{"val-aux/sciknoweval/incorrect_format/worst@8/mean":0.933025,"val-aux/sciknoweval/reward/best@16/std":0.061136881006381635,"val-aux/sciknoweval/reward/best@8/mean":0.8526375000000002,"val-aux/sciknoweval/incorrect_format/best@8/mean":1,"rollout_corr/rollout_is_max":2.3022942543029785,"rollout_corr/training_log_ppl":0.26966726542741526,"val-aux/sciknoweval/reward/worst@2/std":0.15413287282151064,"val-core/sciknoweval/acc/best@16/std":0.061136881006381635,"rollout_corr/log_ppl_diff_min":-0.0106421560049057,"timing_per_token_ms/gen":0.15551790835380125,"val-aux/sciknoweval/acc/worst@8/std":0.08946281937139845,"val-aux/sciknoweval/incorrect_format/maj@16/std":0.0017543234207183663,"training/rollout_probs_diff_valid":1,"perf/time_per_step":41.70409296452999,"val-aux/sciknoweval/score/mean@16":0.6828125,"rollout_corr/rollout_is_seq_max_deviation":0.007250070571899414,"rollout_corr/kl":0.00037315004849824973,"val-aux/sciknoweval/incorrect_format/best@4/mean":0.9998374999999999,"val-aux/sciknoweval/acc/worst@16/std":0.054003469751280585,"perf/max_memory_allocated_gb":123.32309532165527,"actor/pg_clipfrac":0.0001629320678375734,"val-aux/sciknoweval/incorrect_format/worst@4/std":0.04764508939236297,"val-aux/sciknoweval/incorrect_format/worst@4/mean":0.9560875,"val-core/sciknoweval/acc/best@16/mean":0.8910875,"val-aux/sciknoweval/score/best@2/std":0.1476854115376502,"critic/score/max":1,"timing_s/start_profile":8.623674511909485e-05,"critic/advantages/min":-0.875,"val-aux/sciknoweval/score/worst@4/mean":0.5515125000000001,"response/aborted_ratio":0,"val-aux/sciknoweval/score/best@2/mean":0.7532,"val-aux/sciknoweval/reward/maj@8/mean":0.710175,"timing_s/agent_loop/generate_sequences/max":22.433935675770044,"val-aux/sciknoweval/incorrect_format/worst@16/std":0.029568343249240757,"val-aux/sciknoweval/reward/maj@4/mean":0.7018125,"rollout_corr/rollout_is_seq_std":0.0016033316496759653,"val-aux/sciknoweval/incorrect_format/best@8/std":0,"val-aux/sciknoweval/incorrect_format/maj@8/mean":0.9988624999999999,"val-aux/sciknoweval/acc/maj@2/mean":0.6832499999999999,"training/rollout_actor_probs_pearson_corr":0.999374508857727,"val-aux/sciknoweval/incorrect_format/worst@8/std":0.0436722700672832,"val-aux/num_turns/mean":2,"val-aux/sciknoweval/score/worst@16/std":0.054003469751280585,"global_seqlen/balanced_min":31299,"val-aux/sciknoweval/reward/best@4/mean":0.807675,"val-aux/sciknoweval/score/maj@2/std":0.17928347494256552,"val-aux/sciknoweval/score/maj@4/mean":0.7018125,"perf/cpu_memory_used_gb":107.57986450195312,"val-aux/sciknoweval/reward/worst@8/mean":0.4995375,"val-aux/num_turns/min":2,"val-aux/sciknoweval/incorrect_format/maj@16/mean":0.9998750000000001,"val-aux/sciknoweval/reward/maj@16/std":0.05584489583302406,"response_length/mean":712.5625,"val-aux/sciknoweval/acc/best@4/std":0.11309093981217563,"val-aux/sciknoweval/score/best@8/mean":0.8526375000000002,"global_seqlen/min":17010,"val-aux/sciknoweval/reward/best@4/std":0.11309093981217563,"val-aux/sciknoweval/reward/worst@2/mean":0.6133875,"rollout_corr/rollout_is_seq_mean":1.0000271797180176,"val-core/sciknoweval/acc/mean@16":0.6828125,"actor/lr":1e-06,"timing_s/agent_loop/tool_calls/max":0,"val-aux/sciknoweval/reward/maj@2/mean":0.6832499999999999,"response_length_non_aborted/max":3652,"training/rollout_probs_diff_std":0.00978316180408001,"training/rollout_probs_diff_max":0.630595862865448,"prompt_length/clip_ratio":0,"rollout_corr/rollout_is_std":0.031683772802352905,"critic/score/mean":0.7265625,"val-aux/sciknoweval/incorrect_format/std@16":0.03383512807052093,"timing_per_token_ms/update_actor":0.041682880410699334,"timing_s/agent_loop/slowest/response_length":3652,"prompt_length/min":191,"rollout_corr/rollout_ppl":1.3452776027843356,"val-aux/sciknoweval/score/maj@16/mean":0.7157250000000001,"val-aux/sciknoweval/acc/worst@4/mean":0.5515125000000001,"val-core/sciknoweval/acc/maj@16/std":0.05584489583302406,"_runtime":4928.163775548,"val-aux/sciknoweval/score/worst@4/std":0.12241577016075818,"val-aux/sciknoweval/reward/std@16":0.17979486326514935,"actor/pg_loss":-7.347844075411558e-05,"val-aux/sciknoweval/score/worst@2/std":0.15413287282151064,"perf/throughput":753.3792912537954,"val-aux/sciknoweval/acc/best@8/mean":0.8526375000000002,"val-aux/sciknoweval/incorrect_format/maj@4/std":0.021483511676811038,"val-aux/sciknoweval/incorrect_format/worst@2/std":0.042452979144002284,"critic/advantages/mean":-0.0018172748386859894,"rollout_corr/rollout_is_seq_max":1.0072500705718994,"val-aux/sciknoweval/score/maj@4/std":0.13581544160753714,"val-aux/sciknoweval/score/maj@8/mean":0.710175,"timing_per_token_ms/adv":0.00048023693467704956,"val-aux/sciknoweval/score/best@16/mean":0.8910875,"timing_s/testing":119.00530271604657,"val-aux/sciknoweval/score/worst@8/mean":0.4995375,"actor/grad_norm":0.12031856318935752,"val-aux/sciknoweval/acc/worst@2/mean":0.6133875,"val-aux/sciknoweval/score/worst@2/mean":0.6133875,"actor/ppo_kl":-0.00012778303023264215,"val-aux/sciknoweval/incorrect_format/maj@2/std":0.03387729307844336,"timing_s/step":41.70409296452999,"response_length_non_aborted/mean":712.5625,"val-aux/sciknoweval/acc/worst@16/mean":0.46395,"val-aux/sciknoweval/acc/maj@8/std":0.08949529688241312,"response_length_non_aborted/min":110,"perf/mfu/actor_infer":0,"rollout_corr/rollout_is_min":0.3131349980831146,"val-aux/sciknoweval/reward/worst@16/std":0.054003469751280585,"timing_s/adv":0.12070851400494576,"val-aux/sciknoweval/acc/maj@4/std":0.13581544160753714,"val-aux/sciknoweval/reward/worst@16/mean":0.46395,"val-aux/sciknoweval/reward/maj@8/std":0.08949529688241312,"val-aux/sciknoweval/incorrect_format/best@4/std":0.002749370152057837,"global_seqlen/minmax_diff":26423,"val-aux/sciknoweval/score/best@8/std":0.0888708326383256,"timing_s/stop_profile":0.00019830837845802307,"rollout_corr/rollout_is_ratio_fraction_high":1.0963950444420334e-05,"val-aux/sciknoweval/incorrect_format/best@16/std":0,"val-aux/sciknoweval/reward/maj@2/std":0.17928347494256552,"val-aux/sciknoweval/reward/maj@4/std":0.13581544160753714,"global_seqlen/mean":31419,"training/global_step":100,"val-aux/sciknoweval/score/maj@2/mean":0.6832499999999999,"val-aux/sciknoweval/incorrect_format/best@16/mean":1,"rollout_corr/ppl_ratio":1.0003747479058802,"response_length/clip_ratio":0,"rollout_corr/rollout_is_seq_min":0.9959959983825684,"critic/advantages/max":0.875,"timing_s/save_checkpoint":6.902973407879472,"val-aux/sciknoweval/acc/maj@4/mean":0.7018125,"val-aux/sciknoweval/score/std@16":0.17979486326514935,"rollout_corr/rollout_is_seq_fraction_low":0,"val-aux/sciknoweval/incorrect_format/mean@16":0.9859375,"val-aux/sciknoweval/reward/worst@8/std":0.08946281937139845,"rollout_corr/rollout_log_ppl":0.2692941162531497,"val-aux/sciknoweval/incorrect_format/maj@4/mean":0.9946999999999999,"rollout_corr/log_ppl_diff_max":0.0049625784158706665,"critic/score/min":0,"training/rollout_probs_diff_mean":0.0036211644764989614,"timing_s/agent_loop/generate_sequences/min":1.2800081353634596,"rollout_corr/rollout_is_mean":1.0000176429748535,"val-aux/sciknoweval/reward/best@8/std":0.0888708326383256,"actor/entropy":0.37953582406044006,"val-aux/sciknoweval/score/best@4/std":0.11309093981217563,"response_length/min":110,"critic/returns/mean":-0.0018172748386859894,"val-aux/sciknoweval/acc/best@2/std":0.1476854115376502,"timing_s/agent_loop/tool_calls/mean":0,"timing_s/gen":28.36895477026701,"val-aux/sciknoweval/incorrect_format/best@2/mean":0.99745,"perf/total_num_tokens":251352,"val-aux/sciknoweval/score/best@4/mean":0.807675,"val-aux/sciknoweval/acc/best@2/mean":0.7532,"val-aux/sciknoweval/score/worst@8/std":0.08946281937139845,"val-aux/sciknoweval/acc/best@8/std":0.0888708326383256,"val-aux/sciknoweval/incorrect_format/best@2/std":0.014287439995541685,"actor/pg_clipfrac_lower":0,"val-aux/sciknoweval/acc/maj@8/mean":0.710175,"val-aux/sciknoweval/incorrect_format/worst@2/mean":0.9741250000000001,"rollout_corr/training_ppl":1.3458213401027024,"critic/rewards/mean":0.7265625,"rollout_corr/rollout_is_ratio_fraction_low":1.644592521188315e-05,"val-aux/sciknoweval/acc/worst@2/std":0.15413287282151064,"val-aux/sciknoweval/score/best@16/std":0.061136881006381635,"perf/mfu/actor":0.0881824181802986,"val-aux/sciknoweval/reward/worst@4/mean":0.5515125000000001,"response_length_non_aborted/clip_ratio":0,"val-aux/sciknoweval/reward/mean@16":0.6828125,"_timestamp":1.7829825726200204e+09,"rollout_corr/rollout_is_eff_sample_size":0.998997264270168,"val-aux/sciknoweval/score/maj@16/std":0.05584489583302406,"timing_s/update_actor":10.477075356990099,"training/epoch":4,"global_seqlen/max":43433,"val-aux/sciknoweval/incorrect_format/maj@2/mean":0.9857875,"actor/kl_loss":0,"critic/rewards/min":0,"rollout_corr/log_ppl_diff":0.0003731491742655635,"val-aux/sciknoweval/reward/best@2/mean":0.7532,"rollout_corr/log_ppl_abs_diff":0.0012809229083359241,"critic/returns/min":-0.875,"val-aux/sciknoweval/reward/best@2/std":0.1476854115376502,"val-aux/sciknoweval/incorrect_format/worst@16/mean":0.9139250000000001,"critic/returns/max":0.875,"val-aux/sciknoweval/score/maj@8/std":0.08949529688241312,"timing_s/reward":0.07503514550626278,"val-aux/sciknoweval/acc/std@16":0.17979486326514935,"_wandb":{"runtime":4928},"val-aux/sciknoweval/reward/worst@4/std":0.12241577016075818,"prompt_length/max":383,"val-aux/num_turns/max":2,"val-aux/sciknoweval/incorrect_format/maj@8/std":0.009237722502874987,"num_turns/mean":2,"_step":100,"val-aux/sciknoweval/acc/worst@8/mean":0.4995375,"timing_s/old_log_prob":2.573247427120805,"rollout_corr/chi2_seq":1.9170411999803036,"val-aux/sciknoweval/reward/maj@16/mean":0.7157250000000001,"critic/rewards/max":1,"val-aux/sciknoweval/reward/best@16/mean":0.8910875,"val-core/sciknoweval/acc/maj@16/mean":0.7157250000000001,"timing_s/agent_loop/slowest/tool_calls":0,"timing_s/agent_loop/tool_calls/min":0,"global_seqlen/balanced_max":31491,"perf/max_memory_reserved_gb":129.08984375,"num_turns/max":2,"timing_s/agent_loop/slowest/generate_sequences":22.433935675770044,"val-aux/sciknoweval/acc/worst@4/std":0.12241577016075818,"rollout_corr/chi2_token":0.0014326744712889194,"val-aux/sciknoweval/score/worst@16/mean":0.46395,"response_length/max":3652,"num_turns/min":2,"val-aux/sciknoweval/acc/maj@2/std":0.17928347494256552,"rollout_corr/rollout_is_seq_fraction_high":0,"val-aux/sciknoweval/acc/best@4/mean":0.807675,"prompt_length/mean":269.28125,"rollout_corr/k3_kl":0.0005444729254122649,"timing_s/agent_loop/slowest/prompt_length":342,"timing_s/agent_loop/generate_sequences/mean":6.539074072250514}