prompt_length/max":479,"timing_s/agent_loop/slowest/generate_sequences":8.524729896336794,"val-aux/sciknoweval/incorrect_format/maj@4/mean":0.9992021276595746,"val-aux/sciknoweval/score/best@2/std":0.041836077372629385,"val-aux/sciknoweval/score/worst@2/std":0.07115878526747459,"val-aux/sciknoweval/reward/maj@8/std":0.026416313160677763,"timing_s/agent_loop/slowest/prompt_length":276,"critic/returns/min":-0.875,"val-aux/sciknoweval/incorrect_format/best@16/std":0,"val-aux/sciknoweval/score/worst@2/mean":0.7377765957446808,"critic/score/max":1,"critic/score/min":0,"val-aux/sciknoweval/reward/maj@8/mean":0.7723617021276596,"rollout_corr/rollout_is_ratio_fraction_low":0.0001169390743598342,"val-aux/sciknoweval/reward/maj@16/std":0.018618881588821194,"prompt_length/min":190,"val-aux/sciknoweval/acc/best@8/mean":0.8101914893617022,"val-aux/sciknoweval/score/best@4/mean":0.801627659574468,"val-aux/sciknoweval/incorrect_format/worst@2/mean":0.9942765957446809,"global_seqlen/max":33657,"val-aux/sciknoweval/incorrect_format/best@2/mean":0.9995851063829787,"val-aux/sciknoweval/acc/worst@2/std":0.07115878526747459,"critic/rewards/min":0,"val-aux/sciknoweval/score/best@16/mean":0.8149574468085106,"actor/pg_clipfrac_lower":0,"val-aux/sciknoweval/score/maj@2/mean":0.7628297872340426,"timing_s/start_profile":9.438768029212952e-05,"response_length/mean":434.25390625,"rollout_corr/rollout_is_ratio_fraction_high":4.497656846069731e-05,"_wandb":{"runtime":3382},"val-aux/sciknoweval/score/maj@4/mean":0.770308510638298,"val-aux/num_turns/max":2,"timing_s/agent_loop/generate_sequences/mean":4.727881735328992,"val-aux/sciknoweval/score/maj@16/mean":0.7719148936170213,"val-aux/sciknoweval/score/best@8/std":0.014096974337829643,"perf/cpu_memory_used_gb":103.06706619262695,"timing_s/old_log_prob":2.3261021357029676,"timing_s/gen":14.453305700793862,"rollout_corr/k3_kl":0.000666843313986476,"val-aux/sciknoweval/reward/maj@4/mean":0.770308510638298,"val-aux/sciknoweval/reward/best@16/mean":0.8149574468085106,"timing_s/step":26.7111904527992,"rollout_corr/kl":0.0009433553804747419,"perf/time_per_step":26.7111904527992,"val-aux/sciknoweval/reward/worst@16/mean":0.6536170212765957,"val-aux/sciknoweval/incorrect_format/std@16":0.009302503118891777,"val-core/sciknoweval/acc/best@16/mean":0.8149574468085106,"val-aux/sciknoweval/reward/best@4/mean":0.801627659574468,"timing_s/stop_profile":0.00014218315482139587,"val-aux/sciknoweval/score/maj@8/mean":0.7723617021276596,"val-aux/sciknoweval/incorrect_format/worst@8/mean":0.9830851063829789,"timing_s/testing":94.38914828561246,"prompt_length/mean":391.125,"timing_s/agent_loop/slowest/tool_calls":0,"response_length_non_aborted/min":197,"val-aux/sciknoweval/score/maj@2/std":0.06820989645423031,"val-aux/sciknoweval/acc/maj@8/mean":0.7723617021276596,"val-aux/sciknoweval/score/mean@16":0.7626329787234043,"val-aux/sciknoweval/reward/worst@8/mean":0.6826808510638298,"val-aux/sciknoweval/acc/std@16":0.06729335213589911,"val-aux/sciknoweval/acc/maj@4/std":0.04686638157004486,"val-core/sciknoweval/acc/maj@16/std":0.018618881588821194,"val-aux/sciknoweval/acc/worst@4/std":0.07037925820603312,"timing_s/adv":0.12102740071713924,"rollout_corr/rollout_is_seq_max_deviation":0.005619168281555176,"val-aux/sciknoweval/incorrect_format/maj@2/std":0.009074902857638772,"rollout_corr/log_ppl_diff_min":-0.004364762455224991,"timing_per_token_ms/adv":0.0005727833368061982,"rollout_corr/log_ppl_diff":0.0009433558898308547,"global_seqlen/min":21007,"val-aux/sciknoweval/incorrect_format/maj@8/mean":0.9998617021276596,"rollout_corr/rollout_is_seq_std":0.0016542637022212148,"timing_s/agent_loop/generate_sequences/max":8.524729896336794,"rollout_corr/log_ppl_abs_diff":0.0016063227940321667,"rollout_corr/rollout_ppl":1.0636031506583095,"critic/returns/max":0.375,"val-aux/sciknoweval/reward/mean@16":0.7626329787234043,"val-aux/sciknoweval/acc/worst@16/std":0.05400056483953881,"rollout_corr/training_ppl":1.0646160240285099,"perf/total_num_tokens":211297,"val-aux/sciknoweval/score/worst@4/mean":0.7114680851063832,"val-aux/sciknoweval/acc/worst@8/mean":0.6826808510638298,"rollout_corr/rollout_is_seq_mean":0.9998504519462585,"val-aux/sciknoweval/reward/worst@16/std":0.05400056483953881,"response_length_non_aborted/clip_ratio":0,"val-aux/num_turns/min":2,"training/rollout_actor_probs_pearson_corr":0.9959689378738403,"num_turns/max":2,"response_length_non_aborted/mean":434.25390625,"critic/rewards/max":1,"val-aux/sciknoweval/incorrect_format/best@4/mean":0.9999787234042554,"timing_s/agent_loop/tool_calls/mean":0,"global_seqlen/balanced_min":26409,"val-aux/sciknoweval/incorrect_format/worst@16/mean":0.976063829787234,"val-aux/sciknoweval/acc/maj@4/mean":0.770308510638298,"training/rollout_probs_diff_mean":0.0023399998899549246,"num_turns/mean":2,"actor/kl_loss":0,"actor/ppo_kl":0.00029493772322253253,"val-aux/sciknoweval/score/worst@8/mean":0.6826808510638298,"val-aux/sciknoweval/acc/best@2/mean":0.7876382978723404,"val-aux/sciknoweval/score/worst@16/mean":0.6536170212765957,"val-aux/sciknoweval/incorrect_format/worst@8/std":0.01468444636139687,"val-aux/sciknoweval/reward/maj@2/std":0.06820989645423031,"val-core/sciknoweval/acc/maj@16/mean":0.7719148936170213,"critic/score/mean":0.671875,"val-aux/sciknoweval/reward/best@8/mean":0.8101914893617022,"timing_s/reward":0.06965606473386288,"val-aux/sciknoweval/incorrect_format/mean@16":0.9966755319148937,"response_length/max":1110,"val-aux/sciknoweval/acc/best@4/std":0.023418987057881077,"global_seqlen/balanced_max":26415,"critic/advantages/mean":0.000756730732973665,"val-aux/sciknoweval/acc/best@2/std":0.041836077372629385,"timing_s/agent_loop/slowest/response_length":1110,"val-aux/sciknoweval/incorrect_format/best@16/mean":1,"val-aux/sciknoweval/score/maj@8/std":0.026416313160677763,"global_seqlen/mean":26412.125,"training/rollout_probs_diff_max":0.37930381298065186,"perf/mfu/actor":0.07842677559658048,"perf/mfu/actor_infer":0,"timing_per_token_ms/update_actor":0.04569008660838502,"rollout_corr/rollout_is_seq_max":1.0056191682815552,"training/epoch":3,"timing_s/agent_loop/tool_calls/min":0,"_runtime":3382.692562051,"timing_s/agent_loop/generate_sequences/min":2.1825265772640705,"val-aux/sciknoweval/acc/worst@16/mean":0.6536170212765957,"rollout_corr/ppl_ratio":1.0009459347929806,"timing_s/save_checkpoint":6.867023587226868,"val-aux/sciknoweval/score/best@16/std":0.00848440146040565,"val-aux/sciknoweval/score/worst@4/std":0.07037925820603312,"perf/max_memory_allocated_gb":123.45145273208618,"response_length/min":197,"val-aux/sciknoweval/score/best@4/std":0.023418987057881077,"val-aux/sciknoweval/score/maj@4/std":0.04686638157004486,"rollout_corr/rollout_is_mean":0.999850869178772,"val-aux/sciknoweval/acc/maj@2/mean":0.7628297872340426,"val-aux/sciknoweval/score/best@8/mean":0.8101914893617022,"_timestamp":1.7830003110546255e+09,"val-aux/sciknoweval/acc/maj@8/std":0.026416313160677763,"rollout_corr/rollout_is_min":0.28976842761039734,"rollout_corr/rollout_is_max":2.267052173614502,"response_length/clip_ratio":0,"rollout_corr/training_log_ppl":0.06111104129377054,"timing_per_token_ms/gen":0.1300120150473051,"critic/returns/mean":0.000756730732973665,"_step":100,"val-aux/sciknoweval/acc/worst@4/mean":0.7114680851063832,"rollout_corr/rollout_is_seq_fraction_high":0,"training/rollout_probs_diff_std":0.011555762961506844,"val-aux/sciknoweval/acc/best@8/std":0.014096974337829643,"num_turns/min":2,"rollout_corr/rollout_is_seq_min":0.9947155714035034,"actor/pg_clipfrac":0.00022653777296000044,"perf/throughput":988.8037392669685,"val-aux/sciknoweval/reward/worst@4/mean":0.7114680851063832,"rollout_corr/rollout_is_std":0.03037368506193161,"val-aux/sciknoweval/acc/best@4/mean":0.801627659574468,"val-aux/sciknoweval/acc/worst@2/mean":0.7377765957446808,"val-aux/sciknoweval/reward/worst@2/mean":0.7377765957446808,"val-aux/sciknoweval/score/worst@16/std":0.05400056483953881,"training/global_step":100,"critic/advantages/max":0.375,"val-aux/sciknoweval/incorrect_format/worst@4/std":0.014164656923311533,"val-aux/sciknoweval/reward/worst@8/std":0.06494348791272907,"val-aux/sciknoweval/reward/maj@2/mean":0.7628297872340426,"val-aux/sciknoweval/incorrect_format/maj@4/std":0.004446232243099193,"val-aux/sciknoweval/reward/worst@4/std":0.07037925820603312,"val-aux/sciknoweval/reward/worst@2/std":0.07115878526747459,"actor/grad_norm":0.14011954190209508,"val-aux/sciknoweval/acc/worst@8/std":0.06494348791272907,"training/rollout_probs_diff_valid":1,"val-aux/sciknoweval/incorrect_format/best@8/std":0,"val-aux/sciknoweval/reward/best@4/std":0.023418987057881077,"val-aux/sciknoweval/incorrect_format/best@2/std":0.00298506157652289,"rollout_corr/chi2_seq":0.7943548788316548,"timing_s/update_actor":9.65417823009193,"critic/rewards/mean":0.671875,"val-aux/sciknoweval/score/std@16":0.06729335213589911,"val-aux/sciknoweval/incorrect_format/worst@16/std":0.01202692239227068,"response/aborted_ratio":0,"rollout_corr/log_ppl_diff_max":0.009745059534907341,"val-aux/sciknoweval/score/worst@8/std":0.06494348791272907,"actor/pg_loss":4.458578769117594e-05,"val-aux/sciknoweval/score/best@2/mean":0.7876382978723404,"rollout_corr/rollout_is_seq_fraction_low":0,"prompt_length/clip_ratio":0,"val-aux/sciknoweval/incorrect_format/maj@16/std":0,"val-aux/sciknoweval/incorrect_format/best@4/std":0.00047528314688699314,"rollout_corr/rollout_log_ppl":0.060167685403939686,"actor/lr":1e-06}