670 lines
426 KiB
Plaintext
670 lines
426 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: /workspace/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-SDPO_TR-Qwen-Qwen3-4B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/latest_checkpointed_iteration.txt
|
||
|
|
Training from scratch
|
||
|
|
Training Progress: 9%|▉ | 9/100 [04:57<44:24, 29.28s/it]
|
||
|
|
step:1 - global_seqlen/min:15718 - global_seqlen/max:19322 - global_seqlen/minmax_diff:3604 - global_seqlen/balanced_min:18449 - global_seqlen/balanced_max:18455 - global_seqlen/mean:18453.25 - actor/entropy:0.22047318518161774 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6983712911605835 - training/rollout_probs_diff_mean:0.006159528158605099 - training/rollout_probs_diff_std:0.01634596474468708 - training/rollout_actor_probs_pearson_corr:0.997281551361084 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.859375 - self_distillation/correct_sample_fraction:0.59765625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.859375 - rollout_corr/rollout_is_mean:1.0004948377609253 - rollout_corr/rollout_is_ratio_fraction_high:8.58074490679428e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0001716148981358856 - rollout_corr/rollout_is_max:7.675390720367432 - rollout_corr/rollout_is_min:0.17976222932338715 - rollout_corr/rollout_is_std:0.048191651701927185 - rollout_corr/rollout_is_eff_sample_size:0.9976848443531777 - rollout_corr/rollout_is_seq_mean:1.0004761219024658 - rollout_corr/rollout_is_seq_std:0.0032042181119322777 - rollout_corr/rollout_is_seq_max:1.0277915000915527 - rollout_corr/rollout_is_seq_min:0.9921735525131226 - rollout_corr/rollout_is_seq_max_deviation:0.027791500091552734 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2663253066129982) - rollout_corr/training_log_ppl:np.float64(0.2329175515624229) - rollout_corr/kl:np.float64(0.0008614983111225882) - rollout_corr/k3_kl:np.float64(0.001337601991849624) - rollout_corr/rollout_ppl:np.float64(1.2652166304178536) - rollout_corr/rollout_log_ppl:np.float64(0.23205605207476765) - rollout_corr/log_ppl_diff:np.float64(0.0008614994876552373) - rollout_corr/log_ppl_abs_diff:np.float64(0.0022081050265114754) - rollout_corr/log_ppl_diff_max:np.float64(0.013160884380340576) - rollout_corr/log_ppl_diff_min:np.float64(-0.005962729454040527) - rollout_corr/ppl_ratio:np.float64(1.000866234768182) - rollout_corr/chi2_token:np.float64(0.004201937932521105) - rollout_corr/chi2_seq:np.float64(1.8266779333353043) - actor/pg_loss:np.float64(0.04125104084960185) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.140625) - actor/grad_norm:np.float64(0.797112762928009) - perf/mfu/actor:np.float64(0.026825697315170465) - perf/max_memory_allocated_gb:np.float64(116.06170654296875) - perf/max_memory_reserved_gb:np.float64(121.2890625) - perf/cpu_memory_used_gb:np.float64(155.41648197174072) - actor/lr:np.float64(0.0) - training/global_step:1 - training/epoch:0 - critic/score/mean:0.59765625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.59765625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.006173539441078901 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.006173539441078901 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:318.6640625 - response_length/max:884.0 - response_length/min:145.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:318.6640625 - response_length_non_aborted/max:884.0 - response_length_non_aborted/min:145.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:258.0 - prompt_length/max:491.0 - prompt_length/min:161.0 - 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:0.00023197568953037262 - timing_s/agent_loop/generate_sequences/min:np.float64(11.510200671851635) - timing_s/agent_loop/generate_sequences/max:np.float64(21.49765637330711) - timing_s/agent_loop/generate_sequences/mean:np.float64(16.448533715833037) - timing_s/
|
||
|
|
step:2 - global_seqlen/min:16549 - global_seqlen/max:27156 - global_seqlen/minmax_diff:10607 - global_seqlen/balanced_min:20751 - global_seqlen/balanced_max:20755 - global_seqlen/mean:20752.5 - actor/entropy:0.22268252074718475 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5311315655708313 - training/rollout_probs_diff_mean:0.006163785699754953 - training/rollout_probs_diff_std:0.016367914155125618 - training/rollout_actor_probs_pearson_corr:0.9972761273384094 - self_distillation/success_group_fraction:0.5625 - self_distillation/success_sample_fraction:0.5546875 - self_distillation/correct_sample_fraction:0.3828125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.5546875 - rollout_corr/rollout_is_mean:0.9998151659965515 - rollout_corr/rollout_is_ratio_fraction_high:3.561422636266798e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0001661997230257839 - rollout_corr/rollout_is_max:4.228911399841309 - rollout_corr/rollout_is_min:0.06694962829351425 - rollout_corr/rollout_is_std:0.047955457121133804 - rollout_corr/rollout_is_eff_sample_size:0.9977046606248142 - rollout_corr/rollout_is_seq_mean:0.9998382329940796 - rollout_corr/rollout_is_seq_std:0.003227078355848789 - rollout_corr/rollout_is_seq_max:1.0077893733978271 - rollout_corr/rollout_is_seq_min:0.9878386855125427 - rollout_corr/rollout_is_seq_max_deviation:0.012161314487457275 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2757157147862017) - rollout_corr/training_log_ppl:np.float64(0.24050413677468896) - rollout_corr/kl:np.float64(0.001426290413434117) - rollout_corr/k3_kl:np.float64(0.0012644620865103207) - rollout_corr/rollout_ppl:np.float64(1.2738774102181196) - rollout_corr/rollout_log_ppl:np.float64(0.2390778448607307) - rollout_corr/log_ppl_diff:np.float64(0.0014262919139582664) - rollout_corr/log_ppl_abs_diff:np.float64(0.002835439838236198) - rollout_corr/log_ppl_diff_max:np.float64(0.014306873083114624) - rollout_corr/log_ppl_diff_min:np.float64(-0.006757855415344238) - rollout_corr/ppl_ratio:np.float64(1.0014327545650303) - rollout_corr/chi2_token:np.float64(0.0021655673626810312) - rollout_corr/chi2_seq:np.float64(1.164014260750264) - actor/pg_loss:np.float64(0.027266473734925967) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.4453125) - actor/grad_norm:np.float64(0.7218142151832581) - perf/mfu/actor:np.float64(0.037373811583577915) - perf/max_memory_allocated_gb:np.float64(118.07343626022339) - perf/max_memory_reserved_gb:np.float64(123.22265625) - perf/cpu_memory_used_gb:np.float64(148.7036018371582) - actor/lr:np.float64(1.0000000000000002e-06) - training/global_step:2 - training/epoch:0 - critic/score/mean:0.3828125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.3828125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.010374127887189388 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.010374127887189388 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:329.046875 - response_length/max:810.0 - response_length/min:133.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:329.046875 - response_length_non_aborted/max:810.0 - response_length_non_aborted/min:133.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:319.46875 - prompt_length/max:818.0 - prompt_length/min:166.0 - 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:0.00016235560178756714 - timing_s/agent_loop/generate_sequences/min:np.float64(1.724344789981842) - timing_s/agent_loop/generate_sequences/max:np.float64(6.477641565725207) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.687603
|
||
|
|
step:3 - global_seqlen/min:16739 - global_seqlen/max:24204 - global_seqlen/minmax_diff:7465 - global_seqlen/balanced_min:20076 - global_seqlen/balanced_max:20080 - global_seqlen/mean:20078.875 - actor/entropy:0.23162508010864258 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.546594500541687 - training/rollout_probs_diff_mean:0.0060343313962221146 - training/rollout_probs_diff_std:0.015935607254505157 - training/rollout_actor_probs_pearson_corr:0.9974333643913269 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.734375 - self_distillation/correct_sample_fraction:0.4296875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.734375 - rollout_corr/rollout_is_mean:1.0001411437988281 - rollout_corr/rollout_is_ratio_fraction_high:7.211625052150339e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0001562518737046048 - rollout_corr/rollout_is_max:4.887171268463135 - rollout_corr/rollout_is_min:0.0004863159265369177 - rollout_corr/rollout_is_std:0.04782913625240326 - rollout_corr/rollout_is_eff_sample_size:0.9977179510204519 - rollout_corr/rollout_is_seq_mean:1.0001037120819092 - rollout_corr/rollout_is_seq_std:0.002903043758124113 - rollout_corr/rollout_is_seq_max:1.0123001337051392 - rollout_corr/rollout_is_seq_min:0.9921995401382446 - rollout_corr/rollout_is_seq_max_deviation:0.01230013370513916 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2821754957549274) - rollout_corr/training_log_ppl:np.float64(0.24516546617087442) - rollout_corr/kl:np.float64(0.0012725044054349155) - rollout_corr/k3_kl:np.float64(0.0013762065538571733) - rollout_corr/rollout_ppl:np.float64(1.2805130160413682) - rollout_corr/rollout_log_ppl:np.float64(0.2438929595227819) - rollout_corr/log_ppl_diff:np.float64(0.0012725066480925307) - rollout_corr/log_ppl_abs_diff:np.float64(0.0024485511967213824) - rollout_corr/log_ppl_diff_max:np.float64(0.03015953302383423) - rollout_corr/log_ppl_diff_min:np.float64(-0.006423190236091614) - rollout_corr/ppl_ratio:np.float64(1.0012787310406566) - rollout_corr/chi2_token:np.float64(0.003098314395174384) - rollout_corr/chi2_seq:np.float64(0.5920473046135157) - actor/pg_loss:np.float64(0.0362066325469641) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.265625) - actor/grad_norm:np.float64(0.8659180402755737) - perf/mfu/actor:np.float64(0.036224287443068236) - perf/max_memory_allocated_gb:np.float64(118.11889410018921) - perf/max_memory_reserved_gb:np.float64(123.3203125) - perf/cpu_memory_used_gb:np.float64(147.53532791137695) - actor/lr:np.float64(2.0000000000000003e-06) - training/global_step:3 - training/epoch:0 - critic/score/mean:0.4296875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.4296875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.009714660234749317 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.009714660234749317 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:324.99609375 - response_length/max:758.0 - response_length/min:118.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:324.99609375 - response_length_non_aborted/max:758.0 - response_length_non_aborted/min:118.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:302.46875 - prompt_length/max:557.0 - prompt_length/min:174.0 - 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:2.8790906071662903e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.4628903046250343) - timing_s/agent_loop/generate_sequences/max:np.float64(6.519654957577586) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.679532
|
||
|
|
step:4 - global_seqlen/min:15139 - global_seqlen/max:22092 - global_seqlen/minmax_diff:6953 - global_seqlen/balanced_min:19327 - global_seqlen/balanced_max:19332 - global_seqlen/mean:19330.125 - actor/entropy:0.27335160970687866 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3534165024757385 - training/rollout_probs_diff_mean:0.006507571786642075 - training/rollout_probs_diff_std:0.01534856203943491 - training/rollout_actor_probs_pearson_corr:0.9978428483009338 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.6328125 - self_distillation/correct_sample_fraction:0.33203125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.6328125 - rollout_corr/rollout_is_mean:0.9999235272407532 - rollout_corr/rollout_is_ratio_fraction_high:2.6273957701050676e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0001050958308042027 - rollout_corr/rollout_is_max:5.859992980957031 - rollout_corr/rollout_is_min:0.313480406999588 - rollout_corr/rollout_is_std:0.047640588134527206 - rollout_corr/rollout_is_eff_sample_size:0.9977348018525503 - rollout_corr/rollout_is_seq_mean:1.000044822692871 - rollout_corr/rollout_is_seq_std:0.003276826348155737 - rollout_corr/rollout_is_seq_max:1.010356068611145 - rollout_corr/rollout_is_seq_min:0.991000235080719 - rollout_corr/rollout_is_seq_max_deviation:0.01035606861114502 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.327166048809886) - rollout_corr/training_log_ppl:np.float64(0.27987524002674036) - rollout_corr/kl:np.float64(0.0011668601695653535) - rollout_corr/k3_kl:np.float64(0.0012117288727040432) - rollout_corr/rollout_ppl:np.float64(1.325589036103338) - rollout_corr/rollout_log_ppl:np.float64(0.2787083765870193) - rollout_corr/log_ppl_diff:np.float64(0.0011668634397210553) - rollout_corr/log_ppl_abs_diff:np.float64(0.0026237912970827892) - rollout_corr/log_ppl_diff_max:np.float64(0.010405957698822021) - rollout_corr/log_ppl_diff_min:np.float64(-0.007944285869598389) - rollout_corr/ppl_ratio:np.float64(1.0011725865770131) - rollout_corr/chi2_token:np.float64(0.0027140690945088863) - rollout_corr/chi2_seq:np.float64(0.9943714232649654) - actor/pg_loss:np.float64(0.030126508223474957) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.3671875) - actor/grad_norm:np.float64(0.6699938774108887) - perf/mfu/actor:np.float64(0.03500774102350318) - perf/max_memory_allocated_gb:np.float64(118.11889410018921) - perf/max_memory_reserved_gb:np.float64(123.3203125) - perf/cpu_memory_used_gb:np.float64(145.8373146057129) - actor/lr:np.float64(3e-06) - training/global_step:4 - training/epoch:0 - critic/score/mean:0.33203125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.33203125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0037752394564449787 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0037752394564449787 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:297.34765625 - response_length/max:728.0 - response_length/min:113.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:297.34765625 - response_length_non_aborted/max:728.0 - response_length_non_aborted/min:113.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:306.71875 - prompt_length/max:604.0 - prompt_length/min:173.0 - 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:0.00011218339204788208 - timing_s/agent_loop/generate_sequences/min:np.float64(1.4537068381905556) - timing_s/agent_loop/generate_sequences/max:np.float64(5.953867848962545) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.435880078803166) -
|
||
|
|
step:5 - global_seqlen/min:15086 - global_seqlen/max:20715 - global_seqlen/minmax_diff:5629 - global_seqlen/balanced_min:18157 - global_seqlen/balanced_max:18184 - global_seqlen/mean:18161.875 - actor/entropy:0.3271985352039337 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3507917523384094 - training/rollout_probs_diff_mean:0.006202250253409147 - training/rollout_probs_diff_std:0.013846217654645443 - training/rollout_actor_probs_pearson_corr:0.9984614849090576 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.81640625 - self_distillation/correct_sample_fraction:0.4609375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.81640625 - rollout_corr/rollout_is_mean:0.9997859001159668 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.658195808180608e-05 - rollout_corr/rollout_is_max:1.982527256011963 - rollout_corr/rollout_is_min:0.2963576316833496 - rollout_corr/rollout_is_std:0.04396497458219528 - rollout_corr/rollout_is_eff_sample_size:0.9980699193202004 - rollout_corr/rollout_is_seq_mean:0.9997933506965637 - rollout_corr/rollout_is_seq_std:0.002784077310934663 - rollout_corr/rollout_is_seq_max:1.0092980861663818 - rollout_corr/rollout_is_seq_min:0.9903117418289185 - rollout_corr/rollout_is_seq_max_deviation:0.009688258171081543 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4110902701504529) - rollout_corr/training_log_ppl:np.float64(0.339965935156215) - rollout_corr/kl:np.float64(0.001222059527655972) - rollout_corr/k3_kl:np.float64(0.0010154702323461606) - rollout_corr/rollout_ppl:np.float64(1.409346668049693) - rollout_corr/rollout_log_ppl:np.float64(0.3387438757927157) - rollout_corr/log_ppl_diff:np.float64(0.0012220593634992838) - rollout_corr/log_ppl_abs_diff:np.float64(0.0023589552147313952) - rollout_corr/log_ppl_diff_max:np.float64(0.011199504137039185) - rollout_corr/log_ppl_diff_min:np.float64(-0.008273154497146606) - rollout_corr/ppl_ratio:np.float64(1.0012267152778804) - rollout_corr/chi2_token:np.float64(0.0016158667858690023) - rollout_corr/chi2_seq:np.float64(0.3493654781486839) - actor/pg_loss:np.float64(0.04200247688277159) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.18359375) - actor/grad_norm:np.float64(0.49761390686035156) - perf/mfu/actor:np.float64(0.033210177701025974) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(123.3203125) - perf/cpu_memory_used_gb:np.float64(146.1409101486206) - actor/lr:np.float64(4.000000000000001e-06) - training/global_step:5 - training/epoch:0 - critic/score/mean:0.4609375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.4609375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.002729634987190366 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.002729634987190366 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:293.90234375 - response_length/max:990.0 - response_length/min:134.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:293.90234375 - response_length_non_aborted/max:990.0 - response_length_non_aborted/min:134.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:273.65625 - prompt_length/max:520.0 - prompt_length/min:166.0 - 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:0.00011437572538852692 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9757468439638615) - timing_s/agent_loop/generate_sequences/max:np.float64(9.223095504567027) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.2668456753890496) -
|
||
|
|
step:6 - global_seqlen/min:14100 - global_seqlen/max:17630 - global_seqlen/minmax_diff:3530 - global_seqlen/balanced_min:16190 - global_seqlen/balanced_max:16194 - global_seqlen/mean:16192.625 - actor/entropy:0.43406882882118225 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3764858841896057 - training/rollout_probs_diff_mean:0.006105047184973955 - training/rollout_probs_diff_std:0.012925864197313786 - training/rollout_actor_probs_pearson_corr:0.9988852143287659 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.7421875 - self_distillation/correct_sample_fraction:0.50390625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7421875 - rollout_corr/rollout_is_mean:1.0000123977661133 - rollout_corr/rollout_is_ratio_fraction_high:1.689731470833067e-05 - rollout_corr/rollout_is_ratio_fraction_low:5.069194594398141e-05 - rollout_corr/rollout_is_max:2.24564790725708 - rollout_corr/rollout_is_min:0.4247460663318634 - rollout_corr/rollout_is_std:0.04330727085471153 - rollout_corr/rollout_is_eff_sample_size:0.9981279912369095 - rollout_corr/rollout_is_seq_mean:1.000095248222351 - rollout_corr/rollout_is_seq_std:0.0030568751972168684 - rollout_corr/rollout_is_seq_max:1.0088915824890137 - rollout_corr/rollout_is_seq_min:0.9878351092338562 - rollout_corr/rollout_is_seq_max_deviation:0.012164890766143799 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5746679143048823) - rollout_corr/training_log_ppl:np.float64(0.4427429129136726) - rollout_corr/kl:np.float64(0.0008846975621445985) - rollout_corr/k3_kl:np.float64(0.0009799356969324435) - rollout_corr/rollout_ppl:np.float64(1.573122835252434) - rollout_corr/rollout_log_ppl:np.float64(0.4418582087964751) - rollout_corr/log_ppl_diff:np.float64(0.0008847041171975434) - rollout_corr/log_ppl_abs_diff:np.float64(0.0024411255144514143) - rollout_corr/log_ppl_diff_max:np.float64(0.014471888542175293) - rollout_corr/log_ppl_diff_min:np.float64(-0.007067233324050903) - rollout_corr/ppl_ratio:np.float64(1.000889919232577) - rollout_corr/chi2_token:np.float64(0.0021435413509607315) - rollout_corr/chi2_seq:np.float64(0.3887940801214427) - actor/pg_loss:np.float64(0.04280667805869598) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.2578125) - actor/grad_norm:np.float64(0.7476726770401001) - perf/mfu/actor:np.float64(0.029758548920456464) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(123.3203125) - perf/cpu_memory_used_gb:np.float64(145.06639766693115) - actor/lr:np.float64(5e-06) - training/global_step:6 - 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:-0.004900221247226 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.004900221247226 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:231.17578125 - response_length/max:629.0 - response_length/min:111.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:231.17578125 - response_length_non_aborted/max:629.0 - response_length_non_aborted/min:111.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:274.84375 - prompt_length/max:450.0 - prompt_length/min:173.0 - 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:3.952533006668091e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.2639408763498068) - timing_s/agent_loop/generate_sequences/max:np.float64(4.928012106567621) - timing_s/agent_loop/generate_sequences/mean:np.float64(2.5894566696006223) - timin
|
||
|
|
step:7 - global_seqlen/min:13062 - global_seqlen/max:17397 - global_seqlen/minmax_diff:4335 - global_seqlen/balanced_min:15200 - global_seqlen/balanced_max:15201 - global_seqlen/mean:15200.875 - actor/entropy:0.4080820679664612 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.30160319805145264 - training/rollout_probs_diff_mean:0.0062041026540100574 - training/rollout_probs_diff_std:0.012883169576525688 - training/rollout_actor_probs_pearson_corr:0.9988173842430115 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65234375 - self_distillation/correct_sample_fraction:0.4453125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.65234375 - rollout_corr/rollout_is_mean:0.9999308586120605 - rollout_corr/rollout_is_ratio_fraction_high:2.0998257241444662e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.0903611183166504 - rollout_corr/rollout_is_min:0.5379146933555603 - rollout_corr/rollout_is_std:0.04245239868760109 - rollout_corr/rollout_is_eff_sample_size:0.9982006797220615 - rollout_corr/rollout_is_seq_mean:0.9999265670776367 - rollout_corr/rollout_is_seq_std:0.0033248269464820623 - rollout_corr/rollout_is_seq_max:1.011274814605713 - rollout_corr/rollout_is_seq_min:0.9910061955451965 - rollout_corr/rollout_is_seq_max_deviation:0.01127481460571289 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5216813334263861) - rollout_corr/training_log_ppl:np.float64(0.413758575625252) - rollout_corr/kl:np.float64(0.0009868774781249812) - rollout_corr/k3_kl:np.float64(0.000913497415240272) - rollout_corr/rollout_ppl:np.float64(1.520152977667749) - rollout_corr/rollout_log_ppl:np.float64(0.4127716938382946) - rollout_corr/log_ppl_diff:np.float64(0.0009868817869573832) - rollout_corr/log_ppl_abs_diff:np.float64(0.002786013879813254) - rollout_corr/log_ppl_diff_max:np.float64(0.01028035581111908) - rollout_corr/log_ppl_diff_min:np.float64(-0.008775472640991211) - rollout_corr/ppl_ratio:np.float64(1.0009928627405316) - rollout_corr/chi2_token:np.float64(0.0016839783638715744) - rollout_corr/chi2_seq:np.float64(0.3703045754227787) - actor/pg_loss:np.float64(0.03941588809539098) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.34765625) - actor/grad_norm:np.float64(0.5641831159591675) - perf/mfu/actor:np.float64(0.027968700723935275) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(123.3203125) - perf/cpu_memory_used_gb:np.float64(144.2918930053711) - actor/lr:np.float64(6e-06) - training/global_step:7 - training/epoch:0 - critic/score/mean:0.4453125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.4453125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.000837305502500385 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.000837305502500385 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:186.02734375 - response_length/max:423.0 - response_length/min:89.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:186.02734375 - response_length_non_aborted/max:423.0 - response_length_non_aborted/min:89.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:289.0 - prompt_length/max:662.0 - prompt_length/min:169.0 - 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:0.0001510120928287506 - timing_s/agent_loop/generate_sequences/min:np.float64(0.7814213987439871) - timing_s/agent_loop/generate_sequences/max:np.float64(3.5883758813142776) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.9397942674841033) - timing_s/agent_loop/
|
||
|
|
step:8 - global_seqlen/min:11672 - global_seqlen/max:17308 - global_seqlen/minmax_diff:5636 - global_seqlen/balanced_min:13460 - global_seqlen/balanced_max:13462 - global_seqlen/mean:13461.375 - actor/entropy:0.27522507309913635 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7064226865768433 - training/rollout_probs_diff_mean:0.006937616970390081 - training/rollout_probs_diff_std:0.019912919029593468 - training/rollout_actor_probs_pearson_corr:0.9966567158699036 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.7109375 - self_distillation/correct_sample_fraction:0.57421875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7109375 - rollout_corr/rollout_is_mean:0.999876856803894 - rollout_corr/rollout_is_ratio_fraction_high:0.00018291009473614395 - rollout_corr/rollout_is_ratio_fraction_low:0.0003658201894722879 - rollout_corr/rollout_is_max:4.260913372039795 - rollout_corr/rollout_is_min:0.10023043304681778 - rollout_corr/rollout_is_std:0.05164194479584694 - rollout_corr/rollout_is_eff_sample_size:0.9973387205694644 - rollout_corr/rollout_is_seq_mean:0.9997835159301758 - rollout_corr/rollout_is_seq_std:0.005452709272503853 - rollout_corr/rollout_is_seq_max:1.026816487312317 - rollout_corr/rollout_is_seq_min:0.9780719876289368 - rollout_corr/rollout_is_seq_max_deviation:0.026816487312316895 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3366427612490952) - rollout_corr/training_log_ppl:np.float64(0.2849578551540617) - rollout_corr/kl:np.float64(0.0019782749035015534) - rollout_corr/k3_kl:np.float64(0.0017617729112089364) - rollout_corr/rollout_ppl:np.float64(1.333981245290488) - rollout_corr/rollout_log_ppl:np.float64(0.28297958042821847) - rollout_corr/log_ppl_diff:np.float64(0.001978274725843221) - rollout_corr/log_ppl_abs_diff:np.float64(0.004137693555094302) - rollout_corr/log_ppl_diff_max:np.float64(0.038967013359069824) - rollout_corr/log_ppl_diff_min:np.float64(-0.011893272399902344) - rollout_corr/ppl_ratio:np.float64(1.0019974003080279) - rollout_corr/chi2_token:np.float64(0.003419409738853574) - rollout_corr/chi2_seq:np.float64(0.3262675201985985) - actor/pg_loss:np.float64(0.05070620784681523) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.2890625) - actor/grad_norm:np.float64(0.8857548832893372) - perf/mfu/actor:np.float64(0.024539861708888486) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(123.3203125) - perf/cpu_memory_used_gb:np.float64(144.2158465385437) - actor/lr:np.float64(7e-06) - training/global_step:8 - training/epoch:0 - critic/score/mean:0.57421875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.57421875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0004115477204322815 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0004115477204322815 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:128.13671875 - response_length/max:252.0 - response_length/min:71.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:128.13671875 - response_length_non_aborted/max:252.0 - response_length_non_aborted/min:71.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:292.53125 - prompt_length/max:720.0 - prompt_length/min:174.0 - 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:0.00011466443538665771 - timing_s/agent_loop/generate_sequences/min:np.float64(0.8904057834297419) - timing_s/agent_loop/generate_sequences/max:np.float64(2.274159148335457) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.4984804912091931)
|
||
|
|
step:9 - global_seqlen/min:11409 - global_seqlen/max:16207 - global_seqlen/minmax_diff:4798 - global_seqlen/balanced_min:13540 - global_seqlen/balanced_max:13541 - global_seqlen/mean:13540.625 - actor/entropy:0.2833990454673767 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.25833648443222046 - training/rollout_probs_diff_mean:0.006623577326536179 - training/rollout_probs_diff_std:0.01580883003771305 - training/rollout_actor_probs_pearson_corr:0.9977806210517883 - self_distillation/success_group_fraction:0.625 - self_distillation/success_sample_fraction:0.6171875 - self_distillation/correct_sample_fraction:0.3515625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.6171875 - rollout_corr/rollout_is_mean:0.9999385476112366 - rollout_corr/rollout_is_ratio_fraction_high:0.00011634332622634247 - rollout_corr/rollout_is_ratio_fraction_low:8.725749648874626e-05 - rollout_corr/rollout_is_max:2.649088144302368 - rollout_corr/rollout_is_min:0.4432407319545746 - rollout_corr/rollout_is_std:0.04844822734594345 - rollout_corr/rollout_is_eff_sample_size:0.9976576724221284 - rollout_corr/rollout_is_seq_mean:1.0000391006469727 - rollout_corr/rollout_is_seq_std:0.004701168742030859 - rollout_corr/rollout_is_seq_max:1.0184412002563477 - rollout_corr/rollout_is_seq_min:0.9831221699714661 - rollout_corr/rollout_is_seq_max_deviation:0.018441200256347656 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.3369720592163503) - rollout_corr/training_log_ppl:np.float64(0.2847705118911108) - rollout_corr/kl:np.float64(0.0011767218074965058) - rollout_corr/k3_kl:np.float64(0.0012158670938049454) - rollout_corr/rollout_ppl:np.float64(1.3353116400539875) - rollout_corr/rollout_log_ppl:np.float64(0.283593786531128) - rollout_corr/log_ppl_diff:np.float64(0.001176725359982811) - rollout_corr/log_ppl_abs_diff:np.float64(0.003689644145197235) - rollout_corr/log_ppl_diff_max:np.float64(0.018553972244262695) - rollout_corr/log_ppl_diff_min:np.float64(-0.013477608561515808) - rollout_corr/ppl_ratio:np.float64(1.0011883135885) - rollout_corr/chi2_token:np.float64(0.00258400640450418) - rollout_corr/chi2_seq:np.float64(0.6911876178346574) - actor/pg_loss:np.float64(0.04373098009091336) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.3828125) - actor/grad_norm:np.float64(0.6020697951316833) - perf/mfu/actor:np.float64(0.024140655048275776) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(123.3203125) - perf/cpu_memory_used_gb:np.float64(146.42454147338867) - actor/lr:np.float64(8.000000000000001e-06) - training/global_step:9 - training/epoch:0 - critic/score/mean:0.3515625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.3515625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.003544835839420557 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.003544835839420557 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:134.30078125 - response_length/max:312.0 - response_length/min:65.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:134.30078125 - response_length_non_aborted/max:312.0 - response_length_non_aborted/min:65.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:288.84375 - prompt_length/max:735.0 - prompt_length/min:167.0 - 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:0.00020870566368103027 - timing_s/agent_loop/generate_sequences/min:np.float64(0.8015764281153679) - timing_s/agent_loop/generate_sequences/max:np.float64(2.5898707658052444) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.547275834709
|
||
|
|
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 10}
|
||
|
|
validation generation end
|
||
|
|
[prompt] system
|
||
|
|
|
||
|
|
Given a question and four options, please select the right answer. Respond in the following format:
|
||
|
|
<reasoning>
|
||
|
|
...
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
...
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
user
|
||
|
|
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||
|
|
|
||
|
|
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||
|
|
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||
|
|
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||
|
|
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||
|
|
Please reason step by step.
|
||
|
|
assistant
|
||
|
|
<think>
|
||
|
|
|
||
|
|
</think>
|
||
|
|
|
||
|
|
|
||
|
|
[response] <reasoning>
|
||
|
|
The product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" contains a nitrogen atom bonded to a chlorine-containing aromatic ring and a trifluorinated group. Option A includes a nitrogen atom, a chlorine-containing aromatic ring, and a trifluorinated group, matching the product's structure. The other options either lack the correct functional groups or have incorrect structural features. Therefore, the correct answer is A.
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
[ground_truth] A
|
||
|
|
[score] 1.0
|
||
|
|
[acc] 1.0
|
||
|
|
[pred] A
|
||
|
|
[incorrect_format] 1
|
||
|
|
[feedback]
|
||
|
|
local_global_step_folder: /workspace/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-SDPO_TR-Qwen-Qwen3-4B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_10
|
||
|
|
step:10 - global_seqlen/min:12737 - global_seqlen/max:17597 - global_seqlen/minmax_diff:4860 - global_seqlen/balanced_min:14199 - global_seqlen/balanced_max:14200 - global_seqlen/mean:14199.125 - actor/entropy:0.2620508670806885 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.35831528902053833 - training/rollout_probs_diff_mean:0.005596653092652559 - training/rollout_probs_diff_std:0.013680397532880306 - training/rollout_actor_probs_pearson_corr:0.9982901811599731 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.64453125 - self_distillation/correct_sample_fraction:0.5390625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.64453125 - rollout_corr/rollout_is_mean:1.000045657157898 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:4.48098944616504e-05 - rollout_corr/rollout_is_max:1.6804662942886353 - rollout_corr/rollout_is_min:0.40561527013778687 - rollout_corr/rollout_is_std:0.04217348247766495 - rollout_corr/rollout_is_eff_sample_size:0.9982247927834103 - rollout_corr/rollout_is_seq_mean:1.00006103515625 - rollout_corr/rollout_is_seq_std:0.003654802218079567 - rollout_corr/rollout_is_seq_max:1.011189341545105 - rollout_corr/rollout_is_seq_min:0.9871388077735901 - rollout_corr/rollout_is_seq_max_deviation:0.012861192226409912 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.311476762406528) - rollout_corr/training_log_ppl:np.float64(0.26662622918956913) - rollout_corr/kl:np.float64(0.0008443429046884887) - rollout_corr/k3_kl:np.float64(0.0009053539913281838) - rollout_corr/rollout_ppl:np.float64(1.3103391309268773) - rollout_corr/rollout_log_ppl:np.float64(0.2657818852458149) - rollout_corr/log_ppl_diff:np.float64(0.0008443439437542111) - rollout_corr/log_ppl_abs_diff:np.float64(0.002889614464947954) - rollout_corr/log_ppl_diff_max:np.float64(0.014694839715957642) - rollout_corr/log_ppl_diff_min:np.float64(-0.009645640850067139) - rollout_corr/ppl_ratio:np.float64(1.0008515079971403) - rollout_corr/chi2_token:np.float64(0.0019165368285030127) - rollout_corr/chi2_seq:np.float64(0.6327612979803234) - actor/pg_loss:np.float64(0.036829199954809155) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.35546875) - actor/grad_norm:np.float64(0.41817277669906616) - perf/mfu/actor:np.float64(0.025895922503516626) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(123.3203125) - perf/cpu_memory_used_gb:np.float64(143.05279350280762) - actor/lr:np.float64(9e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.5589285714285714) - val-aux/sciknoweval/reward/std@16:np.float64(0.10730025207526221) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.6016333333333332) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.08997715767418266) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.5164190476190476) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.09089759358428012) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.5585952380952381) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.10754743025252103) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.6360095238095237) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.07011498544412628) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.48081904761904765) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.07012747146136279) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.5627333333333333) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.0821466285341773) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.6638857142857143) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.052625599160734865) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.4516809523809524) - val-aux/
|
||
|
|
step:11 - global_seqlen/min:13239 - global_seqlen/max:17141 - global_seqlen/minmax_diff:3902 - global_seqlen/balanced_min:15477 - global_seqlen/balanced_max:15479 - global_seqlen/mean:15478.0 - actor/entropy:0.245832622051239 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.17985576391220093 - training/rollout_probs_diff_mean:0.005206643138080835 - training/rollout_probs_diff_std:0.013023940846323967 - training/rollout_actor_probs_pearson_corr:0.9983668327331543 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.671875 - self_distillation/correct_sample_fraction:0.45703125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.671875 - rollout_corr/rollout_is_mean:0.999755322933197 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.00010580667003523558 - rollout_corr/rollout_is_max:1.8242530822753906 - rollout_corr/rollout_is_min:0.3737630546092987 - rollout_corr/rollout_is_std:0.04024915769696236 - rollout_corr/rollout_is_eff_sample_size:0.9983818531423422 - rollout_corr/rollout_is_seq_mean:0.9997861981391907 - rollout_corr/rollout_is_seq_std:0.003018396906554699 - rollout_corr/rollout_is_seq_max:1.0083627700805664 - rollout_corr/rollout_is_seq_min:0.9905517101287842 - rollout_corr/rollout_is_seq_max_deviation:0.00944828987121582 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2874454036355019) - rollout_corr/training_log_ppl:np.float64(0.24915726587641984) - rollout_corr/kl:np.float64(0.0010533498188713963) - rollout_corr/k3_kl:np.float64(0.0008395280552235818) - rollout_corr/rollout_ppl:np.float64(1.286038350313902) - rollout_corr/rollout_log_ppl:np.float64(0.24810391839127988) - rollout_corr/log_ppl_diff:np.float64(0.001053347485139966) - rollout_corr/log_ppl_abs_diff:np.float64(0.0025448576197959483) - rollout_corr/log_ppl_diff_max:np.float64(0.010707318782806396) - rollout_corr/log_ppl_diff_min:np.float64(-0.0073602646589279175) - rollout_corr/ppl_ratio:np.float64(1.0010586213320494) - rollout_corr/chi2_token:np.float64(0.0012290575541555882) - rollout_corr/chi2_seq:np.float64(0.1873805020004511) - actor/pg_loss:np.float64(0.03902373217715649) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.328125) - actor/grad_norm:np.float64(0.33753350377082825) - perf/mfu/actor:np.float64(0.0283027398106647) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(146.58620071411133) - actor/lr:np.float64(1e-05) - training/global_step:11 - training/epoch:0 - critic/score/mean:0.45703125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.45703125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.001724648755043745 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.001724648755043745 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:184.59375 - response_length/max:392.0 - response_length/min:94.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:184.59375 - response_length_non_aborted/max:392.0 - response_length_non_aborted/min:94.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:299.09375 - prompt_length/max:620.0 - prompt_length/min:172.0 - 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:7.653236389160156e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9252954758703709) - timing_s/agent_loop/generate_sequences/max:np.float64(3.379142278805375) - timing_s/agent_loop/generate_sequences/mean:np.float64(2.0594222103318316) - timing_s/agent_loop/tool_c
|
||
|
|
step:12 - global_seqlen/min:12918 - global_seqlen/max:16363 - global_seqlen/minmax_diff:3445 - global_seqlen/balanced_min:14135 - global_seqlen/balanced_max:14135 - global_seqlen/mean:14135.0 - actor/entropy:0.20064345002174377 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.21863090991973877 - training/rollout_probs_diff_mean:0.004423043690621853 - training/rollout_probs_diff_std:0.012335534207522869 - training/rollout_actor_probs_pearson_corr:0.998301088809967 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71875 - self_distillation/correct_sample_fraction:0.62109375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.71875 - rollout_corr/rollout_is_mean:1.0000988245010376 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.838175892829895 - rollout_corr/rollout_is_min:0.507088840007782 - rollout_corr/rollout_is_std:0.036836929619312286 - rollout_corr/rollout_is_eff_sample_size:0.9986452361598612 - rollout_corr/rollout_is_seq_mean:1.000101089477539 - rollout_corr/rollout_is_seq_std:0.0029456084594130516 - rollout_corr/rollout_is_seq_max:1.009503722190857 - rollout_corr/rollout_is_seq_min:0.9902786612510681 - rollout_corr/rollout_is_seq_max_deviation:0.009721338748931885 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2255469034425914) - rollout_corr/training_log_ppl:np.float64(0.20098816481186077) - rollout_corr/kl:np.float64(0.0005868169383180799) - rollout_corr/k3_kl:np.float64(0.0006878988006917552) - rollout_corr/rollout_ppl:np.float64(1.2248260988853872) - rollout_corr/rollout_log_ppl:np.float64(0.20040134745067917) - rollout_corr/log_ppl_diff:np.float64(0.0005868173611816019) - rollout_corr/log_ppl_abs_diff:np.float64(0.002345234708627686) - rollout_corr/log_ppl_diff_max:np.float64(0.011413633823394775) - rollout_corr/log_ppl_diff_min:np.float64(-0.008571013808250427) - rollout_corr/ppl_ratio:np.float64(1.0005913574714214) - rollout_corr/chi2_token:np.float64(0.001575350295752287) - rollout_corr/chi2_seq:np.float64(0.34188920934684575) - actor/pg_loss:np.float64(0.03507724807423074) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.28125) - actor/grad_norm:np.float64(0.3015119433403015) - perf/mfu/actor:np.float64(0.025797420252322314) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(146.13171768188477) - actor/lr:np.float64(1e-05) - training/global_step:12 - 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:-0.005813500843942165 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:-0.005813500843942165 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:175.625 - response_length/max:350.0 - response_length/min:106.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:175.625 - response_length_non_aborted/max:350.0 - response_length_non_aborted/min:106.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:266.09375 - prompt_length/max:665.0 - prompt_length/min:166.0 - 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:0.00011731870472431183 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1603798028081656) - timing_s/agent_loop/generate_sequences/max:np.float64(3.1245807837694883) - timing_s/agent_loop/generate_sequences/mean:np.float64(2.004497386064031) - timing_s/agent_loop/tool_calls/min:np.float6
|
||
|
|
step:13 - global_seqlen/min:12168 - global_seqlen/max:15328 - global_seqlen/minmax_diff:3160 - global_seqlen/balanced_min:13977 - global_seqlen/balanced_max:13978 - global_seqlen/mean:13977.875 - actor/entropy:0.18061794340610504 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1791781783103943 - training/rollout_probs_diff_mean:0.003985490649938583 - training/rollout_probs_diff_std:0.011418158188462257 - training/rollout_actor_probs_pearson_corr:0.9983763694763184 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.7421875 - self_distillation/correct_sample_fraction:0.6171875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7421875 - rollout_corr/rollout_is_mean:0.9999973177909851 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5526758432388306 - rollout_corr/rollout_is_min:0.6075006127357483 - rollout_corr/rollout_is_std:0.032720308750867844 - rollout_corr/rollout_is_eff_sample_size:0.9989304074145082 - rollout_corr/rollout_is_seq_mean:0.9999994039535522 - rollout_corr/rollout_is_seq_std:0.002727840794250369 - rollout_corr/rollout_is_seq_max:1.010908842086792 - rollout_corr/rollout_is_seq_min:0.9896724224090576 - rollout_corr/rollout_is_seq_max_deviation:0.010908842086791992 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.2000994230620563) - rollout_corr/training_log_ppl:np.float64(0.17956839984981343) - rollout_corr/kl:np.float64(0.000547381288143356) - rollout_corr/k3_kl:np.float64(0.000546786294762569) - rollout_corr/rollout_ppl:np.float64(1.1994146686047316) - rollout_corr/rollout_log_ppl:np.float64(0.17902101844083518) - rollout_corr/log_ppl_diff:np.float64(0.0005473814089782536) - rollout_corr/log_ppl_abs_diff:np.float64(0.0021539503359235823) - rollout_corr/log_ppl_diff_max:np.float64(0.011421173810958862) - rollout_corr/log_ppl_diff_min:np.float64(-0.009220808744430542) - rollout_corr/ppl_ratio:np.float64(1.000551262870431) - rollout_corr/chi2_token:np.float64(0.0010887153912335634) - rollout_corr/chi2_seq:np.float64(0.22349174809642136) - actor/pg_loss:np.float64(0.03645641265029553) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.2578125) - actor/grad_norm:np.float64(0.3401016891002655) - perf/mfu/actor:np.float64(0.025697298178361496) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.42429733276367) - actor/lr:np.float64(1e-05) - training/global_step:13 - 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:0.0006632840377278626 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0006632840377278626 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:159.74609375 - response_length/max:253.0 - response_length/min:101.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:159.74609375 - response_length_non_aborted/max:253.0 - response_length_non_aborted/min:101.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:277.0625 - prompt_length/max:612.0 - prompt_length/min:174.0 - 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:4.112161695957184e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.245823673903942) - timing_s/agent_loop/generate_sequences/max:np.float64(2.6271581407636404) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.921308981756738) - timing_s/agent_loop/tool_calls/min
|
||
|
|
step:14 - global_seqlen/min:11529 - global_seqlen/max:16640 - global_seqlen/minmax_diff:5111 - global_seqlen/balanced_min:13516 - global_seqlen/balanced_max:13517 - global_seqlen/mean:13516.375 - actor/entropy:0.14009033143520355 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3569830060005188 - training/rollout_probs_diff_mean:0.0033288171980530024 - training/rollout_probs_diff_std:0.011109726503491402 - training/rollout_actor_probs_pearson_corr:0.9982218742370605 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.6484375 - self_distillation/correct_sample_fraction:0.53125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.6484375 - rollout_corr/rollout_is_mean:0.9999690055847168 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:5.041974509367719e-05 - rollout_corr/rollout_is_max:1.597582459449768 - rollout_corr/rollout_is_min:0.141954705119133 - rollout_corr/rollout_is_std:0.03222101181745529 - rollout_corr/rollout_is_eff_sample_size:0.998962883025614 - rollout_corr/rollout_is_seq_mean:1.0000187158584595 - rollout_corr/rollout_is_seq_std:0.0030057828407734632 - rollout_corr/rollout_is_seq_max:1.0104013681411743 - rollout_corr/rollout_is_seq_min:0.9906222820281982 - rollout_corr/rollout_is_seq_max_deviation:0.010401368141174316 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1617755475454032) - rollout_corr/training_log_ppl:np.float64(0.14752614268218167) - rollout_corr/kl:np.float64(0.0005585174466666842) - rollout_corr/k3_kl:np.float64(0.0005772542892117372) - rollout_corr/rollout_ppl:np.float64(1.1610777680762112) - rollout_corr/rollout_log_ppl:np.float64(0.14696762428502552) - rollout_corr/log_ppl_diff:np.float64(0.0005585183971561491) - rollout_corr/log_ppl_abs_diff:np.float64(0.002331730880541727) - rollout_corr/log_ppl_diff_max:np.float64(0.012382254004478455) - rollout_corr/log_ppl_diff_min:np.float64(-0.00909518450498581) - rollout_corr/ppl_ratio:np.float64(1.0005634522531182) - rollout_corr/chi2_token:np.float64(0.0011569156777113676) - rollout_corr/chi2_seq:np.float64(0.17804991081357002) - actor/pg_loss:np.float64(0.02626531153509859) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.3515625) - actor/grad_norm:np.float64(0.28266641497612) - perf/mfu/actor:np.float64(0.025021210941076795) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.05509185791016) - actor/lr:np.float64(1e-05) - training/global_step:14 - training/epoch:0 - 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:-0.001392845413647592 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.001392845413647592 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:154.94921875 - response_length/max:266.0 - response_length/min:78.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:154.94921875 - response_length_non_aborted/max:266.0 - response_length_non_aborted/min:78.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:267.4375 - prompt_length/max:666.0 - prompt_length/min:172.0 - 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:0.00016381405293941498 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9650154206901789) - timing_s/agent_loop/generate_sequences/max:np.float64(2.542610328644514) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.8114038738858653) - timing_s/agent_loop/t
|
||
|
|
step:15 - global_seqlen/min:12377 - global_seqlen/max:15116 - global_seqlen/minmax_diff:2739 - global_seqlen/balanced_min:13733 - global_seqlen/balanced_max:13734 - global_seqlen/mean:13733.5 - actor/entropy:0.13181030750274658 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.16888189315795898 - training/rollout_probs_diff_mean:0.0032676400151103735 - training/rollout_probs_diff_std:0.011030502617359161 - training/rollout_actor_probs_pearson_corr:0.9981563091278076 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.6875 - self_distillation/correct_sample_fraction:0.65625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.6875 - rollout_corr/rollout_is_mean:0.9999507665634155 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.7457441319711506e-05 - rollout_corr/rollout_is_max:1.8359973430633545 - rollout_corr/rollout_is_min:0.44858238101005554 - rollout_corr/rollout_is_std:0.0322265625 - rollout_corr/rollout_is_eff_sample_size:0.9989624071772414 - rollout_corr/rollout_is_seq_mean:0.9999912977218628 - rollout_corr/rollout_is_seq_std:0.0027697246987372637 - rollout_corr/rollout_is_seq_max:1.0073965787887573 - rollout_corr/rollout_is_seq_min:0.9921057224273682 - rollout_corr/rollout_is_seq_max_deviation:0.007894277572631836 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1501588304527104) - rollout_corr/training_log_ppl:np.float64(0.13819306234654505) - rollout_corr/kl:np.float64(0.000540232164041754) - rollout_corr/k3_kl:np.float64(0.0005315027029126895) - rollout_corr/rollout_ppl:np.float64(1.1495192013680935) - rollout_corr/rollout_log_ppl:np.float64(0.13765282883832697) - rollout_corr/log_ppl_diff:np.float64(0.0005402335082180798) - rollout_corr/log_ppl_abs_diff:np.float64(0.00213820839417167) - rollout_corr/log_ppl_diff_max:np.float64(0.00860181450843811) - rollout_corr/log_ppl_diff_min:np.float64(-0.005442321300506592) - rollout_corr/ppl_ratio:np.float64(1.0005441138055176) - rollout_corr/chi2_token:np.float64(0.001065451418980956) - rollout_corr/chi2_seq:np.float64(0.08899404341354966) - actor/pg_loss:np.float64(0.02839167289494071) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.3125) - actor/grad_norm:np.float64(0.2899867296218872) - perf/mfu/actor:np.float64(0.025021423802485433) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(144.90049362182617) - actor/lr:np.float64(1e-05) - training/global_step:15 - training/epoch:0 - critic/score/mean:0.65625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.65625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0009095277055166662 - critic/advantages/max:0.5 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0009095277055166662 - critic/returns/max:0.5 - critic/returns/min:-0.875 - response_length/mean:142.265625 - response_length/max:270.0 - response_length/min:94.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:142.265625 - response_length_non_aborted/max:270.0 - response_length_non_aborted/min:94.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:286.90625 - prompt_length/max:494.0 - prompt_length/min:172.0 - 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:0.0001357048749923706 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1607603561133146) - timing_s/agent_loop/generate_sequences/max:np.float64(2.5490803066641092) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.692190005313023) - timing_s/agent_loop/tool_calls/min:np.flo
|
||
|
|
step:16 - global_seqlen/min:12398 - global_seqlen/max:16595 - global_seqlen/minmax_diff:4197 - global_seqlen/balanced_min:13722 - global_seqlen/balanced_max:13723 - global_seqlen/mean:13722.75 - actor/entropy:0.12278510630130768 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.17904314398765564 - training/rollout_probs_diff_mean:0.003188177477568388 - training/rollout_probs_diff_std:0.010851170867681503 - training/rollout_actor_probs_pearson_corr:0.9980733394622803 - self_distillation/success_group_fraction:0.625 - self_distillation/success_sample_fraction:0.62109375 - self_distillation/correct_sample_fraction:0.4765625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.62109375 - rollout_corr/rollout_is_mean:0.999843180179596 - rollout_corr/rollout_is_ratio_fraction_high:2.637269972183276e-05 - rollout_corr/rollout_is_ratio_fraction_low:7.911809370853007e-05 - rollout_corr/rollout_is_max:2.046469211578369 - rollout_corr/rollout_is_min:0.43286189436912537 - rollout_corr/rollout_is_std:0.03147994354367256 - rollout_corr/rollout_is_eff_sample_size:0.9990096373386255 - rollout_corr/rollout_is_seq_mean:0.999832034111023 - rollout_corr/rollout_is_seq_std:0.0025905396323651075 - rollout_corr/rollout_is_seq_max:1.0108973979949951 - rollout_corr/rollout_is_seq_min:0.9911273717880249 - rollout_corr/rollout_is_seq_max_deviation:0.010897397994995117 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.137071009259671) - rollout_corr/training_log_ppl:np.float64(0.1268506429187255) - rollout_corr/kl:np.float64(0.0006901941260615274) - rollout_corr/k3_kl:np.float64(0.0005222326775822239) - rollout_corr/rollout_ppl:np.float64(1.1362546873278916) - rollout_corr/rollout_log_ppl:np.float64(0.12616044832975604) - rollout_corr/log_ppl_diff:np.float64(0.0006901945889694616) - rollout_corr/log_ppl_abs_diff:np.float64(0.002068179557682015) - rollout_corr/log_ppl_diff_max:np.float64(0.012592673301696777) - rollout_corr/log_ppl_diff_min:np.float64(-0.009294696152210236) - rollout_corr/ppl_ratio:np.float64(1.000693894457072) - rollout_corr/chi2_token:np.float64(0.0007051748689264059) - rollout_corr/chi2_seq:np.float64(0.06535047944635153) - actor/pg_loss:np.float64(0.025590789773559663) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.37890625) - actor/grad_norm:np.float64(0.2646765112876892) - perf/mfu/actor:np.float64(0.025085303146117302) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(144.98699951171875) - actor/lr:np.float64(1e-05) - training/global_step:16 - training/epoch:0 - critic/score/mean:0.4765625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.4765625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:9.230444993590936e-05 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:9.230444993590936e-05 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:148.1171875 - response_length/max:243.0 - response_length/min:92.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:148.1171875 - response_length_non_aborted/max:243.0 - response_length_non_aborted/min:92.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:280.71875 - prompt_length/max:859.0 - prompt_length/min:172.0 - 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:3.0074268579483032e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1252746246755123) - timing_s/agent_loop/generate_sequences/max:np.float64(2.525325696915388) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.786482279538177)
|
||
|
|
step:17 - global_seqlen/min:12549 - global_seqlen/max:18425 - global_seqlen/minmax_diff:5876 - global_seqlen/balanced_min:15298 - global_seqlen/balanced_max:15299 - global_seqlen/mean:15298.75 - actor/entropy:0.12111552804708481 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18452316522598267 - training/rollout_probs_diff_mean:0.0030965162441134453 - training/rollout_probs_diff_std:0.010908165946602821 - training/rollout_actor_probs_pearson_corr:0.9980682730674744 - self_distillation/success_group_fraction:0.53125 - self_distillation/success_sample_fraction:0.52734375 - self_distillation/correct_sample_fraction:0.41796875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.52734375 - rollout_corr/rollout_is_mean:0.9998281002044678 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.6964353310177103e-05 - rollout_corr/rollout_is_max:1.5695059299468994 - rollout_corr/rollout_is_min:0.44206640124320984 - rollout_corr/rollout_is_std:0.03112003020942211 - rollout_corr/rollout_is_eff_sample_size:0.9990323617530771 - rollout_corr/rollout_is_seq_mean:0.999880313873291 - rollout_corr/rollout_is_seq_std:0.0026432638987898827 - rollout_corr/rollout_is_seq_max:1.0078970193862915 - rollout_corr/rollout_is_seq_min:0.9897641539573669 - rollout_corr/rollout_is_seq_max_deviation:0.010235846042633057 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1366667286492884) - rollout_corr/training_log_ppl:np.float64(0.1264654328697361) - rollout_corr/kl:np.float64(0.0006132450980693704) - rollout_corr/k3_kl:np.float64(0.0004935829724210805) - rollout_corr/rollout_ppl:np.float64(1.1359531278721988) - rollout_corr/rollout_log_ppl:np.float64(0.12585218709136825) - rollout_corr/log_ppl_diff:np.float64(0.0006132457783678547) - rollout_corr/log_ppl_abs_diff:np.float64(0.0021326909627532586) - rollout_corr/log_ppl_diff_max:np.float64(0.012381181120872498) - rollout_corr/log_ppl_diff_min:np.float64(-0.00694902241230011) - rollout_corr/ppl_ratio:np.float64(1.000617123907432) - rollout_corr/chi2_token:np.float64(0.0007337189745157957) - rollout_corr/chi2_seq:np.float64(0.10327930352650583) - actor/pg_loss:np.float64(0.021802378425491042) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.47265625) - actor/grad_norm:np.float64(0.2583705186843872) - perf/mfu/actor:np.float64(0.02777774504408589) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.104829788208) - actor/lr:np.float64(1e-05) - training/global_step:17 - training/epoch:0 - critic/score/mean:0.41796875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.41796875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0025245374999940395 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0025245374999940395 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:144.8671875 - response_length/max:290.0 - response_length/min:97.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:144.8671875 - response_length_non_aborted/max:290.0 - response_length_non_aborted/min:97.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:333.21875 - prompt_length/max:617.0 - 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:9.709596633911133e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.101479910314083) - timing_s/agent_loop/generate_sequences/max:np.float64(2.685517242178321) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.7000412998677348) - timing_
|
||
|
|
step:18 - global_seqlen/min:11361 - global_seqlen/max:15184 - global_seqlen/minmax_diff:3823 - global_seqlen/balanced_min:13517 - global_seqlen/balanced_max:13518 - global_seqlen/mean:13517.625 - actor/entropy:0.12062487006187439 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2007083296775818 - training/rollout_probs_diff_mean:0.003090393263846636 - training/rollout_probs_diff_std:0.010683072730898857 - training/rollout_actor_probs_pearson_corr:0.9981490969657898 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65234375 - self_distillation/correct_sample_fraction:0.515625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.65234375 - rollout_corr/rollout_is_mean:0.9999257326126099 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5891640186309814 - rollout_corr/rollout_is_min:0.6099143624305725 - rollout_corr/rollout_is_std:0.030991438776254654 - rollout_corr/rollout_is_eff_sample_size:0.9990403333894667 - rollout_corr/rollout_is_seq_mean:0.9998982548713684 - rollout_corr/rollout_is_seq_std:0.002511096652597189 - rollout_corr/rollout_is_seq_max:1.0092535018920898 - rollout_corr/rollout_is_seq_min:0.9916408658027649 - rollout_corr/rollout_is_seq_max_deviation:0.009253501892089844 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1343218823894858) - rollout_corr/training_log_ppl:np.float64(0.12449584400019376) - rollout_corr/kl:np.float64(0.0005990976934242553) - rollout_corr/k3_kl:np.float64(0.0004973417921476653) - rollout_corr/rollout_ppl:np.float64(1.1336332079954445) - rollout_corr/rollout_log_ppl:np.float64(0.12389674458245281) - rollout_corr/log_ppl_diff:np.float64(0.0005990994177409448) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019938167752115987) - rollout_corr/log_ppl_diff_max:np.float64(0.00943070650100708) - rollout_corr/log_ppl_diff_min:np.float64(-0.00719742476940155) - rollout_corr/ppl_ratio:np.float64(1.0006026013288647) - rollout_corr/chi2_token:np.float64(0.000771090155467391) - rollout_corr/chi2_seq:np.float64(0.09980663354508579) - actor/pg_loss:np.float64(0.02603117700709845) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.34765625) - actor/grad_norm:np.float64(0.2760481536388397) - perf/mfu/actor:np.float64(0.02431310694798927) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.16433334350586) - actor/lr:np.float64(1e-05) - training/global_step:18 - training/epoch:0 - critic/score/mean:0.515625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.515625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.007624865975230932 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.007624865975230932 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:149.14453125 - response_length/max:447.0 - response_length/min:91.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:149.14453125 - response_length_non_aborted/max:447.0 - response_length_non_aborted/min:91.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:273.28125 - prompt_length/max:540.0 - prompt_length/min:170.0 - 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:0.00011593848466873169 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1219863444566727) - timing_s/agent_loop/generate_sequences/max:np.float64(3.275597967207432) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.7206968119644444) - timing_s/agent_loop/tool_calls/m
|
||
|
|
step:19 - global_seqlen/min:11305 - global_seqlen/max:18037 - global_seqlen/minmax_diff:6732 - global_seqlen/balanced_min:14431 - global_seqlen/balanced_max:14433 - global_seqlen/mean:14432.25 - actor/entropy:0.11311642080545425 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2954539358615875 - training/rollout_probs_diff_mean:0.0028050660621374846 - training/rollout_probs_diff_std:0.010392368771135807 - training/rollout_actor_probs_pearson_corr:0.9981169104576111 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.68359375 - self_distillation/correct_sample_fraction:0.54296875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.68359375 - rollout_corr/rollout_is_mean:1.0001578330993652 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.9132332801818848 - rollout_corr/rollout_is_min:0.5190552473068237 - rollout_corr/rollout_is_std:0.029610544443130493 - rollout_corr/rollout_is_eff_sample_size:0.9991243407543628 - rollout_corr/rollout_is_seq_mean:1.0002073049545288 - rollout_corr/rollout_is_seq_std:0.002169467508792877 - rollout_corr/rollout_is_seq_max:1.0070828199386597 - rollout_corr/rollout_is_seq_min:0.9940499663352966 - rollout_corr/rollout_is_seq_max_deviation:0.007082819938659668 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1249200101010501) - rollout_corr/training_log_ppl:np.float64(0.11611939397334936) - rollout_corr/kl:np.float64(0.0002200406781796005) - rollout_corr/k3_kl:np.float64(0.0004273890697863436) - rollout_corr/rollout_ppl:np.float64(1.1246625245548785) - rollout_corr/rollout_log_ppl:np.float64(0.11589935216397862) - rollout_corr/log_ppl_diff:np.float64(0.00022004180937074125) - rollout_corr/log_ppl_abs_diff:np.float64(0.001736158737912774) - rollout_corr/log_ppl_diff_max:np.float64(0.006521761417388916) - rollout_corr/log_ppl_diff_min:np.float64(-0.005666404962539673) - rollout_corr/ppl_ratio:np.float64(1.000222405884415) - rollout_corr/chi2_token:np.float64(0.001288639148697257) - rollout_corr/chi2_seq:np.float64(0.12795519595965743) - actor/pg_loss:np.float64(0.02657115027977852) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.31640625) - actor/grad_norm:np.float64(0.3237147331237793) - perf/mfu/actor:np.float64(0.026402980232713265) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.23960876464844) - actor/lr:np.float64(1e-05) - training/global_step:19 - 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:0.0017250878736376762 - critic/advantages/max:0.875 - critic/advantages/min:-0.625 - critic/returns/mean:0.0017250878736376762 - critic/returns/max:0.875 - critic/returns/min:-0.625 - response_length/mean:148.8828125 - response_length/max:458.0 - response_length/min:92.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:148.8828125 - response_length_non_aborted/max:458.0 - response_length_non_aborted/min:92.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:302.125 - prompt_length/max:836.0 - prompt_length/min:167.0 - 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:0.00010825507342815399 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1860189270228148) - timing_s/agent_loop/generate_sequences/max:np.float64(3.3845436926931143) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.762579758069478) - timing_s/agent_loop/tool_ca
|
||
|
|
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 20}
|
||
|
|
validation generation end
|
||
|
|
[prompt] system
|
||
|
|
|
||
|
|
Given a question and four options, please select the right answer. Respond in the following format:
|
||
|
|
<reasoning>
|
||
|
|
...
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
...
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
user
|
||
|
|
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||
|
|
|
||
|
|
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||
|
|
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||
|
|
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||
|
|
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||
|
|
Please reason step by step.
|
||
|
|
assistant
|
||
|
|
<think>
|
||
|
|
|
||
|
|
</think>
|
||
|
|
|
||
|
|
|
||
|
|
[response] <reasoning>
|
||
|
|
The product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" contains a nitrogen-containing group (Nc1ccn), a chlorine-substituted aromatic ring (Cc2ccc(Cl)cc2), and a trifluorinated group (C(F)(F)F). Option A contains a nitrogen-containing group, a chlorine-substituted aromatic ring, and a trifluorinated group, matching the product's structure. The other options do not include all the necessary functional groups or structural features of the product. Therefore, the correct answer is Option A.
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
[ground_truth] A
|
||
|
|
[score] 1.0
|
||
|
|
[acc] 1.0
|
||
|
|
[pred] A
|
||
|
|
[incorrect_format] 1
|
||
|
|
[feedback]
|
||
|
|
local_global_step_folder: /workspace/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-SDPO_TR-Qwen-Qwen3-4B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_20
|
||
|
|
step:20 - global_seqlen/min:13351 - global_seqlen/max:14693 - global_seqlen/minmax_diff:1342 - global_seqlen/balanced_min:14087 - global_seqlen/balanced_max:14088 - global_seqlen/mean:14087.125 - actor/entropy:0.10291962325572968 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2411690056324005 - training/rollout_probs_diff_mean:0.002671134192496538 - training/rollout_probs_diff_std:0.010336983948946 - training/rollout_actor_probs_pearson_corr:0.9979697465896606 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.68359375 - self_distillation/correct_sample_fraction:0.609375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.68359375 - rollout_corr/rollout_is_mean:1.0000052452087402 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:4.987904321751557e-05 - rollout_corr/rollout_is_max:1.768110752105713 - rollout_corr/rollout_is_min:0.44401484727859497 - rollout_corr/rollout_is_std:0.030318688601255417 - rollout_corr/rollout_is_eff_sample_size:0.9990816213628854 - rollout_corr/rollout_is_seq_mean:1.000036358833313 - rollout_corr/rollout_is_seq_std:0.002488189609721303 - rollout_corr/rollout_is_seq_max:1.0084283351898193 - rollout_corr/rollout_is_seq_min:0.990229606628418 - rollout_corr/rollout_is_seq_max_deviation:0.009770393371582031 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1156582874245942) - rollout_corr/training_log_ppl:np.float64(0.10805689961853204) - rollout_corr/kl:np.float64(0.0004265682127737591) - rollout_corr/k3_kl:np.float64(0.0004629471573593946) - rollout_corr/rollout_ppl:np.float64(1.1151698753237724) - rollout_corr/rollout_log_ppl:np.float64(0.10763033049443038) - rollout_corr/log_ppl_diff:np.float64(0.00042656912410166115) - rollout_corr/log_ppl_abs_diff:np.float64(0.0019028567912755534) - rollout_corr/log_ppl_diff_max:np.float64(0.012332558631896973) - rollout_corr/log_ppl_diff_min:np.float64(-0.007378920912742615) - rollout_corr/ppl_ratio:np.float64(1.0004298442509025) - rollout_corr/chi2_token:np.float64(0.001001996686682105) - rollout_corr/chi2_seq:np.float64(0.13334895949810743) - actor/pg_loss:np.float64(0.022642171283223433) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.31640625) - actor/grad_norm:np.float64(0.25195199251174927) - perf/mfu/actor:np.float64(0.02569860407457783) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.36027908325195) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.669345238095238) - val-aux/sciknoweval/reward/std@16:np.float64(0.07007327594457972) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.6965619047619047) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.0592021949443194) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6420809523809523) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.059596460256278135) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.6692476190476189) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.06996767258090676) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7187714285714285) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.04567011702994718) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6198142857142858) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.048007989430369456) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.6710095238095237) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.05258048053057565) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7373476190476191) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.034267714925014334) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.5991285714285713) - val-aux
|
||
|
|
step:21 - global_seqlen/min:12465 - global_seqlen/max:15755 - global_seqlen/minmax_diff:3290 - global_seqlen/balanced_min:13890 - global_seqlen/balanced_max:13892 - global_seqlen/mean:13891.0 - actor/entropy:0.12715807557106018 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3997354507446289 - training/rollout_probs_diff_mean:0.0031383021268993616 - training/rollout_probs_diff_std:0.011023439466953278 - training/rollout_actor_probs_pearson_corr:0.9980828166007996 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65234375 - self_distillation/correct_sample_fraction:0.5703125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.65234375 - rollout_corr/rollout_is_mean:0.99989253282547 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.00010354110418120399 - rollout_corr/rollout_is_max:1.7020074129104614 - rollout_corr/rollout_is_min:0.25794777274131775 - rollout_corr/rollout_is_std:0.03204479441046715 - rollout_corr/rollout_is_eff_sample_size:0.9989739466280761 - rollout_corr/rollout_is_seq_mean:0.9999291896820068 - rollout_corr/rollout_is_seq_std:0.002610393799841404 - rollout_corr/rollout_is_seq_max:1.008002758026123 - rollout_corr/rollout_is_seq_min:0.9939199090003967 - rollout_corr/rollout_is_seq_max_deviation:0.008002758026123047 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.137966176494956) - rollout_corr/training_log_ppl:np.float64(0.12763843349239323) - rollout_corr/kl:np.float64(0.0005952823944248564) - rollout_corr/k3_kl:np.float64(0.000524450517502828) - rollout_corr/rollout_ppl:np.float64(1.1372648905962706) - rollout_corr/rollout_log_ppl:np.float64(0.1270431500452105) - rollout_corr/log_ppl_diff:np.float64(0.0005952834471827373) - rollout_corr/log_ppl_abs_diff:np.float64(0.0021610268886433914) - rollout_corr/log_ppl_diff_max:np.float64(0.00770944356918335) - rollout_corr/log_ppl_diff_min:np.float64(-0.006601907312870026) - rollout_corr/ppl_ratio:np.float64(1.0005990418139845) - rollout_corr/chi2_token:np.float64(0.000879399012774229) - rollout_corr/chi2_seq:np.float64(0.12246706057339907) - actor/pg_loss:np.float64(0.026511556054174434) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.34765625) - actor/grad_norm:np.float64(0.24646398425102234) - perf/mfu/actor:np.float64(0.025638268640872645) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(147.7473030090332) - actor/lr:np.float64(1e-05) - training/global_step:21 - training/epoch:0 - critic/score/mean:0.5703125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5703125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.003326257923617959 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.003326257923617959 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:150.90625 - response_length/max:436.0 - response_length/min:89.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:150.90625 - response_length_non_aborted/max:436.0 - response_length_non_aborted/min:89.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:283.1875 - prompt_length/max:669.0 - prompt_length/min:169.0 - 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:0.00012489594519138336 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0946341436356306) - timing_s/agent_loop/generate_sequences/max:np.float64(3.3644345086067915) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.8024905884376494) - timing_s/agent_loo
|
||
|
|
step:22 - global_seqlen/min:11390 - global_seqlen/max:14766 - global_seqlen/minmax_diff:3376 - global_seqlen/balanced_min:13097 - global_seqlen/balanced_max:13098 - global_seqlen/mean:13097.5 - actor/entropy:0.12969043850898743 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.16502171754837036 - training/rollout_probs_diff_mean:0.00297151249833405 - training/rollout_probs_diff_std:0.01003173366189003 - training/rollout_actor_probs_pearson_corr:0.9984333515167236 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65625 - self_distillation/correct_sample_fraction:0.6015625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.65625 - rollout_corr/rollout_is_mean:1.0000358819961548 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6442673206329346 - rollout_corr/rollout_is_min:0.6136547923088074 - rollout_corr/rollout_is_std:0.02887885458767414 - rollout_corr/rollout_is_eff_sample_size:0.9991667067670176 - rollout_corr/rollout_is_seq_mean:0.9999986290931702 - rollout_corr/rollout_is_seq_std:0.0024282780941575766 - rollout_corr/rollout_is_seq_max:1.0079325437545776 - rollout_corr/rollout_is_seq_min:0.9940003752708435 - rollout_corr/rollout_is_seq_max_deviation:0.007932543754577637 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1420031026937068) - rollout_corr/training_log_ppl:np.float64(0.13079696985369083) - rollout_corr/kl:np.float64(0.00042907467788211306) - rollout_corr/k3_kl:np.float64(0.00042768119634040147) - rollout_corr/rollout_ppl:np.float64(1.1415011072531343) - rollout_corr/rollout_log_ppl:np.float64(0.13036789452598896) - rollout_corr/log_ppl_diff:np.float64(0.0004290753277018666) - rollout_corr/log_ppl_abs_diff:np.float64(0.001964787210454233) - rollout_corr/log_ppl_diff_max:np.float64(0.00667533278465271) - rollout_corr/log_ppl_diff_min:np.float64(-0.00683671236038208) - rollout_corr/ppl_ratio:np.float64(1.000432118307799) - rollout_corr/chi2_token:np.float64(0.0008504178840667009) - rollout_corr/chi2_seq:np.float64(0.1430925950407982) - actor/pg_loss:np.float64(0.024926225378294475) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.34375) - actor/grad_norm:np.float64(0.24053427577018738) - perf/mfu/actor:np.float64(0.024446148742741376) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(147.3327407836914) - actor/lr:np.float64(1e-05) - training/global_step:22 - training/epoch:0 - critic/score/mean:0.6015625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6015625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-6.352271884679794e-05 - critic/advantages/max:0.5 - critic/advantages/min:-0.875 - critic/returns/mean:-6.352271884679794e-05 - critic/returns/max:0.5 - critic/returns/min:-0.875 - response_length/mean:153.734375 - response_length/max:368.0 - response_length/min:99.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:153.734375 - response_length_non_aborted/max:368.0 - response_length_non_aborted/min:99.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:255.5625 - prompt_length/max:592.0 - prompt_length/min:168.0 - 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:0.00011480599641799927 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0853225961327553) - timing_s/agent_loop/generate_sequences/max:np.float64(3.028973763808608) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.7479954948430532) - timing_s/agent_loop/tool_calls/min:np.float6
|
||
|
|
step:23 - global_seqlen/min:11421 - global_seqlen/max:18267 - global_seqlen/minmax_diff:6846 - global_seqlen/balanced_min:14110 - global_seqlen/balanced_max:14111 - global_seqlen/mean:14110.375 - actor/entropy:0.1080777645111084 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2246091365814209 - training/rollout_probs_diff_mean:0.0027116136625409126 - training/rollout_probs_diff_std:0.010239354334771633 - training/rollout_actor_probs_pearson_corr:0.9981417059898376 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - self_distillation/correct_sample_fraction:0.70703125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8359375 - rollout_corr/rollout_is_mean:0.9999159574508667 - rollout_corr/rollout_is_ratio_fraction_high:2.3930888346512802e-05 - rollout_corr/rollout_is_ratio_fraction_low:4.7861776693025604e-05 - rollout_corr/rollout_is_max:2.1670210361480713 - rollout_corr/rollout_is_min:0.2894376814365387 - rollout_corr/rollout_is_std:0.02980516105890274 - rollout_corr/rollout_is_eff_sample_size:0.9991123218375855 - rollout_corr/rollout_is_seq_mean:0.999924898147583 - rollout_corr/rollout_is_seq_std:0.0024112204555422068 - rollout_corr/rollout_is_seq_max:1.008556842803955 - rollout_corr/rollout_is_seq_min:0.9931722283363342 - rollout_corr/rollout_is_seq_max_deviation:0.008556842803955078 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1239025029353797) - rollout_corr/training_log_ppl:np.float64(0.1146908018636168) - rollout_corr/kl:np.float64(0.0005215470958859214) - rollout_corr/k3_kl:np.float64(0.000446425497727887) - rollout_corr/rollout_ppl:np.float64(1.1232929304242134) - rollout_corr/rollout_log_ppl:np.float64(0.11416925430239644) - rollout_corr/log_ppl_diff:np.float64(0.0005215475612203591) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018498990757507272) - rollout_corr/log_ppl_diff_max:np.float64(0.010816395282745361) - rollout_corr/log_ppl_diff_min:np.float64(-0.005996249616146088) - rollout_corr/ppl_ratio:np.float64(1.000524623086676) - rollout_corr/chi2_token:np.float64(0.0007619133684784174) - rollout_corr/chi2_seq:np.float64(0.10674174036830664) - actor/pg_loss:np.float64(0.031138732349063503) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1640625) - actor/grad_norm:np.float64(0.2722848653793335) - perf/mfu/actor:np.float64(0.026003424276398353) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(146.20458221435547) - actor/lr:np.float64(1e-05) - 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:-0.0015794385690242052 - critic/advantages/max:0.875 - critic/advantages/min:-0.625 - critic/returns/mean:-0.0015794385690242052 - critic/returns/max:0.875 - critic/returns/min:-0.625 - response_length/mean:163.23046875 - response_length/max:442.0 - response_length/min:89.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:163.23046875 - response_length_non_aborted/max:442.0 - response_length_non_aborted/min:89.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:277.71875 - prompt_length/max:766.0 - prompt_length/min:172.0 - 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:0.00012568384408950806 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1158825401216745) - timing_s/agent_loop/generate_sequences/max:np.float64(3.3835852332413197) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.896834
|
||
|
|
step:24 - global_seqlen/min:12296 - global_seqlen/max:15857 - global_seqlen/minmax_diff:3561 - global_seqlen/balanced_min:13850 - global_seqlen/balanced_max:13852 - global_seqlen/mean:13851.25 - actor/entropy:0.11131434887647629 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2899988889694214 - training/rollout_probs_diff_mean:0.0026358836330473423 - training/rollout_probs_diff_std:0.010087487287819386 - training/rollout_actor_probs_pearson_corr:0.998238205909729 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.75 - self_distillation/correct_sample_fraction:0.67578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.75 - rollout_corr/rollout_is_mean:1.0000332593917847 - rollout_corr/rollout_is_ratio_fraction_high:5.123738446854986e-05 - rollout_corr/rollout_is_ratio_fraction_low:2.561869223427493e-05 - rollout_corr/rollout_is_max:2.222957134246826 - rollout_corr/rollout_is_min:0.4245928227901459 - rollout_corr/rollout_is_std:0.028796177357435226 - rollout_corr/rollout_is_eff_sample_size:0.9991717052414121 - rollout_corr/rollout_is_seq_mean:1.0000531673431396 - rollout_corr/rollout_is_seq_std:0.0024456840474158525 - rollout_corr/rollout_is_seq_max:1.0085785388946533 - rollout_corr/rollout_is_seq_min:0.992300808429718 - rollout_corr/rollout_is_seq_max_deviation:0.00857853889465332 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1233669430948794) - rollout_corr/training_log_ppl:np.float64(0.11432940145823522) - rollout_corr/kl:np.float64(0.0003661767964899809) - rollout_corr/k3_kl:np.float64(0.00041931639723902236) - rollout_corr/rollout_ppl:np.float64(1.1229447140358388) - rollout_corr/rollout_log_ppl:np.float64(0.11396322471046005) - rollout_corr/log_ppl_diff:np.float64(0.00036617674777517095) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018628619000082836) - rollout_corr/log_ppl_diff_max:np.float64(0.008764177560806274) - rollout_corr/log_ppl_diff_min:np.float64(-0.007174365222454071) - rollout_corr/ppl_ratio:np.float64(1.0003691541496664) - rollout_corr/chi2_token:np.float64(0.000972954323515296) - rollout_corr/chi2_seq:np.float64(0.13390384265221655) - actor/pg_loss:np.float64(0.024783174568256072) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.25) - actor/grad_norm:np.float64(0.2703450620174408) - perf/mfu/actor:np.float64(0.02549099285083658) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(146.08948135375977) - actor/lr:np.float64(1e-05) - training/global_step:24 - training/epoch:0 - 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:-0.002702771918848157 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.002702771918848157 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:152.4765625 - response_length/max:355.0 - response_length/min:83.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:152.4765625 - response_length_non_aborted/max:355.0 - response_length_non_aborted/min:83.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:280.375 - prompt_length/max:848.0 - prompt_length/min:171.0 - 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:0.00010621920228004456 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9996157940477133) - timing_s/agent_loop/generate_sequences/max:np.float64(2.9736707489937544) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.7879874280115473) - timing_s/age
|
||
|
|
step:25 - global_seqlen/min:12484 - global_seqlen/max:17641 - global_seqlen/minmax_diff:5157 - global_seqlen/balanced_min:15070 - global_seqlen/balanced_max:15072 - global_seqlen/mean:15071.25 - actor/entropy:0.11880117654800415 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1537410020828247 - training/rollout_probs_diff_mean:0.0024952711537480354 - training/rollout_probs_diff_std:0.00913560576736927 - training/rollout_actor_probs_pearson_corr:0.9986276030540466 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65234375 - self_distillation/correct_sample_fraction:0.5703125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.65234375 - rollout_corr/rollout_is_mean:1.0000261068344116 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5713075399398804 - rollout_corr/rollout_is_min:0.5040717720985413 - rollout_corr/rollout_is_std:0.026172056794166565 - rollout_corr/rollout_is_eff_sample_size:0.9993154922939871 - rollout_corr/rollout_is_seq_mean:1.000041127204895 - rollout_corr/rollout_is_seq_std:0.0020636694971472025 - rollout_corr/rollout_is_seq_max:1.0094728469848633 - rollout_corr/rollout_is_seq_min:0.9938651323318481 - rollout_corr/rollout_is_seq_max_deviation:0.009472846984863281 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1283396580256522) - rollout_corr/training_log_ppl:np.float64(0.11889193205570336) - rollout_corr/kl:np.float64(0.0003068465399636011) - rollout_corr/k3_kl:np.float64(0.0003479610317675963) - rollout_corr/rollout_ppl:np.float64(1.1279965857975185) - rollout_corr/rollout_log_ppl:np.float64(0.11858508509612875) - rollout_corr/log_ppl_diff:np.float64(0.0003068469595746137) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015656692776246928) - rollout_corr/log_ppl_diff_max:np.float64(0.0072449445724487305) - rollout_corr/log_ppl_diff_min:np.float64(-0.008152216672897339) - rollout_corr/ppl_ratio:np.float64(1.0003090088721365) - rollout_corr/chi2_token:np.float64(0.0007727986667305231) - rollout_corr/chi2_seq:np.float64(0.08230392076075077) - actor/pg_loss:np.float64(0.022842508642497705) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.34765625) - actor/grad_norm:np.float64(0.21252715587615967) - perf/mfu/actor:np.float64(0.02782771860808586) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.6954107284546) - actor/lr:np.float64(1e-05) - training/global_step:25 - training/epoch:0 - critic/score/mean:0.5703125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5703125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0047833556309342384 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0047833556309342384 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:168.2265625 - response_length/max:553.0 - response_length/min:91.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:168.2265625 - response_length_non_aborted/max:553.0 - response_length_non_aborted/min:91.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:302.75 - prompt_length/max:646.0 - prompt_length/min:168.0 - 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:0.00010803714394569397 - timing_s/agent_loop/generate_sequences/min:np.float64(0.6494101472198963) - timing_s/agent_loop/generate_sequences/max:np.float64(3.827692398801446) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.6794912924742675) - timing_s/agent_loop/tool_cal
|
||
|
|
step:26 - global_seqlen/min:12559 - global_seqlen/max:16221 - global_seqlen/minmax_diff:3662 - global_seqlen/balanced_min:14217 - global_seqlen/balanced_max:14220 - global_seqlen/mean:14218.25 - actor/entropy:0.11071880906820297 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2101230025291443 - training/rollout_probs_diff_mean:0.0024308613501489162 - training/rollout_probs_diff_std:0.00905881728976965 - training/rollout_actor_probs_pearson_corr:0.998513400554657 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71484375 - self_distillation/correct_sample_fraction:0.67578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.71484375 - rollout_corr/rollout_is_mean:1.0000576972961426 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.9612475633621216 - rollout_corr/rollout_is_min:0.5377657413482666 - rollout_corr/rollout_is_std:0.025781989097595215 - rollout_corr/rollout_is_eff_sample_size:0.9993358495996573 - rollout_corr/rollout_is_seq_mean:1.0001106262207031 - rollout_corr/rollout_is_seq_std:0.0023443789687007666 - rollout_corr/rollout_is_seq_max:1.0127758979797363 - rollout_corr/rollout_is_seq_min:0.9946098923683167 - rollout_corr/rollout_is_seq_max_deviation:0.012775897979736328 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.11772670596838) - rollout_corr/training_log_ppl:np.float64(0.10903704005431791) - rollout_corr/kl:np.float64(0.00022227377597694442) - rollout_corr/k3_kl:np.float64(0.0003328241111422159) - rollout_corr/rollout_ppl:np.float64(1.1174973263405263) - rollout_corr/rollout_log_ppl:np.float64(0.10881476599388407) - rollout_corr/log_ppl_diff:np.float64(0.0002222740604338469) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017394657206750708) - rollout_corr/log_ppl_diff_max:np.float64(0.006034031510353088) - rollout_corr/log_ppl_diff_min:np.float64(-0.010783441364765167) - rollout_corr/ppl_ratio:np.float64(1.000224873656407) - rollout_corr/chi2_token:np.float64(0.0008987907785922289) - rollout_corr/chi2_seq:np.float64(0.21288909576833248) - actor/pg_loss:np.float64(0.021539170916184958) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.28515625) - actor/grad_norm:np.float64(0.218305304646492) - perf/mfu/actor:np.float64(0.02595944657374498) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.77664184570312) - actor/lr:np.float64(1e-05) - training/global_step:26 - training/epoch:0 - 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:0.0013369513908401132 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.0013369513908401132 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:147.9140625 - response_length/max:401.0 - response_length/min:87.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:147.9140625 - response_length_non_aborted/max:401.0 - response_length_non_aborted/min:87.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:296.40625 - prompt_length/max:709.0 - prompt_length/min:170.0 - 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:0.0001330636441707611 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0405685920268297) - timing_s/agent_loop/generate_sequences/max:np.float64(3.0143740121275187) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.7115694091116893) - timing_s/agent_loop/tool_cal
|
||
|
|
step:27 - global_seqlen/min:11816 - global_seqlen/max:15880 - global_seqlen/minmax_diff:4064 - global_seqlen/balanced_min:13531 - global_seqlen/balanced_max:13533 - global_seqlen/mean:13532.0 - actor/entropy:0.12660694122314453 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1674748659133911 - training/rollout_probs_diff_mean:0.002562726614996791 - training/rollout_probs_diff_std:0.009022769518196583 - training/rollout_actor_probs_pearson_corr:0.9987052083015442 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80078125 - self_distillation/correct_sample_fraction:0.62109375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.80078125 - rollout_corr/rollout_is_mean:0.9999564290046692 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.593483328819275 - rollout_corr/rollout_is_min:0.6646667718887329 - rollout_corr/rollout_is_std:0.025559086352586746 - rollout_corr/rollout_is_eff_sample_size:0.9993470405186317 - rollout_corr/rollout_is_seq_mean:0.9999009370803833 - rollout_corr/rollout_is_seq_std:0.0021186915691941977 - rollout_corr/rollout_is_seq_max:1.005697250366211 - rollout_corr/rollout_is_seq_min:0.9928781390190125 - rollout_corr/rollout_is_seq_max_deviation:0.007121860980987549 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.134166210424155) - rollout_corr/training_log_ppl:np.float64(0.1229278872924624) - rollout_corr/kl:np.float64(0.0004327786916462628) - rollout_corr/k3_kl:np.float64(0.00033370639718377504) - rollout_corr/rollout_ppl:np.float64(1.133664627559483) - rollout_corr/rollout_log_ppl:np.float64(0.12249510837864364) - rollout_corr/log_ppl_diff:np.float64(0.00043277891381876543) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016269541301880963) - rollout_corr/log_ppl_diff_max:np.float64(0.007538795471191406) - rollout_corr/log_ppl_diff_min:np.float64(-0.005172327160835266) - rollout_corr/ppl_ratio:np.float64(1.0004351157695055) - rollout_corr/chi2_token:np.float64(0.00048030796460807323) - rollout_corr/chi2_seq:np.float64(0.0803564318921417) - actor/pg_loss:np.float64(0.028563630504322646) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.19921875) - actor/grad_norm:np.float64(0.2247341424226761) - perf/mfu/actor:np.float64(0.02477846242173555) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.68768692016602) - actor/lr:np.float64(1e-05) - training/global_step:27 - 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:0.00339339766651392 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.00339339766651392 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:165.1875 - response_length/max:429.0 - response_length/min:83.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:165.1875 - response_length_non_aborted/max:429.0 - response_length_non_aborted/min:83.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:257.6875 - prompt_length/max:380.0 - prompt_length/min:160.0 - 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:0.00015437975525856018 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0639190394431353) - timing_s/agent_loop/generate_sequences/max:np.float64(3.4738897550851107) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.8571415548649384) - timing_s/agent_loop/tool_calls/min:np.f
|
||
|
|
step:28 - global_seqlen/min:11848 - global_seqlen/max:14958 - global_seqlen/minmax_diff:3110 - global_seqlen/balanced_min:13677 - global_seqlen/balanced_max:13678 - global_seqlen/mean:13677.875 - actor/entropy:0.11580180376768112 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3394787311553955 - training/rollout_probs_diff_mean:0.0024618746247142553 - training/rollout_probs_diff_std:0.009092049673199654 - training/rollout_actor_probs_pearson_corr:0.9986160397529602 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - self_distillation/correct_sample_fraction:0.63671875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83203125 - rollout_corr/rollout_is_mean:0.9999212622642517 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5526795387268066 - rollout_corr/rollout_is_min:0.5632566213607788 - rollout_corr/rollout_is_std:0.025374187156558037 - rollout_corr/rollout_is_eff_sample_size:0.9993564458465024 - rollout_corr/rollout_is_seq_mean:0.9999768733978271 - rollout_corr/rollout_is_seq_std:0.002216876717284322 - rollout_corr/rollout_is_seq_max:1.0070844888687134 - rollout_corr/rollout_is_seq_min:0.9902510046958923 - rollout_corr/rollout_is_seq_max_deviation:0.009748995304107666 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1243466497398913) - rollout_corr/training_log_ppl:np.float64(0.11503954643922043) - rollout_corr/kl:np.float64(0.0003474501511453454) - rollout_corr/k3_kl:np.float64(0.00032437793313278007) - rollout_corr/rollout_ppl:np.float64(1.1239472315646708) - rollout_corr/rollout_log_ppl:np.float64(0.11469209423739812) - rollout_corr/log_ppl_diff:np.float64(0.00034745220182230696) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016906235905480571) - rollout_corr/log_ppl_diff_max:np.float64(0.010576978325843811) - rollout_corr/log_ppl_diff_min:np.float64(-0.005915537476539612) - rollout_corr/ppl_ratio:np.float64(1.000349999172613) - rollout_corr/chi2_token:np.float64(0.0006061324384063482) - rollout_corr/chi2_seq:np.float64(0.05213409336283803) - actor/pg_loss:np.float64(0.02973498885967274) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.16796875) - actor/grad_norm:np.float64(0.2632063925266266) - perf/mfu/actor:np.float64(0.025162563702191672) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.68184518814087) - actor/lr:np.float64(1e-05) - training/global_step:28 - 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:-0.0039811525493860245 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0039811525493860245 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:155.02734375 - response_length/max:513.0 - response_length/min:86.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:155.02734375 - response_length_non_aborted/max:513.0 - response_length_non_aborted/min:86.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:272.40625 - prompt_length/max:443.0 - prompt_length/min:162.0 - 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:8.878111839294434e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9449766818434) - timing_s/agent_loop/generate_sequences/max:np.float64(3.789089487865567) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.753637524045189) - timing_s/agent_loop/t
|
||
|
|
step:29 - global_seqlen/min:11293 - global_seqlen/max:17568 - global_seqlen/minmax_diff:6275 - global_seqlen/balanced_min:14050 - global_seqlen/balanced_max:14061 - global_seqlen/mean:14051.625 - actor/entropy:0.1358516812324524 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.15126264095306396 - training/rollout_probs_diff_mean:0.0026747887022793293 - training/rollout_probs_diff_std:0.00899449922144413 - training/rollout_actor_probs_pearson_corr:0.9988278150558472 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.63671875 - self_distillation/correct_sample_fraction:0.44921875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.63671875 - rollout_corr/rollout_is_mean:1.000122308731079 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.612297534942627 - rollout_corr/rollout_is_min:0.5690498352050781 - rollout_corr/rollout_is_std:0.026746494695544243 - rollout_corr/rollout_is_eff_sample_size:0.9992853745267646 - rollout_corr/rollout_is_seq_mean:1.0001204013824463 - rollout_corr/rollout_is_seq_std:0.002428042469546199 - rollout_corr/rollout_is_seq_max:1.0088955163955688 - rollout_corr/rollout_is_seq_min:0.9914945363998413 - rollout_corr/rollout_is_seq_max_deviation:0.008895516395568848 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1481166468001902) - rollout_corr/training_log_ppl:np.float64(0.13577204996181536) - rollout_corr/kl:np.float64(0.000252926797355002) - rollout_corr/k3_kl:np.float64(0.00037337523420255536) - rollout_corr/rollout_ppl:np.float64(1.1478161658160388) - rollout_corr/rollout_log_ppl:np.float64(0.13551912197908678) - rollout_corr/log_ppl_diff:np.float64(0.0002529279827285791) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018354955518589122) - rollout_corr/log_ppl_diff_max:np.float64(0.009352564811706543) - rollout_corr/log_ppl_diff_min:np.float64(-0.007123738527297974) - rollout_corr/ppl_ratio:np.float64(1.0002559435088187) - rollout_corr/chi2_token:np.float64(0.000995169160887599) - rollout_corr/chi2_seq:np.float64(0.13358463742770255) - actor/pg_loss:np.float64(0.02372536157781724) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.36328125) - actor/grad_norm:np.float64(0.2024281769990921) - perf/mfu/actor:np.float64(0.02607972880057848) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.61566925048828) - actor/lr:np.float64(1e-05) - training/global_step:29 - training/epoch:0 - critic/score/mean:0.44921875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.44921875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.01037230808287859 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.01037230808287859 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:162.36328125 - response_length/max:627.0 - response_length/min:73.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:162.36328125 - response_length_non_aborted/max:627.0 - response_length_non_aborted/min:73.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:276.75 - prompt_length/max:492.0 - prompt_length/min:171.0 - 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:0.00012026727199554443 - timing_s/agent_loop/generate_sequences/min:np.float64(0.850072855129838) - timing_s/agent_loop/generate_sequences/max:np.float64(4.276421539485455) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.7775436455267482) - timing_s/agent_loop/tool_calls
|
||
|
|
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 30}
|
||
|
|
validation generation end
|
||
|
|
[prompt] system
|
||
|
|
|
||
|
|
Given a question and four options, please select the right answer. Respond in the following format:
|
||
|
|
<reasoning>
|
||
|
|
...
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
...
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
user
|
||
|
|
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||
|
|
|
||
|
|
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||
|
|
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||
|
|
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||
|
|
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||
|
|
Please reason step by step.
|
||
|
|
assistant
|
||
|
|
<think>
|
||
|
|
|
||
|
|
</think>
|
||
|
|
|
||
|
|
|
||
|
|
[response] <reasoning>
|
||
|
|
The product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" contains a chlorine-substituted benzene ring (Cc2ccc(Cl)cc2), a trifluorinated group (C(F)(F)F), and a nitrogen-containing group (Nc1ccn). Option A contains the same functional groups as the product and matches the product's structure exactly. The other options either have different functional groups or incorrect arrangements. Therefore, the correct answer is Option A.
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
[ground_truth] A
|
||
|
|
[score] 1.0
|
||
|
|
[acc] 1.0
|
||
|
|
[pred] A
|
||
|
|
[incorrect_format] 1
|
||
|
|
[feedback]
|
||
|
|
local_global_step_folder: /workspace/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-SDPO_TR-Qwen-Qwen3-4B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_30
|
||
|
|
step:30 - global_seqlen/min:11733 - global_seqlen/max:17677 - global_seqlen/minmax_diff:5944 - global_seqlen/balanced_min:14388 - global_seqlen/balanced_max:14390 - global_seqlen/mean:14389.5 - actor/entropy:0.10702905058860779 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.14319461584091187 - training/rollout_probs_diff_mean:0.0022296635434031487 - training/rollout_probs_diff_std:0.008453376591205597 - training/rollout_actor_probs_pearson_corr:0.9986989498138428 - self_distillation/success_group_fraction:0.5625 - self_distillation/success_sample_fraction:0.5625 - self_distillation/correct_sample_fraction:0.5 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.5625 - rollout_corr/rollout_is_mean:0.9999619722366333 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.7817959562526084e-05 - rollout_corr/rollout_is_max:1.5303750038146973 - rollout_corr/rollout_is_min:0.4321787357330322 - rollout_corr/rollout_is_std:0.0234425850212574 - rollout_corr/rollout_is_eff_sample_size:0.999450627941437 - rollout_corr/rollout_is_seq_mean:0.9999107122421265 - rollout_corr/rollout_is_seq_std:0.002046900102868676 - rollout_corr/rollout_is_seq_max:1.0060863494873047 - rollout_corr/rollout_is_seq_min:0.9930869340896606 - rollout_corr/rollout_is_seq_max_deviation:0.0069130659103393555 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.1123732789419591) - rollout_corr/training_log_ppl:np.float64(0.10452871808956843) - rollout_corr/kl:np.float64(0.00037610777075336443) - rollout_corr/k3_kl:np.float64(0.0002867263211012716) - rollout_corr/rollout_ppl:np.float64(1.1119311721995473) - rollout_corr/rollout_log_ppl:np.float64(0.10415261046364321) - rollout_corr/log_ppl_diff:np.float64(0.00037610762592521496) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016263087054539938) - rollout_corr/log_ppl_diff_max:np.float64(0.007853642106056213) - rollout_corr/log_ppl_diff_min:np.float64(-0.005079112946987152) - rollout_corr/ppl_ratio:np.float64(1.0003784170839936) - rollout_corr/chi2_token:np.float64(0.0003811225760728121) - rollout_corr/chi2_seq:np.float64(0.06720940652303398) - actor/pg_loss:np.float64(0.018335329927140265) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.4375) - actor/grad_norm:np.float64(0.19319124519824982) - perf/mfu/actor:np.float64(0.026866365728809155) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.73289489746094) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.6898809523809524) - val-aux/sciknoweval/reward/std@16:np.float64(0.06221026595170445) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7151142857142857) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.05824940288412261) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6656666666666666) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.046875006990806246) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.6906380952380952) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.062325289853640845) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7387809523809524) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.0496842967650804) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6487714285714287) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.030945064629849427) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.6885523809523809) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.04925178732123632) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7596333333333333) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.03738632084597657) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6377238095238095) - val-aux/sciknowe
|
||
|
|
step:31 - global_seqlen/min:11479 - global_seqlen/max:16925 - global_seqlen/minmax_diff:5446 - global_seqlen/balanced_min:14289 - global_seqlen/balanced_max:14291 - global_seqlen/mean:14290.0 - actor/entropy:0.0960652306675911 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3103688955307007 - training/rollout_probs_diff_mean:0.002089801477268338 - training/rollout_probs_diff_std:0.008831391111016273 - training/rollout_actor_probs_pearson_corr:0.9984229207038879 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.7421875 - self_distillation/correct_sample_fraction:0.62109375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7421875 - rollout_corr/rollout_is_mean:0.9999995231628418 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0001171508920378983 - rollout_corr/rollout_is_max:1.8267484903335571 - rollout_corr/rollout_is_min:0.4158477485179901 - rollout_corr/rollout_is_std:0.024116843938827515 - rollout_corr/rollout_is_eff_sample_size:0.9994188350346924 - rollout_corr/rollout_is_seq_mean:1.0000255107879639 - rollout_corr/rollout_is_seq_std:0.0020597896073013544 - rollout_corr/rollout_is_seq_max:1.007808804512024 - rollout_corr/rollout_is_seq_min:0.9946039319038391 - rollout_corr/rollout_is_seq_max_deviation:0.007808804512023926 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0970559134148061) - rollout_corr/training_log_ppl:np.float64(0.09053667732223403) - rollout_corr/kl:np.float64(0.00026855524223989846) - rollout_corr/k3_kl:np.float64(0.0002940440506223885) - rollout_corr/rollout_ppl:np.float64(1.0967666669748724) - rollout_corr/rollout_log_ppl:np.float64(0.09026812213505764) - rollout_corr/log_ppl_diff:np.float64(0.00026855518717638915) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015296213668989367) - rollout_corr/log_ppl_diff_max:np.float64(0.006772458553314209) - rollout_corr/log_ppl_diff_min:np.float64(-0.006242543458938599) - rollout_corr/ppl_ratio:np.float64(1.0002707564271986) - rollout_corr/chi2_token:np.float64(0.0006230971775949001) - rollout_corr/chi2_seq:np.float64(0.12701442232355475) - actor/pg_loss:np.float64(0.01858664442806912) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.2578125) - actor/grad_norm:np.float64(0.18856680393218994) - perf/mfu/actor:np.float64(0.02619266701042223) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(147.01678085327148) - actor/lr:np.float64(1e-05) - training/global_step:31 - 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:0.006718603428453207 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.006718603428453207 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:166.71875 - response_length/max:506.0 - response_length/min:76.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:166.71875 - response_length_non_aborted/max:506.0 - response_length_non_aborted/min:76.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:279.84375 - prompt_length/max:635.0 - prompt_length/min:176.0 - 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:0.00012081488966941833 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9543891865760088) - timing_s/agent_loop/generate_sequences/max:np.float64(3.8332445546984673) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.8908400539803552) - timing_s/agent_l
|
||
|
|
step:32 - global_seqlen/min:10807 - global_seqlen/max:15269 - global_seqlen/minmax_diff:4462 - global_seqlen/balanced_min:12927 - global_seqlen/balanced_max:12929 - global_seqlen/mean:12927.75 - actor/entropy:0.10790649056434631 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1901467740535736 - training/rollout_probs_diff_mean:0.002296852646395564 - training/rollout_probs_diff_std:0.008958207443356514 - training/rollout_actor_probs_pearson_corr:0.9985485076904297 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.68359375 - self_distillation/correct_sample_fraction:0.6015625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.68359375 - rollout_corr/rollout_is_mean:0.999981701374054 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6220754384994507 - rollout_corr/rollout_is_min:0.6155729293823242 - rollout_corr/rollout_is_std:0.024825632572174072 - rollout_corr/rollout_is_eff_sample_size:0.999384067579581 - rollout_corr/rollout_is_seq_mean:0.9999926686286926 - rollout_corr/rollout_is_seq_std:0.002221285132691264 - rollout_corr/rollout_is_seq_max:1.0090365409851074 - rollout_corr/rollout_is_seq_min:0.9937421679496765 - rollout_corr/rollout_is_seq_max_deviation:0.009036540985107422 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.114644323475659) - rollout_corr/training_log_ppl:np.float64(0.10616074145946186) - rollout_corr/kl:np.float64(0.0003164141928273523) - rollout_corr/k3_kl:np.float64(0.0003090736178548159) - rollout_corr/rollout_ppl:np.float64(1.1142864963039756) - rollout_corr/rollout_log_ppl:np.float64(0.10584432619634754) - rollout_corr/log_ppl_diff:np.float64(0.00031641526311432244) - rollout_corr/log_ppl_abs_diff:np.float64(0.001670944516263262) - rollout_corr/log_ppl_diff_max:np.float64(0.006982266902923584) - rollout_corr/log_ppl_diff_min:np.float64(-0.008343696594238281) - rollout_corr/ppl_ratio:np.float64(1.0003189065027982) - rollout_corr/chi2_token:np.float64(0.0006077890284359455) - rollout_corr/chi2_seq:np.float64(0.10511020314879715) - actor/pg_loss:np.float64(0.021079552939227142) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.31640625) - actor/grad_norm:np.float64(0.24713245034217834) - perf/mfu/actor:np.float64(0.02394769878606099) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(146.72430801391602) - actor/lr:np.float64(1e-05) - training/global_step:32 - training/epoch:0 - critic/score/mean:0.6015625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6015625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.01309016440063715 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.01309016440063715 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:141.1484375 - response_length/max:435.0 - response_length/min:65.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:141.1484375 - response_length_non_aborted/max:435.0 - response_length_non_aborted/min:65.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:262.84375 - prompt_length/max:404.0 - prompt_length/min:162.0 - 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:0.00011385418474674225 - timing_s/agent_loop/generate_sequences/min:np.float64(0.769508745521307) - timing_s/agent_loop/generate_sequences/max:np.float64(3.2758901808410883) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.6268715770493145) - timing_s/agent_loop/tool_calls/mi
|
||
|
|
step:33 - global_seqlen/min:11512 - global_seqlen/max:15756 - global_seqlen/minmax_diff:4244 - global_seqlen/balanced_min:13931 - global_seqlen/balanced_max:13989 - global_seqlen/mean:13939.0 - actor/entropy:0.08723315596580505 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24040323495864868 - training/rollout_probs_diff_mean:0.0019417706644162536 - training/rollout_probs_diff_std:0.008372211828827858 - training/rollout_actor_probs_pearson_corr:0.9984833598136902 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71484375 - self_distillation/correct_sample_fraction:0.67578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.71484375 - rollout_corr/rollout_is_mean:1.0000035762786865 - rollout_corr/rollout_is_ratio_fraction_high:2.7539104848983698e-05 - rollout_corr/rollout_is_ratio_fraction_low:8.26173200039193e-05 - rollout_corr/rollout_is_max:2.0409646034240723 - rollout_corr/rollout_is_min:0.1927628070116043 - rollout_corr/rollout_is_std:0.024789592251181602 - rollout_corr/rollout_is_eff_sample_size:0.9993858535200482 - rollout_corr/rollout_is_seq_mean:1.0000815391540527 - rollout_corr/rollout_is_seq_std:0.0022696962114423513 - rollout_corr/rollout_is_seq_max:1.008997917175293 - rollout_corr/rollout_is_seq_min:0.9935104250907898 - rollout_corr/rollout_is_seq_max_deviation:0.008997917175292969 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0901623209938407) - rollout_corr/training_log_ppl:np.float64(0.08444468390734983) - rollout_corr/kl:np.float64(0.00023199306088761418) - rollout_corr/k3_kl:np.float64(0.000313557187648578) - rollout_corr/rollout_ppl:np.float64(1.0899134599603713) - rollout_corr/rollout_log_ppl:np.float64(0.08421269063455838) - rollout_corr/log_ppl_diff:np.float64(0.000231993272791442) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016191630265893764) - rollout_corr/log_ppl_diff_max:np.float64(0.010110177099704742) - rollout_corr/log_ppl_diff_min:np.float64(-0.008274316787719727) - rollout_corr/ppl_ratio:np.float64(1.0002347407862544) - rollout_corr/chi2_token:np.float64(0.0007759747095406055) - rollout_corr/chi2_seq:np.float64(0.12760675442405045) - actor/pg_loss:np.float64(0.018758283783427032) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.28515625) - actor/grad_norm:np.float64(0.22452300786972046) - perf/mfu/actor:np.float64(0.02609049070098661) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.99493026733398) - actor/lr:np.float64(1e-05) - training/global_step:33 - training/epoch:0 - 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:-0.010964006185531616 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.010964006185531616 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:141.84375 - response_length/max:561.0 - response_length/min:64.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:141.84375 - response_length_non_aborted/max:561.0 - response_length_non_aborted/min:64.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:293.75 - prompt_length/max:474.0 - prompt_length/min:172.0 - 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:0.00015580281615257263 - timing_s/agent_loop/generate_sequences/min:np.float64(0.8279736675322056) - timing_s/agent_loop/generate_sequences/max:np.float64(3.753238506615162) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.6938069750613
|
||
|
|
step:34 - global_seqlen/min:13258 - global_seqlen/max:17390 - global_seqlen/minmax_diff:4132 - global_seqlen/balanced_min:14227 - global_seqlen/balanced_max:14229 - global_seqlen/mean:14228.625 - actor/entropy:0.09306639432907104 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.15485137701034546 - training/rollout_probs_diff_mean:0.0021454168017953634 - training/rollout_probs_diff_std:0.008605170994997025 - training/rollout_actor_probs_pearson_corr:0.9984543919563293 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80859375 - self_distillation/correct_sample_fraction:0.6640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.80859375 - rollout_corr/rollout_is_mean:0.9999364614486694 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5268398523330688 - rollout_corr/rollout_is_min:0.5899041891098022 - rollout_corr/rollout_is_std:0.023531407117843628 - rollout_corr/rollout_is_eff_sample_size:0.9994464602167408 - rollout_corr/rollout_is_seq_mean:0.9999260902404785 - rollout_corr/rollout_is_seq_std:0.0019278281833976507 - rollout_corr/rollout_is_seq_max:1.0062973499298096 - rollout_corr/rollout_is_seq_min:0.9934419989585876 - rollout_corr/rollout_is_seq_max_deviation:0.0065580010414123535 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0937934298999608) - rollout_corr/training_log_ppl:np.float64(0.0876305461163156) - rollout_corr/kl:np.float64(0.0003366993573479249) - rollout_corr/k3_kl:np.float64(0.00026279278095601294) - rollout_corr/rollout_ppl:np.float64(1.0934107997454703) - rollout_corr/rollout_log_ppl:np.float64(0.0872938462066486) - rollout_corr/log_ppl_diff:np.float64(0.00033669990966700425) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014094624555127666) - rollout_corr/log_ppl_diff_max:np.float64(0.0070962607860565186) - rollout_corr/log_ppl_diff_min:np.float64(-0.005653634667396545) - rollout_corr/ppl_ratio:np.float64(1.0003386216703802) - rollout_corr/chi2_token:np.float64(0.00038006436079740524) - rollout_corr/chi2_seq:np.float64(0.0699941769707948) - actor/pg_loss:np.float64(0.023825400812654607) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.19140625) - actor/grad_norm:np.float64(0.2171851098537445) - perf/mfu/actor:np.float64(0.02659338955433433) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.97358322143555) - actor/lr:np.float64(1e-05) - training/global_step:34 - training/epoch:0 - critic/score/mean:0.6640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0023529944010078907 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0023529944010078907 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:150.86328125 - response_length/max:561.0 - response_length/min:62.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:150.86328125 - response_length_non_aborted/max:561.0 - response_length_non_aborted/min:62.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:293.78125 - prompt_length/max:621.0 - prompt_length/min:170.0 - 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:0.0001813899725675583 - timing_s/agent_loop/generate_sequences/min:np.float64(0.8421060629189014) - timing_s/agent_loop/generate_sequences/max:np.float64(3.951502872630954) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.7443187848839443) - timing_s/agent_loo
|
||
|
|
step:35 - global_seqlen/min:12764 - global_seqlen/max:15213 - global_seqlen/minmax_diff:2449 - global_seqlen/balanced_min:13885 - global_seqlen/balanced_max:13887 - global_seqlen/mean:13886.0 - actor/entropy:0.08516640216112137 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2932351231575012 - training/rollout_probs_diff_mean:0.0019224673742428422 - training/rollout_probs_diff_std:0.009027084335684776 - training/rollout_actor_probs_pearson_corr:0.9982297420501709 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - self_distillation/correct_sample_fraction:0.6015625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83203125 - rollout_corr/rollout_is_mean:0.9999315142631531 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.7523740530014038 - rollout_corr/rollout_is_min:0.6194095015525818 - rollout_corr/rollout_is_std:0.02400536835193634 - rollout_corr/rollout_is_eff_sample_size:0.9994237169600801 - rollout_corr/rollout_is_seq_mean:0.9999163150787354 - rollout_corr/rollout_is_seq_std:0.0020165895111858845 - rollout_corr/rollout_is_seq_max:1.010898232460022 - rollout_corr/rollout_is_seq_min:0.9941428899765015 - rollout_corr/rollout_is_seq_max_deviation:0.010898232460021973 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.082754431758076) - rollout_corr/training_log_ppl:np.float64(0.0776018513965937) - rollout_corr/kl:np.float64(0.0003592223224668789) - rollout_corr/k3_kl:np.float64(0.0002755671693871211) - rollout_corr/rollout_ppl:np.float64(1.0823510992340744) - rollout_corr/rollout_log_ppl:np.float64(0.07724262839201401) - rollout_corr/log_ppl_diff:np.float64(0.0003592230045796896) - rollout_corr/log_ppl_abs_diff:np.float64(0.001318233206802688) - rollout_corr/log_ppl_diff_max:np.float64(0.007056728005409241) - rollout_corr/log_ppl_diff_min:np.float64(-0.008488357067108154) - rollout_corr/ppl_ratio:np.float64(1.0003613005392253) - rollout_corr/chi2_token:np.float64(0.00038353120908141136) - rollout_corr/chi2_seq:np.float64(0.07222988177090883) - actor/pg_loss:np.float64(0.021845255453797563) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.16796875) - actor/grad_norm:np.float64(0.21970264613628387) - perf/mfu/actor:np.float64(0.025891345216215084) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.9612693786621) - actor/lr:np.float64(1e-05) - training/global_step:35 - training/epoch:0 - critic/score/mean:0.6015625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6015625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0016584319528192282 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0016584319528192282 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:155.75 - response_length/max:452.0 - response_length/min:84.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:155.75 - response_length_non_aborted/max:452.0 - response_length_non_aborted/min:84.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:278.1875 - prompt_length/max:514.0 - prompt_length/min:181.0 - 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:0.0001240130513906479 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9371265154331923) - timing_s/agent_loop/generate_sequences/max:np.float64(3.1848833598196507) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.6985925326662255) - timing_s/agent_loop/tool_calls/min:np.f
|
||
|
|
step:36 - global_seqlen/min:10361 - global_seqlen/max:16127 - global_seqlen/minmax_diff:5766 - global_seqlen/balanced_min:13496 - global_seqlen/balanced_max:13498 - global_seqlen/mean:13497.25 - actor/entropy:0.0783422514796257 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23827695846557617 - training/rollout_probs_diff_mean:0.0018704815302044153 - training/rollout_probs_diff_std:0.008860025554895401 - training/rollout_actor_probs_pearson_corr:0.9981427192687988 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.73046875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8125 - rollout_corr/rollout_is_mean:0.9998724460601807 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:8.149959467118606e-05 - rollout_corr/rollout_is_max:1.741719365119934 - rollout_corr/rollout_is_min:0.3717479705810547 - rollout_corr/rollout_is_std:0.023533940315246582 - rollout_corr/rollout_is_eff_sample_size:0.9994461029848127 - rollout_corr/rollout_is_seq_mean:0.9998472332954407 - rollout_corr/rollout_is_seq_std:0.0019082835642620921 - rollout_corr/rollout_is_seq_max:1.0055783987045288 - rollout_corr/rollout_is_seq_min:0.9935135245323181 - rollout_corr/rollout_is_seq_max_deviation:0.006486475467681885 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.082965727429837) - rollout_corr/training_log_ppl:np.float64(0.0774456966391881) - rollout_corr/kl:np.float64(0.0004325241008764913) - rollout_corr/k3_kl:np.float64(0.0002797218442157027) - rollout_corr/rollout_ppl:np.float64(1.0824668062850833) - rollout_corr/rollout_log_ppl:np.float64(0.07701317118176121) - rollout_corr/log_ppl_diff:np.float64(0.0004325254574268911) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014095314179485285) - rollout_corr/log_ppl_diff_max:np.float64(0.009368028491735458) - rollout_corr/log_ppl_diff_min:np.float64(-0.005114108324050903) - rollout_corr/ppl_ratio:np.float64(1.0004347544163465) - rollout_corr/chi2_token:np.float64(0.00022734561935067177) - rollout_corr/chi2_seq:np.float64(0.06286093988455832) - actor/pg_loss:np.float64(0.020643032783567605) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.221962109208107) - perf/mfu/actor:np.float64(0.025198055405465096) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(146.00308990478516) - actor/lr:np.float64(1e-05) - training/global_step:36 - training/epoch:0 - critic/score/mean:0.73046875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.73046875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.003239608835428953 - critic/advantages/max:0.625 - critic/advantages/min:-0.875 - critic/returns/mean:0.003239608835428953 - critic/returns/max:0.625 - critic/returns/min:-0.875 - response_length/mean:143.7890625 - response_length/max:452.0 - response_length/min:76.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:143.7890625 - response_length_non_aborted/max:452.0 - response_length_non_aborted/min:76.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:278.0 - prompt_length/max:704.0 - prompt_length/min:165.0 - 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:0.00010615400969982147 - timing_s/agent_loop/generate_sequences/min:np.float64(0.980858201161027) - timing_s/agent_loop/generate_sequences/max:np.float64(3.2110455986112356) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.6668950048624538) - timing_s/agent_loop/tool
|
||
|
|
step:37 - global_seqlen/min:11720 - global_seqlen/max:17741 - global_seqlen/minmax_diff:6021 - global_seqlen/balanced_min:14414 - global_seqlen/balanced_max:14417 - global_seqlen/mean:14416.0 - actor/entropy:0.0887036845088005 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.15928596258163452 - training/rollout_probs_diff_mean:0.0019779072608798742 - training/rollout_probs_diff_std:0.008288256824016571 - training/rollout_actor_probs_pearson_corr:0.9985271692276001 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8046875 - self_distillation/correct_sample_fraction:0.609375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8046875 - rollout_corr/rollout_is_mean:1.0000797510147095 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6178338527679443 - rollout_corr/rollout_is_min:0.5753598809242249 - rollout_corr/rollout_is_std:0.023587066680192947 - rollout_corr/rollout_is_eff_sample_size:0.9994440786753774 - rollout_corr/rollout_is_seq_mean:1.0000554323196411 - rollout_corr/rollout_is_seq_std:0.0022047990933060646 - rollout_corr/rollout_is_seq_max:1.0106920003890991 - rollout_corr/rollout_is_seq_min:0.9927466511726379 - rollout_corr/rollout_is_seq_max_deviation:0.010692000389099121 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.096379661001265) - rollout_corr/training_log_ppl:np.float64(0.08951613759018073) - rollout_corr/kl:np.float64(0.00022537233156016967) - rollout_corr/k3_kl:np.float64(0.00028085313970013727) - rollout_corr/rollout_ppl:np.float64(1.096147044096142) - rollout_corr/rollout_log_ppl:np.float64(0.08929076531126157) - rollout_corr/log_ppl_diff:np.float64(0.000225372278919167) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015250365840415725) - rollout_corr/log_ppl_diff_max:np.float64(0.008192993700504303) - rollout_corr/log_ppl_diff_min:np.float64(-0.009256064891815186) - rollout_corr/ppl_ratio:np.float64(1.0002278012689203) - rollout_corr/chi2_token:np.float64(0.0006749788299202919) - rollout_corr/chi2_seq:np.float64(0.15035385033115745) - actor/pg_loss:np.float64(0.023521677601671342) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1953125) - actor/grad_norm:np.float64(0.24731527268886566) - perf/mfu/actor:np.float64(0.026825269182907126) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.96209335327148) - actor/lr:np.float64(1e-05) - training/global_step:37 - training/epoch:0 - critic/score/mean:0.609375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.609375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0007480996428057551 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0007480996428057551 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:137.71875 - response_length/max:408.0 - response_length/min:73.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:137.71875 - response_length_non_aborted/max:408.0 - response_length_non_aborted/min:73.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:312.78125 - prompt_length/max:1009.0 - prompt_length/min:171.0 - 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:9.809434413909912e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(0.7352254409343004) - timing_s/agent_loop/generate_sequences/max:np.float64(3.1295759975910187) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.508918766034185) - timing_s/agent_loop/tool_calls/min:
|
||
|
|
step:38 - global_seqlen/min:10214 - global_seqlen/max:15355 - global_seqlen/minmax_diff:5141 - global_seqlen/balanced_min:12844 - global_seqlen/balanced_max:12848 - global_seqlen/mean:12846.75 - actor/entropy:0.06084660068154335 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24124675989151 - training/rollout_probs_diff_mean:0.0014737353194504976 - training/rollout_probs_diff_std:0.007630929350852966 - training/rollout_actor_probs_pearson_corr:0.9982551336288452 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.74609375 - self_distillation/correct_sample_fraction:0.6484375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.74609375 - rollout_corr/rollout_is_mean:1.0000709295272827 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.82438013528008e-05 - rollout_corr/rollout_is_max:1.4431730508804321 - rollout_corr/rollout_is_min:0.3892325758934021 - rollout_corr/rollout_is_std:0.019956909120082855 - rollout_corr/rollout_is_eff_sample_size:0.9996021185548479 - rollout_corr/rollout_is_seq_mean:1.000087857246399 - rollout_corr/rollout_is_seq_std:0.0016890917904675007 - rollout_corr/rollout_is_seq_max:1.006268858909607 - rollout_corr/rollout_is_seq_min:0.9942615032196045 - rollout_corr/rollout_is_seq_max_deviation:0.006268858909606934 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0605181017890573) - rollout_corr/training_log_ppl:np.float64(0.05722638296333571) - rollout_corr/kl:np.float64(0.00011067048382562561) - rollout_corr/k3_kl:np.float64(0.00019850069808886135) - rollout_corr/rollout_ppl:np.float64(1.06039271177724) - rollout_corr/rollout_log_ppl:np.float64(0.05711571169854324) - rollout_corr/log_ppl_diff:np.float64(0.00011067126479247236) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011355591888104755) - rollout_corr/log_ppl_diff_max:np.float64(0.006475556641817093) - rollout_corr/log_ppl_diff_min:np.float64(-0.005778133869171143) - rollout_corr/ppl_ratio:np.float64(1.000112189212814) - rollout_corr/chi2_token:np.float64(0.0005644641350954771) - rollout_corr/chi2_seq:np.float64(0.08666495815850794) - actor/pg_loss:np.float64(0.014137650663599288) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.25390625) - actor/grad_norm:np.float64(0.18862751126289368) - perf/mfu/actor:np.float64(0.023716524806210452) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.87643814086914) - actor/lr:np.float64(1e-05) - training/global_step:38 - training/epoch:0 - critic/score/mean:0.6484375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6484375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0006390159833244979 - critic/advantages/max:0.875 - critic/advantages/min:-0.5 - critic/returns/mean:0.0006390159833244979 - critic/returns/max:0.875 - critic/returns/min:-0.5 - response_length/mean:138.3046875 - response_length/max:555.0 - response_length/min:78.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:138.3046875 - response_length_non_aborted/max:555.0 - response_length_non_aborted/min:78.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:263.15625 - prompt_length/max:500.0 - prompt_length/min:164.0 - 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:0.00011944584548473358 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9441804923117161) - timing_s/agent_loop/generate_sequences/max:np.float64(3.667453086003661) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.6324499895927147) - timing_s/agent_l
|
||
|
|
step:39 - global_seqlen/min:10600 - global_seqlen/max:15847 - global_seqlen/minmax_diff:5247 - global_seqlen/balanced_min:13805 - global_seqlen/balanced_max:13806 - global_seqlen/mean:13805.25 - actor/entropy:0.06885745376348495 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1470346450805664 - training/rollout_probs_diff_mean:0.0017357806209474802 - training/rollout_probs_diff_std:0.008160529658198357 - training/rollout_actor_probs_pearson_corr:0.9982522130012512 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.68359375 - self_distillation/correct_sample_fraction:0.61328125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.68359375 - rollout_corr/rollout_is_mean:0.9999474287033081 - rollout_corr/rollout_is_ratio_fraction_high:2.619309634610545e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.934377670288086 - rollout_corr/rollout_is_min:0.5325842499732971 - rollout_corr/rollout_is_std:0.02317664958536625 - rollout_corr/rollout_is_eff_sample_size:0.9994630122427728 - rollout_corr/rollout_is_seq_mean:0.9999102354049683 - rollout_corr/rollout_is_seq_std:0.0023426783736795187 - rollout_corr/rollout_is_seq_max:1.0122082233428955 - rollout_corr/rollout_is_seq_min:0.9911507964134216 - rollout_corr/rollout_is_seq_max_deviation:0.012208223342895508 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0735346740111709) - rollout_corr/training_log_ppl:np.float64(0.06930472629528595) - rollout_corr/kl:np.float64(0.000367558998746631) - rollout_corr/k3_kl:np.float64(0.0002777871578185931) - rollout_corr/rollout_ppl:np.float64(1.0731044695712626) - rollout_corr/rollout_log_ppl:np.float64(0.06893716670322192) - rollout_corr/log_ppl_diff:np.float64(0.0003675595920640262) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016067784932829454) - rollout_corr/log_ppl_diff_max:np.float64(0.010068729519844055) - rollout_corr/log_ppl_diff_min:np.float64(-0.008516862988471985) - rollout_corr/ppl_ratio:np.float64(1.0003703352995217) - rollout_corr/chi2_token:np.float64(0.000417166156694293) - rollout_corr/chi2_seq:np.float64(0.09390338184311986) - actor/pg_loss:np.float64(0.015166341426919416) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.31640625) - actor/grad_norm:np.float64(0.21120840311050415) - perf/mfu/actor:np.float64(0.025729783744862792) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.8499641418457) - actor/lr:np.float64(1e-05) - training/global_step:39 - training/epoch:0 - critic/score/mean:0.61328125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.61328125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0027731938753277063 - critic/advantages/max:0.875 - critic/advantages/min:-0.625 - critic/returns/mean:0.0027731938753277063 - critic/returns/max:0.875 - critic/returns/min:-0.625 - response_length/mean:149.1328125 - response_length/max:430.0 - response_length/min:78.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:149.1328125 - response_length_non_aborted/max:430.0 - response_length_non_aborted/min:78.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:282.28125 - prompt_length/max:452.0 - prompt_length/min:181.0 - 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:0.0001134239137172699 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9792734552174807) - timing_s/agent_loop/generate_sequences/max:np.float64(3.2234483510255814) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.720382460676774) - 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': 40}
|
||
|
|
validation generation end
|
||
|
|
[prompt] system
|
||
|
|
|
||
|
|
Given a question and four options, please select the right answer. Respond in the following format:
|
||
|
|
<reasoning>
|
||
|
|
...
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
...
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
user
|
||
|
|
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||
|
|
|
||
|
|
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||
|
|
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||
|
|
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||
|
|
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||
|
|
Please reason step by step.
|
||
|
|
assistant
|
||
|
|
<think>
|
||
|
|
|
||
|
|
</think>
|
||
|
|
|
||
|
|
|
||
|
|
[response] <reasoning>
|
||
|
|
The product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" contains a chlorine-substituted aromatic ring, a fluorine-substituted group (C(F)(F)F), and a nitrogen-containing group. Option A contains a carboxylic acid group (O=C1), a chlorine-substituted aromatic ring, and a nitrogen-containing group. The structure of Option A matches the product's structure, including the chlorine-substituted aromatic ring and the fluorine-substituted group. The other options either contain different functional groups or do not match the product's structure. Therefore, the correct answer is Option A.
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
[ground_truth] A
|
||
|
|
[score] 1.0
|
||
|
|
[acc] 1.0
|
||
|
|
[pred] A
|
||
|
|
[incorrect_format] 1
|
||
|
|
[feedback]
|
||
|
|
local_global_step_folder: /workspace/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-SDPO_TR-Qwen-Qwen3-4B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_40
|
||
|
|
step:40 - global_seqlen/min:11885 - global_seqlen/max:18697 - global_seqlen/minmax_diff:6812 - global_seqlen/balanced_min:14702 - global_seqlen/balanced_max:14704 - global_seqlen/mean:14703.375 - actor/entropy:0.07293666154146194 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.20892837643623352 - training/rollout_probs_diff_mean:0.0019123838283121586 - training/rollout_probs_diff_std:0.008589479140937328 - training/rollout_actor_probs_pearson_corr:0.9980955719947815 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8046875 - self_distillation/correct_sample_fraction:0.68359375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8046875 - rollout_corr/rollout_is_mean:1.000276803970337 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.9838459491729736 - rollout_corr/rollout_is_min:0.6108250617980957 - rollout_corr/rollout_is_std:0.02510496973991394 - rollout_corr/rollout_is_eff_sample_size:0.999370613699892 - rollout_corr/rollout_is_seq_mean:1.0002444982528687 - rollout_corr/rollout_is_seq_std:0.002133375033736229 - rollout_corr/rollout_is_seq_max:1.010922908782959 - rollout_corr/rollout_is_seq_min:0.9941123127937317 - rollout_corr/rollout_is_seq_max_deviation:0.010922908782958984 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.07762461155653) - rollout_corr/training_log_ppl:np.float64(0.07340155096517265) - rollout_corr/kl:np.float64(5.290125749812802e-05) - rollout_corr/k3_kl:np.float64(0.0002974506734462201) - rollout_corr/rollout_ppl:np.float64(1.0775611237622797) - rollout_corr/rollout_log_ppl:np.float64(0.07334864878237113) - rollout_corr/log_ppl_diff:np.float64(5.290218280151748e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014637900629850265) - rollout_corr/log_ppl_diff_max:np.float64(0.006884455680847168) - rollout_corr/log_ppl_diff_min:np.float64(-0.009787335991859436) - rollout_corr/ppl_ratio:np.float64(1.0000550681725144) - rollout_corr/chi2_token:np.float64(0.0011072962079197168) - rollout_corr/chi2_seq:np.float64(0.18554388917982578) - actor/pg_loss:np.float64(0.02182503892882437) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1953125) - actor/grad_norm:np.float64(0.23505358397960663) - perf/mfu/actor:np.float64(0.02733374828563766) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(145.83687591552734) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.6779761904761905) - val-aux/sciknoweval/reward/std@16:np.float64(0.07735763430637632) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7094714285714284) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.06543023206307134) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6466333333333334) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.06547524288610061) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.6784142857142856) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.07714018918744804) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7340714285714285) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.04892074381822419) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6199380952380953) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.04895557388087405) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.6808428571428572) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.06050268451530412) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.753047619047619) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.035136735215871046) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.5996095238095238) - val-aux/sciknoweval/reward/
|
||
|
|
step:41 - global_seqlen/min:10276 - global_seqlen/max:16700 - global_seqlen/minmax_diff:6424 - global_seqlen/balanced_min:13193 - global_seqlen/balanced_max:13195 - global_seqlen/mean:13193.875 - actor/entropy:0.07079192996025085 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24107179045677185 - training/rollout_probs_diff_mean:0.0018619152251631021 - training/rollout_probs_diff_std:0.008480723947286606 - training/rollout_actor_probs_pearson_corr:0.9981778264045715 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65234375 - self_distillation/correct_sample_fraction:0.5625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.65234375 - rollout_corr/rollout_is_mean:0.999925971031189 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.9824209213256836 - rollout_corr/rollout_is_min:0.570592999458313 - rollout_corr/rollout_is_std:0.024635229259729385 - rollout_corr/rollout_is_eff_sample_size:0.9993934736044263 - rollout_corr/rollout_is_seq_mean:0.9998980164527893 - rollout_corr/rollout_is_seq_std:0.0018514165421947837 - rollout_corr/rollout_is_seq_max:1.0056418180465698 - rollout_corr/rollout_is_seq_min:0.9939931035041809 - rollout_corr/rollout_is_seq_max_deviation:0.006006896495819092 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0739467809908092) - rollout_corr/training_log_ppl:np.float64(0.06925883538679045) - rollout_corr/kl:np.float64(0.0003803577491622856) - rollout_corr/k3_kl:np.float64(0.00027842260249305324) - rollout_corr/rollout_ppl:np.float64(1.0735123353078961) - rollout_corr/rollout_log_ppl:np.float64(0.06887847663040247) - rollout_corr/log_ppl_diff:np.float64(0.00038035875638797734) - rollout_corr/log_ppl_abs_diff:np.float64(0.001282316851302312) - rollout_corr/log_ppl_diff_max:np.float64(0.006815433502197266) - rollout_corr/log_ppl_diff_min:np.float64(-0.004278451204299927) - rollout_corr/ppl_ratio:np.float64(1.000382185447961) - rollout_corr/chi2_token:np.float64(0.0003656598273664713) - rollout_corr/chi2_seq:np.float64(0.042437196243554354) - actor/pg_loss:np.float64(0.014877505851963235) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.34765625) - actor/grad_norm:np.float64(0.19618210196495056) - perf/mfu/actor:np.float64(0.024389583963837343) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(147.83795928955078) - actor/lr:np.float64(1e-05) - training/global_step:41 - training/epoch:0 - critic/score/mean:0.5625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00498144282028079 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00498144282028079 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:148.40234375 - response_length/max:460.0 - response_length/min:96.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:148.40234375 - response_length_non_aborted/max:460.0 - response_length_non_aborted/min:96.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:263.90625 - prompt_length/max:631.0 - prompt_length/min:163.0 - 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:0.00017390400171279907 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1351077985018492) - timing_s/agent_loop/generate_sequences/max:np.float64(3.4039971847087145) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.686211900319904) - timing_s/agent_loop/tool_calls/m
|
||
|
|
step:42 - global_seqlen/min:10429 - global_seqlen/max:16246 - global_seqlen/minmax_diff:5817 - global_seqlen/balanced_min:13529 - global_seqlen/balanced_max:13530 - global_seqlen/mean:13529.75 - actor/entropy:0.06523686647415161 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18463215231895447 - training/rollout_probs_diff_mean:0.0015808126190677285 - training/rollout_probs_diff_std:0.007686233147978783 - training/rollout_actor_probs_pearson_corr:0.998401403427124 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.77734375 - self_distillation/correct_sample_fraction:0.6875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.77734375 - rollout_corr/rollout_is_mean:0.9999479651451111 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4217114448547363 - rollout_corr/rollout_is_min:0.6108585596084595 - rollout_corr/rollout_is_std:0.020862242206931114 - rollout_corr/rollout_is_eff_sample_size:0.9995648371237307 - rollout_corr/rollout_is_seq_mean:0.9999287128448486 - rollout_corr/rollout_is_seq_std:0.0016140977386385202 - rollout_corr/rollout_is_seq_max:1.005111813545227 - rollout_corr/rollout_is_seq_min:0.9936686158180237 - rollout_corr/rollout_is_seq_max_deviation:0.006331384181976318 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0628207307308912) - rollout_corr/training_log_ppl:np.float64(0.05942355485370854) - rollout_corr/kl:np.float64(0.00028260730116114186) - rollout_corr/k3_kl:np.float64(0.00021138309256457655) - rollout_corr/rollout_ppl:np.float64(1.0625111223198473) - rollout_corr/rollout_log_ppl:np.float64(0.059140947250625686) - rollout_corr/log_ppl_diff:np.float64(0.0002826076030828517) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010934863099123504) - rollout_corr/log_ppl_diff_max:np.float64(0.007111489772796631) - rollout_corr/log_ppl_diff_min:np.float64(-0.003931105136871338) - rollout_corr/ppl_ratio:np.float64(1.0002840005327016) - rollout_corr/chi2_token:np.float64(0.0002743923105299473) - rollout_corr/chi2_seq:np.float64(0.03434801124967635) - actor/pg_loss:np.float64(0.015241430047694848) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.22265625) - actor/grad_norm:np.float64(0.20213060081005096) - perf/mfu/actor:np.float64(0.025279196952231256) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(147.57822036743164) - actor/lr:np.float64(1e-05) - training/global_step:42 - training/epoch:0 - critic/score/mean:0.6875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0070900521241128445 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0070900521241128445 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:166.1796875 - response_length/max:522.0 - response_length/min:91.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:166.1796875 - response_length_non_aborted/max:522.0 - response_length_non_aborted/min:91.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:256.625 - prompt_length/max:557.0 - prompt_length/min:175.0 - 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:0.00019638985395431519 - timing_s/agent_loop/generate_sequences/min:np.float64(1.160779232159257) - timing_s/agent_loop/generate_sequences/max:np.float64(3.826703542843461) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.9309289849115885) - timing_s/agent_loop/tool_calls/m
|
||
|
|
step:43 - global_seqlen/min:12618 - global_seqlen/max:18207 - global_seqlen/minmax_diff:5589 - global_seqlen/balanced_min:14763 - global_seqlen/balanced_max:14881 - global_seqlen/mean:14788.75 - actor/entropy:0.06332086026668549 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.5082992315292358 - training/rollout_probs_diff_mean:0.0017634698888286948 - training/rollout_probs_diff_std:0.008931181393563747 - training/rollout_actor_probs_pearson_corr:0.9977215528488159 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.75 - self_distillation/correct_sample_fraction:0.71484375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.75 - rollout_corr/rollout_is_mean:0.9997488856315613 - rollout_corr/rollout_is_ratio_fraction_high:2.4535060219932348e-05 - rollout_corr/rollout_is_ratio_fraction_low:4.9070120439864695e-05 - rollout_corr/rollout_is_max:2.0018973350524902 - rollout_corr/rollout_is_min:0.3460032343864441 - rollout_corr/rollout_is_std:0.023989222943782806 - rollout_corr/rollout_is_eff_sample_size:0.9994245504643032 - rollout_corr/rollout_is_seq_mean:0.999769926071167 - rollout_corr/rollout_is_seq_std:0.0019337874837219715 - rollout_corr/rollout_is_seq_max:1.0095938444137573 - rollout_corr/rollout_is_seq_min:0.9936483502388 - rollout_corr/rollout_is_seq_max_deviation:0.009593844413757324 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.061429635155946) - rollout_corr/training_log_ppl:np.float64(0.05873463886314312) - rollout_corr/kl:np.float64(0.000525905199655341) - rollout_corr/k3_kl:np.float64(0.00029583590612557013) - rollout_corr/rollout_ppl:np.float64(1.0608501550741494) - rollout_corr/rollout_log_ppl:np.float64(0.05820873323040132) - rollout_corr/log_ppl_diff:np.float64(0.0005259056327417966) - rollout_corr/log_ppl_abs_diff:np.float64(0.001470901117215817) - rollout_corr/log_ppl_diff_max:np.float64(0.007815755903720856) - rollout_corr/log_ppl_diff_min:np.float64(-0.007278263568878174) - rollout_corr/ppl_ratio:np.float64(1.0005279961042106) - rollout_corr/chi2_token:np.float64(0.00012022210285067558) - rollout_corr/chi2_seq:np.float64(0.01147706457413733) - actor/pg_loss:np.float64(0.015978831296621365) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.25) - actor/grad_norm:np.float64(0.20858143270015717) - perf/mfu/actor:np.float64(0.027460105038100546) - perf/max_memory_allocated_gb:np.float64(118.25946617126465) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(146.92567443847656) - actor/lr:np.float64(1e-05) - training/global_step:43 - training/epoch:0 - 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:-0.002441238611936569 - critic/advantages/max:0.5 - critic/advantages/min:-0.875 - critic/returns/mean:-0.002441238611936569 - critic/returns/max:0.5 - critic/returns/min:-0.875 - response_length/mean:159.2109375 - response_length/max:757.0 - response_length/min:88.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:159.2109375 - response_length_non_aborted/max:757.0 - response_length_non_aborted/min:88.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:302.9375 - prompt_length/max:813.0 - prompt_length/min:171.0 - 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:5.4856762290000916e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(0.8350387774407864) - timing_s/agent_loop/generate_sequences/max:np.float64(4.893745483830571) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.6857070837795618) - timing_s/age
|
||
|
|
step:44 - global_seqlen/min:12792 - global_seqlen/max:17928 - global_seqlen/minmax_diff:5136 - global_seqlen/balanced_min:15511 - global_seqlen/balanced_max:15952 - global_seqlen/mean:15566.75 - actor/entropy:0.06333281099796295 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4010775685310364 - training/rollout_probs_diff_mean:0.0017267776420339942 - training/rollout_probs_diff_std:0.008502635173499584 - training/rollout_actor_probs_pearson_corr:0.9979615211486816 - self_distillation/success_group_fraction:0.6875 - self_distillation/success_sample_fraction:0.6875 - self_distillation/correct_sample_fraction:0.6171875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.6875 - rollout_corr/rollout_is_mean:1.0000686645507812 - rollout_corr/rollout_is_ratio_fraction_high:2.233439045085106e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.356858253479004 - rollout_corr/rollout_is_min:0.5079668760299683 - rollout_corr/rollout_is_std:0.02274049073457718 - rollout_corr/rollout_is_eff_sample_size:0.9994831373870391 - rollout_corr/rollout_is_seq_mean:1.0001215934753418 - rollout_corr/rollout_is_seq_std:0.0017713712295517325 - rollout_corr/rollout_is_seq_max:1.0070902109146118 - rollout_corr/rollout_is_seq_min:0.9914056062698364 - rollout_corr/rollout_is_seq_max_deviation:0.008594393730163574 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0604945085942745) - rollout_corr/training_log_ppl:np.float64(0.05760557667227317) - rollout_corr/kl:np.float64(0.00011546219154152937) - rollout_corr/k3_kl:np.float64(0.00023692107658696666) - rollout_corr/rollout_ppl:np.float64(1.0603752294555306) - rollout_corr/rollout_log_ppl:np.float64(0.057490114099195466) - rollout_corr/log_ppl_diff:np.float64(0.00011546257307770702) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011617422578780179) - rollout_corr/log_ppl_diff_max:np.float64(0.009427748620510101) - rollout_corr/log_ppl_diff_min:np.float64(-0.006299793720245361) - rollout_corr/ppl_ratio:np.float64(1.0001169571187347) - rollout_corr/chi2_token:np.float64(0.0007438324391841888) - rollout_corr/chi2_seq:np.float64(0.09388145827688277) - actor/pg_loss:np.float64(0.013146525828467759) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.3125) - actor/grad_norm:np.float64(0.18757085502147675) - perf/mfu/actor:np.float64(0.028836037013096352) - perf/max_memory_allocated_gb:np.float64(118.99612092971802) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(146.75523376464844) - actor/lr:np.float64(1e-05) - training/global_step:44 - 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:-0.015759704634547234 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.015759704634547234 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:174.8984375 - response_length/max:1302.0 - response_length/min:98.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:174.8984375 - response_length_non_aborted/max:1302.0 - response_length_non_aborted/min:98.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:311.5625 - prompt_length/max:701.0 - prompt_length/min:176.0 - 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:0.00013153068721294403 - timing_s/agent_loop/generate_sequences/min:np.float64(0.8961184844374657) - timing_s/agent_loop/generate_sequences/max:np.float64(7.549025377258658) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.8505728371674195) - timing_s/agent_l
|
||
|
|
step:45 - global_seqlen/min:14454 - global_seqlen/max:21474 - global_seqlen/minmax_diff:7020 - global_seqlen/balanced_min:16456 - global_seqlen/balanced_max:16462 - global_seqlen/mean:16459.375 - actor/entropy:0.08358263969421387 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.9014339447021484 - training/rollout_probs_diff_mean:0.0020398148335516453 - training/rollout_probs_diff_std:0.00984979048371315 - training/rollout_actor_probs_pearson_corr:0.9979046583175659 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65234375 - self_distillation/correct_sample_fraction:0.5234375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.65234375 - rollout_corr/rollout_is_mean:0.9997007250785828 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.9563729438232258e-05 - rollout_corr/rollout_is_max:1.8632231950759888 - rollout_corr/rollout_is_min:0.09565600752830505 - rollout_corr/rollout_is_std:0.024711325764656067 - rollout_corr/rollout_is_eff_sample_size:0.9993894254201319 - rollout_corr/rollout_is_seq_mean:0.999687135219574 - rollout_corr/rollout_is_seq_std:0.0021446668542921543 - rollout_corr/rollout_is_seq_max:1.006246566772461 - rollout_corr/rollout_is_seq_min:0.9929413199424744 - rollout_corr/rollout_is_seq_max_deviation:0.007058680057525635 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.082862624898553) - rollout_corr/training_log_ppl:np.float64(0.07850344575456347) - rollout_corr/kl:np.float64(0.0006431741838426319) - rollout_corr/k3_kl:np.float64(0.00033030845313335513) - rollout_corr/rollout_ppl:np.float64(1.082145718857646) - rollout_corr/rollout_log_ppl:np.float64(0.07786027109790439) - rollout_corr/log_ppl_diff:np.float64(0.0006431746566590846) - rollout_corr/log_ppl_abs_diff:np.float64(0.001659819078952296) - rollout_corr/log_ppl_diff_max:np.float64(0.009067557752132416) - rollout_corr/log_ppl_diff_min:np.float64(-0.0051116943359375) - rollout_corr/ppl_ratio:np.float64(1.0006458736024797) - rollout_corr/chi2_token:np.float64(-1.3278331607580185e-06) - rollout_corr/chi2_seq:np.float64(0.04086496401578188) - actor/pg_loss:np.float64(0.017752698722702576) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.34765625) - actor/grad_norm:np.float64(0.21230396628379822) - perf/mfu/actor:np.float64(0.030655056665718206) - perf/max_memory_allocated_gb:np.float64(118.99612092971802) - perf/max_memory_reserved_gb:np.float64(126.7265625) - perf/cpu_memory_used_gb:np.float64(146.51731872558594) - actor/lr:np.float64(1e-05) - training/global_step:45 - training/epoch:0 - critic/score/mean:0.5234375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5234375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.010764941573143005 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.010764941573143005 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:199.66796875 - response_length/max:880.0 - response_length/min:77.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:199.66796875 - response_length_non_aborted/max:880.0 - response_length_non_aborted/min:77.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:314.6875 - prompt_length/max:845.0 - prompt_length/min:170.0 - 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:0.00010529346764087677 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9852230623364449) - timing_s/agent_loop/generate_sequences/max:np.float64(5.978739287704229) - timing_s/agent_loop/generate_sequences/mean:np.float64(2.170499387459131) - timing
|
||
|
|
step:46 - global_seqlen/min:12586 - global_seqlen/max:30489 - global_seqlen/minmax_diff:17903 - global_seqlen/balanced_min:14803 - global_seqlen/balanced_max:22222 - global_seqlen/mean:17591.25 - actor/entropy:0.06621301919221878 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.7000388503074646 - training/rollout_probs_diff_mean:0.0016268962062895298 - training/rollout_probs_diff_std:0.008891762234270573 - training/rollout_actor_probs_pearson_corr:0.997891366481781 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - self_distillation/correct_sample_fraction:0.67578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8359375 - rollout_corr/rollout_is_mean:0.9999244213104248 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:5.691357728210278e-05 - rollout_corr/rollout_is_max:1.648771047592163 - rollout_corr/rollout_is_min:0.20913249254226685 - rollout_corr/rollout_is_std:0.021762777119874954 - rollout_corr/rollout_is_eff_sample_size:0.99952648660448 - rollout_corr/rollout_is_seq_mean:0.9998850226402283 - rollout_corr/rollout_is_seq_std:0.0018324963748455048 - rollout_corr/rollout_is_seq_max:1.0079123973846436 - rollout_corr/rollout_is_seq_min:0.9941005706787109 - rollout_corr/rollout_is_seq_max_deviation:0.007912397384643555 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0876929117366672) - rollout_corr/training_log_ppl:np.float64(0.08215764638555356) - rollout_corr/kl:np.float64(0.0004236203267915517) - rollout_corr/k3_kl:np.float64(0.00030869100003624774) - rollout_corr/rollout_ppl:np.float64(1.0872142277657986) - rollout_corr/rollout_log_ppl:np.float64(0.08173402504645821) - rollout_corr/log_ppl_diff:np.float64(0.00042362133909534805) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013981999525611855) - rollout_corr/log_ppl_diff_max:np.float64(0.006462082266807556) - rollout_corr/log_ppl_diff_min:np.float64(-0.006531879305839539) - rollout_corr/ppl_ratio:np.float64(1.0004254570230842) - rollout_corr/chi2_token:np.float64(0.0003700978122651577) - rollout_corr/chi2_seq:np.float64(0.08962852461263537) - actor/pg_loss:np.float64(0.022780441144014674) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1640625) - actor/grad_norm:np.float64(0.2267768681049347) - perf/mfu/actor:np.float64(0.03127041960605621) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(146.43690490722656) - actor/lr:np.float64(1e-05) - training/global_step:46 - training/epoch:0 - 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:-0.09571796655654907 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.09571796655654907 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:274.5390625 - response_length/max:8192.0 - response_length/min:89.0 - response_length/clip_ratio:0.01171875 - response_length_non_aborted/mean:274.5390625 - response_length_non_aborted/max:8192.0 - response_length_non_aborted/min:89.0 - response_length_non_aborted/clip_ratio:0.01171875 - response/aborted_ratio:0.0 - prompt_length/mean:275.1875 - prompt_length/max:595.0 - prompt_length/min:165.0 - 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:0.00010778754949569702 - timing_s/agent_loop/generate_sequences/min:np.float64(0.641911469399929) - timing_s/agent_loop/generate_sequences/max:np.float64(50.53799136355519) - timing_s/agent_loop/generate_sequences/mean:np.float64(2.34711665625945
|
||
|
|
step:47 - global_seqlen/min:11972 - global_seqlen/max:18317 - global_seqlen/minmax_diff:6345 - global_seqlen/balanced_min:14710 - global_seqlen/balanced_max:14758 - global_seqlen/mean:14717.875 - actor/entropy:0.0710403323173523 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.21512258052825928 - training/rollout_probs_diff_mean:0.0018978065345436335 - training/rollout_probs_diff_std:0.0090432558208704 - training/rollout_actor_probs_pearson_corr:0.9979128241539001 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.73828125 - self_distillation/correct_sample_fraction:0.5859375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.73828125 - rollout_corr/rollout_is_mean:0.9999450445175171 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6097140312194824 - rollout_corr/rollout_is_min:0.5686210989952087 - rollout_corr/rollout_is_std:0.023823421448469162 - rollout_corr/rollout_is_eff_sample_size:0.9994328855829984 - rollout_corr/rollout_is_seq_mean:0.9999264478683472 - rollout_corr/rollout_is_seq_std:0.001821366255171597 - rollout_corr/rollout_is_seq_max:1.0064975023269653 - rollout_corr/rollout_is_seq_min:0.9943331480026245 - rollout_corr/rollout_is_seq_max_deviation:0.006497502326965332 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0721829500980675) - rollout_corr/training_log_ppl:np.float64(0.06795982785510546) - rollout_corr/kl:np.float64(0.000347097932527185) - rollout_corr/k3_kl:np.float64(0.0002735320489369577) - rollout_corr/rollout_ppl:np.float64(1.0718094683252275) - rollout_corr/rollout_log_ppl:np.float64(0.06761272950447506) - rollout_corr/log_ppl_diff:np.float64(0.00034709835063040373) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013902388959650125) - rollout_corr/log_ppl_diff_max:np.float64(0.006981182843446732) - rollout_corr/log_ppl_diff_min:np.float64(-0.00567246600985527) - rollout_corr/ppl_ratio:np.float64(1.0003488475922495) - rollout_corr/chi2_token:np.float64(0.00039295689202845097) - rollout_corr/chi2_seq:np.float64(0.07276946469210088) - actor/pg_loss:np.float64(0.01831813951032757) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.26171875) - actor/grad_norm:np.float64(0.21489128470420837) - perf/mfu/actor:np.float64(0.02722952827575236) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(146.43510055541992) - actor/lr:np.float64(1e-05) - training/global_step:47 - training/epoch:0 - 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:0.003287391969934106 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.003287391969934106 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:166.65234375 - response_length/max:592.0 - response_length/min:94.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:166.65234375 - response_length_non_aborted/max:592.0 - response_length_non_aborted/min:94.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:293.28125 - prompt_length/max:950.0 - prompt_length/min:165.0 - 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:9.459815919399261e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1122638117522001) - timing_s/agent_loop/generate_sequences/max:np.float64(4.120073888450861) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.890857254860748) - timing_s/agent_loop/tool_calls/
|
||
|
|
step:48 - global_seqlen/min:10765 - global_seqlen/max:22670 - global_seqlen/minmax_diff:11905 - global_seqlen/balanced_min:13605 - global_seqlen/balanced_max:21003 - global_seqlen/mean:14531.0 - actor/entropy:0.05110728740692139 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3797518312931061 - training/rollout_probs_diff_mean:0.0013046754756942391 - training/rollout_probs_diff_std:0.007485178764909506 - training/rollout_actor_probs_pearson_corr:0.9980976581573486 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.74609375 - self_distillation/correct_sample_fraction:0.66796875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.74609375 - rollout_corr/rollout_is_mean:0.9998725652694702 - rollout_corr/rollout_is_ratio_fraction_high:2.0736562873935327e-05 - rollout_corr/rollout_is_ratio_fraction_low:2.0736562873935327e-05 - rollout_corr/rollout_is_max:2.0332674980163574 - rollout_corr/rollout_is_min:0.4464702308177948 - rollout_corr/rollout_is_std:0.020138273015618324 - rollout_corr/rollout_is_eff_sample_size:0.9995943761758205 - rollout_corr/rollout_is_seq_mean:0.9998383522033691 - rollout_corr/rollout_is_seq_std:0.001851289882324636 - rollout_corr/rollout_is_seq_max:1.0069019794464111 - rollout_corr/rollout_is_seq_min:0.9923463463783264 - rollout_corr/rollout_is_seq_max_deviation:0.007653653621673584 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0641534626483917) - rollout_corr/training_log_ppl:np.float64(0.06074093675891845) - rollout_corr/kl:np.float64(0.0003994460914868436) - rollout_corr/k3_kl:np.float64(0.0002377950319914703) - rollout_corr/rollout_ppl:np.float64(1.0636993627995253) - rollout_corr/rollout_log_ppl:np.float64(0.060341490250493734) - rollout_corr/log_ppl_diff:np.float64(0.00039944650842471674) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013448067429067123) - rollout_corr/log_ppl_diff_max:np.float64(0.008481428027153015) - rollout_corr/log_ppl_diff_min:np.float64(-0.006191611289978027) - rollout_corr/ppl_ratio:np.float64(1.0004012978170067) - rollout_corr/chi2_token:np.float64(0.00015194201841950417) - rollout_corr/chi2_seq:np.float64(0.016621158691123128) - actor/pg_loss:np.float64(0.01742617581219008) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.25390625) - actor/grad_norm:np.float64(0.22651104629039764) - perf/mfu/actor:np.float64(0.02531592628058647) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(146.32642364501953) - actor/lr:np.float64(1e-05) - training/global_step:48 - training/epoch:0 - critic/score/mean:0.66796875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.66796875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.02241881564259529 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.02241881564259529 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:188.375 - response_length/max:8192.0 - response_length/min:86.0 - response_length/clip_ratio:0.00390625 - response_length_non_aborted/mean:188.375 - response_length_non_aborted/max:8192.0 - response_length_non_aborted/min:86.0 - response_length_non_aborted/clip_ratio:0.00390625 - response/aborted_ratio:0.0 - prompt_length/mean:265.71875 - prompt_length/max:563.0 - prompt_length/min:163.0 - 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:9.629316627979279e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(0.864195154979825) - timing_s/agent_loop/generate_sequences/max:np.float64(48.285674726590514) - timing_s/agent_loop/generate_sequences/mean:np.float64(
|
||
|
|
step:49 - global_seqlen/min:13053 - global_seqlen/max:15481 - global_seqlen/minmax_diff:2428 - global_seqlen/balanced_min:14402 - global_seqlen/balanced_max:14404 - global_seqlen/mean:14402.75 - actor/entropy:0.05611808970570564 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1949445605278015 - training/rollout_probs_diff_mean:0.0017106728628277779 - training/rollout_probs_diff_std:0.009412742219865322 - training/rollout_actor_probs_pearson_corr:0.9972189664840698 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.87109375 - self_distillation/correct_sample_fraction:0.73828125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.87109375 - rollout_corr/rollout_is_mean:1.0000537633895874 - rollout_corr/rollout_is_ratio_fraction_high:6.8421293690335e-05 - rollout_corr/rollout_is_ratio_fraction_low:6.8421293690335e-05 - rollout_corr/rollout_is_max:2.0341246128082275 - rollout_corr/rollout_is_min:0.48374107480049133 - rollout_corr/rollout_is_std:0.025453928858041763 - rollout_corr/rollout_is_eff_sample_size:0.9993527551271193 - rollout_corr/rollout_is_seq_mean:1.0001276731491089 - rollout_corr/rollout_is_seq_std:0.0018110385863110423 - rollout_corr/rollout_is_seq_max:1.0062569379806519 - rollout_corr/rollout_is_seq_min:0.9950587153434753 - rollout_corr/rollout_is_seq_max_deviation:0.0062569379806518555 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0543933575972915) - rollout_corr/training_log_ppl:np.float64(0.051720200484851375) - rollout_corr/kl:np.float64(0.0001631105899075891) - rollout_corr/k3_kl:np.float64(0.00029083043149642556) - rollout_corr/rollout_ppl:np.float64(1.0542059908621013) - rollout_corr/rollout_log_ppl:np.float64(0.05155708950525195) - rollout_corr/log_ppl_diff:np.float64(0.00016311097959942344) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012463130854598603) - rollout_corr/log_ppl_diff_max:np.float64(0.0059377700090408325) - rollout_corr/log_ppl_diff_min:np.float64(-0.0053930096328258514) - rollout_corr/ppl_ratio:np.float64(1.0001647043973207) - rollout_corr/chi2_token:np.float64(0.0008709710091352463) - rollout_corr/chi2_seq:np.float64(0.12359315692447126) - actor/pg_loss:np.float64(0.01638743910723406) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.12890625) - actor/grad_norm:np.float64(0.2278270721435547) - perf/mfu/actor:np.float64(0.02660612828082059) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(146.37047576904297) - actor/lr:np.float64(1e-05) - training/global_step:49 - training/epoch:0 - 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:0.001878734678030014 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.001878734678030014 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:171.2734375 - response_length/max:589.0 - response_length/min:91.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:171.2734375 - response_length_non_aborted/max:589.0 - response_length_non_aborted/min:91.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:278.8125 - prompt_length/max:418.0 - prompt_length/min:176.0 - 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:5.387887358665466e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1531347017735243) - timing_s/agent_loop/generate_sequences/max:np.float64(4.218779465183616) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.971782235
|
||
|
|
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 50}
|
||
|
|
validation generation end
|
||
|
|
[prompt] system
|
||
|
|
|
||
|
|
Given a question and four options, please select the right answer. Respond in the following format:
|
||
|
|
<reasoning>
|
||
|
|
...
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
...
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
user
|
||
|
|
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||
|
|
|
||
|
|
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||
|
|
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||
|
|
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||
|
|
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||
|
|
Please reason step by step.
|
||
|
|
assistant
|
||
|
|
<think>
|
||
|
|
|
||
|
|
</think>
|
||
|
|
|
||
|
|
|
||
|
|
[response] <reasoning>
|
||
|
|
The product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" contains a chlorine atom, a fluorine-substituted aromatic ring, and a nitrogen-containing group. Option A contains a carboxylic acid group (O=C), a chlorine-substituted aromatic ring (Cc2ccc(Cl)cc2), and a nitrogen-containing group (N). The fluorine-substituted aromatic ring is not explicitly present in Option A, but the product's structure includes a fluorine-substituted aromatic ring, so Option A is the most consistent with the product's structure. The other options either lack the necessary functional groups or do not match the product's structure. Therefore, the correct answer is Option A.
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
[ground_truth] A
|
||
|
|
[score] 1.0
|
||
|
|
[acc] 1.0
|
||
|
|
[pred] A
|
||
|
|
[incorrect_format] 1
|
||
|
|
[feedback]
|
||
|
|
local_global_step_folder: /workspace/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-SDPO_TR-Qwen-Qwen3-4B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_50
|
||
|
|
step:50 - global_seqlen/min:12273 - global_seqlen/max:20550 - global_seqlen/minmax_diff:8277 - global_seqlen/balanced_min:14684 - global_seqlen/balanced_max:14687 - global_seqlen/mean:14685.375 - actor/entropy:0.07561355829238892 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.29304978251457214 - training/rollout_probs_diff_mean:0.001762501080520451 - training/rollout_probs_diff_std:0.008273256942629814 - training/rollout_actor_probs_pearson_corr:0.9984067678451538 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.89453125 - self_distillation/correct_sample_fraction:0.671875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.89453125 - rollout_corr/rollout_is_mean:1.0000308752059937 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.1802166884299368e-05 - rollout_corr/rollout_is_max:1.5995075702667236 - rollout_corr/rollout_is_min:0.478391170501709 - rollout_corr/rollout_is_std:0.02248215675354004 - rollout_corr/rollout_is_eff_sample_size:0.9994946888714428 - rollout_corr/rollout_is_seq_mean:1.0000187158584595 - rollout_corr/rollout_is_seq_std:0.0017447519348934293 - rollout_corr/rollout_is_seq_max:1.0052552223205566 - rollout_corr/rollout_is_seq_min:0.9940527677536011 - rollout_corr/rollout_is_seq_max_deviation:0.005947232246398926 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.07186457561329) - rollout_corr/training_log_ppl:np.float64(0.0678128774943616) - rollout_corr/kl:np.float64(0.0002231121888520704) - rollout_corr/k3_kl:np.float64(0.00024176058257628467) - rollout_corr/rollout_ppl:np.float64(1.0716128391213715) - rollout_corr/rollout_log_ppl:np.float64(0.06758976506034742) - rollout_corr/log_ppl_diff:np.float64(0.00022311243401418324) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013058617228125513) - rollout_corr/log_ppl_diff_max:np.float64(0.006422758102416992) - rollout_corr/log_ppl_diff_min:np.float64(-0.004845947027206421) - rollout_corr/ppl_ratio:np.float64(1.0002247104421258) - rollout_corr/chi2_token:np.float64(0.0005261811893433332) - rollout_corr/chi2_seq:np.float64(0.09909631311893463) - actor/pg_loss:np.float64(0.020501964903928638) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.10546875) - actor/grad_norm:np.float64(0.24316243827342987) - perf/mfu/actor:np.float64(0.027283914169174982) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(146.43873596191406) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.6779761904761905) - val-aux/sciknoweval/reward/std@16:np.float64(0.07888399639762744) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7098666666666666) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.06784212836896987) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6465238095238094) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.06491490310697885) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.6784380952380952) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.0788026045929127) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7356285714285714) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.052098506036682105) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6202761904761905) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.04708022689022077) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.6832761904761905) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.06202960446036021) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.755395238095238) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.0397951149208317) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6020857142857143) - val
|
||
|
|
step:51 - global_seqlen/min:13785 - global_seqlen/max:17393 - global_seqlen/minmax_diff:3608 - global_seqlen/balanced_min:15157 - global_seqlen/balanced_max:15158 - global_seqlen/mean:15157.75 - actor/entropy:0.0681234747171402 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6351427435874939 - training/rollout_probs_diff_mean:0.0020061070099473 - training/rollout_probs_diff_std:0.012267793528735638 - training/rollout_actor_probs_pearson_corr:0.9959521293640137 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.74609375 - self_distillation/correct_sample_fraction:0.625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.74609375 - rollout_corr/rollout_is_mean:1.0000672340393066 - rollout_corr/rollout_is_ratio_fraction_high:9.696029155747965e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00014544044097419828 - rollout_corr/rollout_is_max:4.48164701461792 - rollout_corr/rollout_is_min:0.2231280654668808 - rollout_corr/rollout_is_std:0.026788802817463875 - rollout_corr/rollout_is_eff_sample_size:0.9992826366379459 - rollout_corr/rollout_is_seq_mean:1.0001052618026733 - rollout_corr/rollout_is_seq_std:0.002782619558274746 - rollout_corr/rollout_is_seq_max:1.0199143886566162 - rollout_corr/rollout_is_seq_min:0.9943129420280457 - rollout_corr/rollout_is_seq_max_deviation:0.01991438865661621 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0677353329956532) - rollout_corr/training_log_ppl:np.float64(0.06431337321657793) - rollout_corr/kl:np.float64(0.0003540552945526798) - rollout_corr/k3_kl:np.float64(0.0004593753830577432) - rollout_corr/rollout_ppl:np.float64(1.0673504914157093) - rollout_corr/rollout_log_ppl:np.float64(0.06395931754764206) - rollout_corr/log_ppl_diff:np.float64(0.0003540556689358709) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018492478317284622) - rollout_corr/log_ppl_diff_max:np.float64(0.008627243340015411) - rollout_corr/log_ppl_diff_min:np.float64(-0.007441259920597076) - rollout_corr/ppl_ratio:np.float64(1.0003571410197765) - rollout_corr/chi2_token:np.float64(0.0013932900037616491) - rollout_corr/chi2_seq:np.float64(0.2077259263023734) - actor/pg_loss:np.float64(0.019668076191010186) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.25390625) - actor/grad_norm:np.float64(0.23474666476249695) - perf/mfu/actor:np.float64(0.027817779555644587) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(147.69947814941406) - actor/lr:np.float64(1e-05) - training/global_step:51 - training/epoch:0 - critic/score/mean:0.625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0002636107965372503 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0002636107965372503 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:161.1484375 - response_length/max:556.0 - response_length/min:94.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:161.1484375 - response_length_non_aborted/max:556.0 - response_length_non_aborted/min:94.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:312.53125 - prompt_length/max:608.0 - prompt_length/min:172.0 - 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:4.118867218494415e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1142488569021225) - timing_s/agent_loop/generate_sequences/max:np.float64(4.022502403706312) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.7922559326543706) - timing_s
|
||
|
|
step:52 - global_seqlen/min:10273 - global_seqlen/max:20280 - global_seqlen/minmax_diff:10007 - global_seqlen/balanced_min:13673 - global_seqlen/balanced_max:19900 - global_seqlen/mean:14452.0 - actor/entropy:0.050270870327949524 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.42640915513038635 - training/rollout_probs_diff_mean:0.0013177337823435664 - training/rollout_probs_diff_std:0.007472561672329903 - training/rollout_actor_probs_pearson_corr:0.9980469346046448 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.640625 - self_distillation/correct_sample_fraction:0.5078125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.640625 - rollout_corr/rollout_is_mean:0.9999501705169678 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:4.1397583117941394e-05 - rollout_corr/rollout_is_max:1.6976341009140015 - rollout_corr/rollout_is_min:0.4572162330150604 - rollout_corr/rollout_is_std:0.020170804113149643 - rollout_corr/rollout_is_eff_sample_size:0.9995931850512308 - rollout_corr/rollout_is_seq_mean:0.9998766183853149 - rollout_corr/rollout_is_seq_std:0.0017649311339482665 - rollout_corr/rollout_is_seq_max:1.0053123235702515 - rollout_corr/rollout_is_seq_min:0.9933837056159973 - rollout_corr/rollout_is_seq_max_deviation:0.0066162943840026855 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0542300478555262) - rollout_corr/training_log_ppl:np.float64(0.051696847780476674) - rollout_corr/kl:np.float64(0.00035296082031877063) - rollout_corr/k3_kl:np.float64(0.0002296415946676407) - rollout_corr/rollout_ppl:np.float64(1.0538452421315014) - rollout_corr/rollout_log_ppl:np.float64(0.051343887027087476) - rollout_corr/log_ppl_diff:np.float64(0.00035296075338919763) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012989518222639163) - rollout_corr/log_ppl_diff_max:np.float64(0.008090570569038391) - rollout_corr/log_ppl_diff_min:np.float64(-0.004333682358264923) - rollout_corr/ppl_ratio:np.float64(1.0003546955995262) - rollout_corr/chi2_token:np.float64(0.00020034168846905231) - rollout_corr/chi2_seq:np.float64(0.0533747102599591) - actor/pg_loss:np.float64(0.010998465798593315) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.359375) - actor/grad_norm:np.float64(0.19872982800006866) - perf/mfu/actor:np.float64(0.025650034611372243) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(147.08522033691406) - actor/lr:np.float64(1e-05) - training/global_step:52 - training/epoch:0 - critic/score/mean:0.5078125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5078125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0194232277572155 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0194232277572155 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:188.71875 - response_length/max:6828.0 - response_length/min:89.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:188.71875 - response_length_non_aborted/max:6828.0 - response_length_non_aborted/min:89.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:262.90625 - prompt_length/max:453.0 - prompt_length/min:163.0 - 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:2.7719885110855103e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0929646994918585) - timing_s/agent_loop/generate_sequences/max:np.float64(33.341456323862076) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.9367772706827964) - timing_
|
||
|
|
step:53 - global_seqlen/min:12263 - global_seqlen/max:16347 - global_seqlen/minmax_diff:4084 - global_seqlen/balanced_min:14451 - global_seqlen/balanced_max:14452 - global_seqlen/mean:14451.5 - actor/entropy:0.06619513779878616 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3348143696784973 - training/rollout_probs_diff_mean:0.0016967304982244968 - training/rollout_probs_diff_std:0.0088465241715312 - training/rollout_actor_probs_pearson_corr:0.997840166091919 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.76953125 - self_distillation/correct_sample_fraction:0.6640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.76953125 - rollout_corr/rollout_is_mean:1.0001972913742065 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.5032542907865718e-05 - rollout_corr/rollout_is_max:1.7752958536148071 - rollout_corr/rollout_is_min:0.42816057801246643 - rollout_corr/rollout_is_std:0.023261362686753273 - rollout_corr/rollout_is_eff_sample_size:0.9994595588971189 - rollout_corr/rollout_is_seq_mean:1.000156283378601 - rollout_corr/rollout_is_seq_std:0.001858926611021161 - rollout_corr/rollout_is_seq_max:1.007813811302185 - rollout_corr/rollout_is_seq_min:0.9958849549293518 - rollout_corr/rollout_is_seq_max_deviation:0.007813811302185059 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0588029497303069) - rollout_corr/training_log_ppl:np.float64(0.05579841929647955) - rollout_corr/kl:np.float64(8.970298647770214e-05) - rollout_corr/k3_kl:np.float64(0.0002460099959896711) - rollout_corr/rollout_ppl:np.float64(1.0587060991674662) - rollout_corr/rollout_log_ppl:np.float64(0.05570871597956284) - rollout_corr/log_ppl_diff:np.float64(8.970331691671163e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.001330124847299885) - rollout_corr/log_ppl_diff_max:np.float64(0.005807593464851379) - rollout_corr/log_ppl_diff_min:np.float64(-0.006357003003358841) - rollout_corr/ppl_ratio:np.float64(1.000091329216957) - rollout_corr/chi2_token:np.float64(0.0008138609118759632) - rollout_corr/chi2_seq:np.float64(0.16345967934466898) - actor/pg_loss:np.float64(0.016017729722989316) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.23046875) - actor/grad_norm:np.float64(0.2413892149925232) - perf/mfu/actor:np.float64(0.02687997963978201) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(146.82293319702148) - actor/lr:np.float64(1e-05) - training/global_step:53 - training/epoch:0 - critic/score/mean:0.6640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.004818764515221119 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.004818764515221119 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:156.046875 - response_length/max:516.0 - response_length/min:90.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:156.046875 - response_length_non_aborted/max:516.0 - response_length_non_aborted/min:90.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:295.5625 - prompt_length/max:532.0 - prompt_length/min:170.0 - 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:4.171207547187805e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(0.7596689406782389) - timing_s/agent_loop/generate_sequences/max:np.float64(3.4429763201624155) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.4928666922351113) - timing_s/agent_loo
|
||
|
|
step:54 - global_seqlen/min:11042 - global_seqlen/max:24654 - global_seqlen/minmax_diff:13612 - global_seqlen/balanced_min:14668 - global_seqlen/balanced_max:21529 - global_seqlen/mean:15891.875 - actor/entropy:0.053803786635398865 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23699387907981873 - training/rollout_probs_diff_mean:0.001173991127870977 - training/rollout_probs_diff_std:0.006440272554755211 - training/rollout_actor_probs_pearson_corr:0.9986345767974854 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.65234375 - self_distillation/correct_sample_fraction:0.51171875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.65234375 - rollout_corr/rollout_is_mean:1.000104308128357 - rollout_corr/rollout_is_ratio_fraction_high:1.729655014059972e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.256138324737549 - rollout_corr/rollout_is_min:0.5504862070083618 - rollout_corr/rollout_is_std:0.018519317731261253 - rollout_corr/rollout_is_eff_sample_size:0.9996570333311645 - rollout_corr/rollout_is_seq_mean:1.000095248222351 - rollout_corr/rollout_is_seq_std:0.001631860970519483 - rollout_corr/rollout_is_seq_max:1.0075827836990356 - rollout_corr/rollout_is_seq_min:0.9951592683792114 - rollout_corr/rollout_is_seq_max_deviation:0.0075827836990356445 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0529556698165834) - rollout_corr/training_log_ppl:np.float64(0.05036401271638624) - rollout_corr/kl:np.float64(8.996897381274849e-05) - rollout_corr/k3_kl:np.float64(0.00018527261378517568) - rollout_corr/rollout_ppl:np.float64(1.0528563912957907) - rollout_corr/rollout_log_ppl:np.float64(0.05027404336578911) - rollout_corr/log_ppl_diff:np.float64(8.996935059712996e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012019992897194243) - rollout_corr/log_ppl_diff_max:np.float64(0.005825638771057129) - rollout_corr/log_ppl_diff_min:np.float64(-0.005735017359256744) - rollout_corr/ppl_ratio:np.float64(1.0000912856776267) - rollout_corr/chi2_token:np.float64(0.0005739200860261917) - rollout_corr/chi2_seq:np.float64(0.13656649249605834) - actor/pg_loss:np.float64(0.011654071255293275) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.34765625) - actor/grad_norm:np.float64(0.2249014675617218) - perf/mfu/actor:np.float64(0.027442123571277822) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(146.78827285766602) - actor/lr:np.float64(1e-05) - training/global_step:54 - training/epoch:0 - critic/score/mean:0.51171875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.51171875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.002460434101521969 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.002460434101521969 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:225.83984375 - response_length/max:8192.0 - response_length/min:90.0 - response_length/clip_ratio:0.00390625 - response_length_non_aborted/mean:225.83984375 - response_length_non_aborted/max:8192.0 - response_length_non_aborted/min:90.0 - response_length_non_aborted/clip_ratio:0.00390625 - response/aborted_ratio:0.0 - prompt_length/mean:270.78125 - prompt_length/max:597.0 - prompt_length/min:172.0 - 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:2.923049032688141e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.149860581383109) - timing_s/agent_loop/generate_sequences/max:np.float64(48.763225726783276) - timing_s/agent_loop/generate_sequences/mean:np.float64(2.13691365
|
||
|
|
step:55 - global_seqlen/min:12834 - global_seqlen/max:16109 - global_seqlen/minmax_diff:3275 - global_seqlen/balanced_min:14748 - global_seqlen/balanced_max:14755 - global_seqlen/mean:14752.5 - actor/entropy:0.057580672204494476 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23105883598327637 - training/rollout_probs_diff_mean:0.0014392229495570064 - training/rollout_probs_diff_std:0.007678190246224403 - training/rollout_actor_probs_pearson_corr:0.9981269240379333 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.8671875 - self_distillation/correct_sample_fraction:0.7421875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8671875 - rollout_corr/rollout_is_mean:1.0000759363174438 - rollout_corr/rollout_is_ratio_fraction_high:2.4957571440609172e-05 - rollout_corr/rollout_is_ratio_fraction_low:2.4957571440609172e-05 - rollout_corr/rollout_is_max:2.0695643424987793 - rollout_corr/rollout_is_min:0.21324193477630615 - rollout_corr/rollout_is_std:0.020574551075696945 - rollout_corr/rollout_is_eff_sample_size:0.9995767478217765 - rollout_corr/rollout_is_seq_mean:1.0000709295272827 - rollout_corr/rollout_is_seq_std:0.001509099267423153 - rollout_corr/rollout_is_seq_max:1.0068371295928955 - rollout_corr/rollout_is_seq_min:0.9955276846885681 - rollout_corr/rollout_is_seq_max_deviation:0.006837129592895508 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0521913673728704) - rollout_corr/training_log_ppl:np.float64(0.04939920129027087) - rollout_corr/kl:np.float64(0.00012914264678798304) - rollout_corr/k3_kl:np.float64(0.0002000106041766675) - rollout_corr/rollout_ppl:np.float64(1.0520513663068414) - rollout_corr/rollout_log_ppl:np.float64(0.04927005856961841) - rollout_corr/log_ppl_diff:np.float64(0.00012914272065245314) - rollout_corr/log_ppl_abs_diff:np.float64(0.001041421282934607) - rollout_corr/log_ppl_diff_max:np.float64(0.006936497986316681) - rollout_corr/log_ppl_diff_min:np.float64(-0.005645032972097397) - rollout_corr/ppl_ratio:np.float64(1.0001303255558014) - rollout_corr/chi2_token:np.float64(0.0005347500555217266) - rollout_corr/chi2_seq:np.float64(0.08593951212242246) - actor/pg_loss:np.float64(0.01740096336334318) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1328125) - actor/grad_norm:np.float64(0.24474173784255981) - perf/mfu/actor:np.float64(0.027080023908404704) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(146.85068893432617) - actor/lr:np.float64(1e-05) - training/global_step:55 - training/epoch:0 - 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:0.021229535341262817 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.021229535341262817 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:156.515625 - response_length/max:662.0 - response_length/min:85.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:156.515625 - response_length_non_aborted/max:662.0 - response_length_non_aborted/min:85.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:304.5 - prompt_length/max:625.0 - prompt_length/min:169.0 - 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:7.290206849575043e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.080026151612401) - timing_s/agent_loop/generate_sequences/max:np.float64(4.484344149008393) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.834276787340059) - t
|
||
|
|
step:56 - global_seqlen/min:12136 - global_seqlen/max:15842 - global_seqlen/minmax_diff:3706 - global_seqlen/balanced_min:13761 - global_seqlen/balanced_max:14110 - global_seqlen/mean:13805.75 - actor/entropy:0.06290365010499954 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2553977966308594 - training/rollout_probs_diff_mean:0.0013078309129923582 - training/rollout_probs_diff_std:0.007081640884280205 - training/rollout_actor_probs_pearson_corr:0.9986499547958374 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.76953125 - self_distillation/correct_sample_fraction:0.58984375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.76953125 - rollout_corr/rollout_is_mean:0.999901533126831 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:4.6985856897663325e-05 - rollout_corr/rollout_is_max:1.6487230062484741 - rollout_corr/rollout_is_min:0.2297939509153366 - rollout_corr/rollout_is_std:0.01874643936753273 - rollout_corr/rollout_is_eff_sample_size:0.9996485753478168 - rollout_corr/rollout_is_seq_mean:0.9998514652252197 - rollout_corr/rollout_is_seq_std:0.0015365928411483765 - rollout_corr/rollout_is_seq_max:1.0061583518981934 - rollout_corr/rollout_is_seq_min:0.9946321845054626 - rollout_corr/rollout_is_seq_max_deviation:0.006158351898193359 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0611660415306687) - rollout_corr/training_log_ppl:np.float64(0.057615773576799256) - rollout_corr/kl:np.float64(0.0003380900895706418) - rollout_corr/k3_kl:np.float64(0.00018951624487079446) - rollout_corr/rollout_ppl:np.float64(1.06080615054816) - rollout_corr/rollout_log_ppl:np.float64(0.057277683103848176) - rollout_corr/log_ppl_diff:np.float64(0.0003380904729510803) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010974310291658185) - rollout_corr/log_ppl_diff_max:np.float64(0.0062238723039627075) - rollout_corr/log_ppl_diff_min:np.float64(-0.005159109830856323) - rollout_corr/ppl_ratio:np.float64(1.0003394377417862) - rollout_corr/chi2_token:np.float64(5.8220699429512024e-05) - rollout_corr/chi2_seq:np.float64(0.04952691262587905) - actor/pg_loss:np.float64(0.014771190783903876) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.23046875) - actor/grad_norm:np.float64(0.21855317056179047) - perf/mfu/actor:np.float64(0.02551060919303675) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(146.8512840270996) - actor/lr:np.float64(1e-05) - training/global_step:56 - training/epoch:0 - critic/score/mean:0.58984375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.58984375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.014527440071105957 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.014527440071105957 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:166.2734375 - response_length/max:878.0 - response_length/min:89.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:166.2734375 - response_length_non_aborted/max:878.0 - response_length_non_aborted/min:89.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:265.15625 - prompt_length/max:458.0 - prompt_length/min:165.0 - 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:0.0001043323427438736 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1515661887824535) - timing_s/agent_loop/generate_sequences/max:np.float64(5.586911920458078) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.8624851164640859) - timi
|
||
|
|
step:57 - global_seqlen/min:12046 - global_seqlen/max:16877 - global_seqlen/minmax_diff:4831 - global_seqlen/balanced_min:13822 - global_seqlen/balanced_max:13822 - global_seqlen/mean:13822.0 - actor/entropy:0.04305357113480568 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3186177611351013 - training/rollout_probs_diff_mean:0.0010850881226360798 - training/rollout_probs_diff_std:0.007101000286638737 - training/rollout_actor_probs_pearson_corr:0.9979764819145203 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.7421875 - self_distillation/correct_sample_fraction:0.671875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7421875 - rollout_corr/rollout_is_mean:1.0000479221343994 - rollout_corr/rollout_is_ratio_fraction_high:2.691065674298443e-05 - rollout_corr/rollout_is_ratio_fraction_low:2.691065674298443e-05 - rollout_corr/rollout_is_max:3.672912120819092 - rollout_corr/rollout_is_min:0.4975969195365906 - rollout_corr/rollout_is_std:0.01947930082678795 - rollout_corr/rollout_is_eff_sample_size:0.9996207007538677 - rollout_corr/rollout_is_seq_mean:1.0000463724136353 - rollout_corr/rollout_is_seq_std:0.0017254749545827508 - rollout_corr/rollout_is_seq_max:1.0114986896514893 - rollout_corr/rollout_is_seq_min:0.9951343536376953 - rollout_corr/rollout_is_seq_max_deviation:0.011498689651489258 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.039251589216292) - rollout_corr/training_log_ppl:np.float64(0.03721142720814896) - rollout_corr/kl:np.float64(0.00014553768819602908) - rollout_corr/k3_kl:np.float64(0.000191883558967966) - rollout_corr/rollout_ppl:np.float64(1.0390931791625917) - rollout_corr/rollout_log_ppl:np.float64(0.037065889293330656) - rollout_corr/log_ppl_diff:np.float64(0.00014553791481830558) - rollout_corr/log_ppl_abs_diff:np.float64(0.001020391137558363) - rollout_corr/log_ppl_diff_max:np.float64(0.0058981627225875854) - rollout_corr/log_ppl_diff_min:np.float64(-0.007298845797777176) - rollout_corr/ppl_ratio:np.float64(1.0001467908732593) - rollout_corr/chi2_token:np.float64(0.0005612140521407127) - rollout_corr/chi2_seq:np.float64(0.08465815661475062) - actor/pg_loss:np.float64(0.008728326913171713) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.2578125) - actor/grad_norm:np.float64(0.18989485502243042) - perf/mfu/actor:np.float64(0.02576324113101456) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(146.82937240600586) - actor/lr:np.float64(1e-05) - training/global_step:57 - 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:-0.001093245460651815 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.001093245460651815 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:145.15625 - response_length/max:412.0 - response_length/min:93.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:145.15625 - response_length_non_aborted/max:412.0 - response_length_non_aborted/min:93.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:286.78125 - prompt_length/max:660.0 - prompt_length/min:172.0 - 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:0.00014624372124671936 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1552259735763073) - timing_s/agent_loop/generate_sequences/max:np.float64(3.243732053786516) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.736313527515449) - tim
|
||
|
|
step:58 - global_seqlen/min:12374 - global_seqlen/max:16584 - global_seqlen/minmax_diff:4210 - global_seqlen/balanced_min:13815 - global_seqlen/balanced_max:13816 - global_seqlen/mean:13815.75 - actor/entropy:0.0564543642103672 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.8388888239860535 - training/rollout_probs_diff_mean:0.0012103425106033683 - training/rollout_probs_diff_std:0.00775111373513937 - training/rollout_actor_probs_pearson_corr:0.9980425238609314 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.87109375 - self_distillation/correct_sample_fraction:0.76171875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.87109375 - rollout_corr/rollout_is_mean:1.0000101327896118 - rollout_corr/rollout_is_ratio_fraction_high:5.043627243139781e-05 - rollout_corr/rollout_is_ratio_fraction_low:7.565441046608612e-05 - rollout_corr/rollout_is_max:4.377130031585693 - rollout_corr/rollout_is_min:0.09207962453365326 - rollout_corr/rollout_is_std:0.01885739341378212 - rollout_corr/rollout_is_eff_sample_size:0.999644405972073 - rollout_corr/rollout_is_seq_mean:0.9999992251396179 - rollout_corr/rollout_is_seq_std:0.002012019045650959 - rollout_corr/rollout_is_seq_max:1.0216212272644043 - rollout_corr/rollout_is_seq_min:0.99472975730896 - rollout_corr/rollout_is_seq_max_deviation:0.021621227264404297 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0488806394860148) - rollout_corr/training_log_ppl:np.float64(0.04639116437715529) - rollout_corr/kl:np.float64(0.0002282902869481518) - rollout_corr/k3_kl:np.float64(0.00022756980897740628) - rollout_corr/rollout_ppl:np.float64(1.0486358972266316) - rollout_corr/rollout_log_ppl:np.float64(0.04616287367167615) - rollout_corr/log_ppl_diff:np.float64(0.00022829070547913943) - rollout_corr/log_ppl_abs_diff:np.float64(0.0009535168990737475) - rollout_corr/log_ppl_diff_max:np.float64(0.006817936897277832) - rollout_corr/log_ppl_diff_min:np.float64(-0.00942557118833065) - rollout_corr/ppl_ratio:np.float64(1.0002296171151102) - rollout_corr/chi2_token:np.float64(0.0006204620003700256) - rollout_corr/chi2_seq:np.float64(0.09274240606464446) - actor/pg_loss:np.float64(0.01682569802665057) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.12890625) - actor/grad_norm:np.float64(0.20225954055786133) - perf/mfu/actor:np.float64(0.025696149196778983) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(146.79572677612305) - actor/lr:np.float64(1e-05) - training/global_step:58 - training/epoch:0 - 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:-0.00717140780761838 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00717140780761838 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:154.8984375 - response_length/max:483.0 - response_length/min:92.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:154.8984375 - response_length_non_aborted/max:483.0 - response_length_non_aborted/min:92.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:276.84375 - prompt_length/max:598.0 - prompt_length/min:170.0 - 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:0.0001271907240152359 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1525416430085897) - timing_s/agent_loop/generate_sequences/max:np.float64(3.565409915521741) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.774068693193839
|
||
|
|
step:59 - global_seqlen/min:12717 - global_seqlen/max:16519 - global_seqlen/minmax_diff:3802 - global_seqlen/balanced_min:14104 - global_seqlen/balanced_max:14104 - global_seqlen/mean:14104.0 - actor/entropy:0.0858306959271431 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4576534628868103 - training/rollout_probs_diff_mean:0.0014705249341204762 - training/rollout_probs_diff_std:0.0070825605653226376 - training/rollout_actor_probs_pearson_corr:0.9989221692085266 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.77734375 - self_distillation/correct_sample_fraction:0.63671875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.77734375 - rollout_corr/rollout_is_mean:1.0000361204147339 - rollout_corr/rollout_is_ratio_fraction_high:2.268602474941872e-05 - rollout_corr/rollout_is_ratio_fraction_low:2.268602474941872e-05 - rollout_corr/rollout_is_max:2.451803207397461 - rollout_corr/rollout_is_min:0.23710320889949799 - rollout_corr/rollout_is_std:0.019549552351236343 - rollout_corr/rollout_is_eff_sample_size:0.9996180801454941 - rollout_corr/rollout_is_seq_mean:1.0000218152999878 - rollout_corr/rollout_is_seq_std:0.0013853474520146847 - rollout_corr/rollout_is_seq_max:1.0061777830123901 - rollout_corr/rollout_is_seq_min:0.9950460195541382 - rollout_corr/rollout_is_seq_max_deviation:0.006177783012390137 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0663618030957878) - rollout_corr/training_log_ppl:np.float64(0.061598015423669494) - rollout_corr/kl:np.float64(0.0001568761066628177) - rollout_corr/k3_kl:np.float64(0.00017866073002048696) - rollout_corr/rollout_ppl:np.float64(1.0661960523575544) - rollout_corr/rollout_log_ppl:np.float64(0.061441138876475065) - rollout_corr/log_ppl_diff:np.float64(0.0001568765471944289) - rollout_corr/log_ppl_abs_diff:np.float64(0.0009393328392519606) - rollout_corr/log_ppl_diff_max:np.float64(0.007103202864527702) - rollout_corr/log_ppl_diff_min:np.float64(-0.004075981676578522) - rollout_corr/ppl_ratio:np.float64(1.000157886883244) - rollout_corr/chi2_token:np.float64(0.0003975392319262028) - rollout_corr/chi2_seq:np.float64(0.06792250112630427) - actor/pg_loss:np.float64(0.011977420326942934) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.22265625) - actor/grad_norm:np.float64(0.17822754383087158) - perf/mfu/actor:np.float64(0.026102517261822555) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(146.77954864501953) - actor/lr:np.float64(1e-05) - training/global_step:59 - 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:0.003170371986925602 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.003170371986925602 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:172.1875 - response_length/max:628.0 - response_length/min:100.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:172.1875 - response_length_non_aborted/max:628.0 - response_length_non_aborted/min:100.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:268.5625 - prompt_length/max:615.0 - prompt_length/min:172.0 - 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:8.427724242210388e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.129058724269271) - timing_s/agent_loop/generate_sequences/max:np.float64(4.439052080735564) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.87158564445
|
||
|
|
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 60}
|
||
|
|
validation generation end
|
||
|
|
[prompt] system
|
||
|
|
|
||
|
|
Given a question and four options, please select the right answer. Respond in the following format:
|
||
|
|
<reasoning>
|
||
|
|
...
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
...
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
user
|
||
|
|
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||
|
|
|
||
|
|
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||
|
|
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||
|
|
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||
|
|
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||
|
|
Please reason step by step.
|
||
|
|
assistant
|
||
|
|
<think>
|
||
|
|
|
||
|
|
</think>
|
||
|
|
|
||
|
|
|
||
|
|
[response] <reasoning>
|
||
|
|
The product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" contains a nitrogen-containing group (N), a chlorine-substituted aromatic ring (Cc2ccc(Cl)cc2C(F)(F)F), and a fluorine-substituted aromatic ring (C(F)(F)F). Option A contains a nitrogen-containing group (N), a chlorine-substituted aromatic ring (Cc2ccc(Cl)cc2C(F)(F)F), and a fluorine-substituted aromatic ring (C(F)(F)F), which matches the product's structure. The other options either lack the nitrogen-containing group or do not match the product's structure. Therefore, the correct answer is Option A.
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
[ground_truth] A
|
||
|
|
[score] 1.0
|
||
|
|
[acc] 1.0
|
||
|
|
[pred] A
|
||
|
|
[incorrect_format] 1
|
||
|
|
[feedback]
|
||
|
|
local_global_step_folder: /workspace/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-SDPO_TR-Qwen-Qwen3-4B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_60
|
||
|
|
step:60 - global_seqlen/min:11744 - global_seqlen/max:15139 - global_seqlen/minmax_diff:3395 - global_seqlen/balanced_min:13349 - global_seqlen/balanced_max:13366 - global_seqlen/mean:13352.0 - actor/entropy:0.06620658934116364 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.29316267371177673 - training/rollout_probs_diff_mean:0.0013730258215218782 - training/rollout_probs_diff_std:0.008046751841902733 - training/rollout_actor_probs_pearson_corr:0.9982349276542664 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.875 - self_distillation/correct_sample_fraction:0.69140625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.875 - rollout_corr/rollout_is_mean:0.9999560117721558 - rollout_corr/rollout_is_ratio_fraction_high:5.163155583431944e-05 - rollout_corr/rollout_is_ratio_fraction_low:5.163155583431944e-05 - rollout_corr/rollout_is_max:2.089754343032837 - rollout_corr/rollout_is_min:0.46648940443992615 - rollout_corr/rollout_is_std:0.020114580169320107 - rollout_corr/rollout_is_eff_sample_size:0.9995953290775362 - rollout_corr/rollout_is_seq_mean:0.9999569654464722 - rollout_corr/rollout_is_seq_std:0.0016783947357907891 - rollout_corr/rollout_is_seq_max:1.0077065229415894 - rollout_corr/rollout_is_seq_min:0.9930129051208496 - rollout_corr/rollout_is_seq_max_deviation:0.0077065229415893555 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0649936739355326) - rollout_corr/training_log_ppl:np.float64(0.060476485850031736) - rollout_corr/kl:np.float64(0.00024308692715080404) - rollout_corr/k3_kl:np.float64(0.00020008101425591552) - rollout_corr/rollout_ppl:np.float64(1.0647307941690087) - rollout_corr/rollout_log_ppl:np.float64(0.06023339794489857) - rollout_corr/log_ppl_diff:np.float64(0.0002430879051331658) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010607241043913973) - rollout_corr/log_ppl_diff_max:np.float64(0.008530355989933014) - rollout_corr/log_ppl_diff_min:np.float64(-0.005089845508337021) - rollout_corr/ppl_ratio:np.float64(1.0002445632126182) - rollout_corr/chi2_token:np.float64(0.00032317149452865124) - rollout_corr/chi2_seq:np.float64(0.045475199818611145) - actor/pg_loss:np.float64(0.014449079333580528) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.125) - actor/grad_norm:np.float64(0.2238793820142746) - perf/mfu/actor:np.float64(0.024764171802649664) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(147.60906982421875) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.680952380952381) - val-aux/sciknoweval/reward/std@16:np.float64(0.06463960621060422) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7058809523809525) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.06235704865545992) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6561285714285713) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.046362899514005344) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.6813761904761905) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.06482839891835555) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7304761904761905) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.055523436616170284) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6403523809523809) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.030870032418694168) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.677757142857143) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.048198906964557434) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7542333333333334) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.045197200263492526) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.628714285
|
||
|
|
step:61 - global_seqlen/min:10531 - global_seqlen/max:16318 - global_seqlen/minmax_diff:5787 - global_seqlen/balanced_min:14074 - global_seqlen/balanced_max:14078 - global_seqlen/mean:14076.75 - actor/entropy:0.059575147926807404 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.47440558671951294 - training/rollout_probs_diff_mean:0.0012028295313939452 - training/rollout_probs_diff_std:0.007416948210448027 - training/rollout_actor_probs_pearson_corr:0.998302698135376 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8046875 - self_distillation/correct_sample_fraction:0.69140625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8046875 - rollout_corr/rollout_is_mean:0.9997712969779968 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0001098080538213253 - rollout_corr/rollout_is_max:1.5607712268829346 - rollout_corr/rollout_is_min:0.19285432994365692 - rollout_corr/rollout_is_std:0.01838202215731144 - rollout_corr/rollout_is_eff_sample_size:0.9996621558413732 - rollout_corr/rollout_is_seq_mean:0.9998639822006226 - rollout_corr/rollout_is_seq_std:0.0013256744714453816 - rollout_corr/rollout_is_seq_max:1.004337191581726 - rollout_corr/rollout_is_seq_min:0.9934490323066711 - rollout_corr/rollout_is_seq_max_deviation:0.006550967693328857 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0465027811005712) - rollout_corr/training_log_ppl:np.float64(0.04388999806269567) - rollout_corr/kl:np.float64(0.00031523574030645207) - rollout_corr/k3_kl:np.float64(0.00017924927377371467) - rollout_corr/rollout_ppl:np.float64(1.0461620846763253) - rollout_corr/rollout_log_ppl:np.float64(0.0435747618906106) - rollout_corr/log_ppl_diff:np.float64(0.00031523617208506494) - rollout_corr/log_ppl_abs_diff:np.float64(0.0009351294846027258) - rollout_corr/log_ppl_diff_max:np.float64(0.008085810579359531) - rollout_corr/log_ppl_diff_min:np.float64(-0.0034272000193595886) - rollout_corr/ppl_ratio:np.float64(1.0003164175432175) - rollout_corr/chi2_token:np.float64(3.889226354658604e-05) - rollout_corr/chi2_seq:np.float64(-0.011462690308690071) - actor/pg_loss:np.float64(0.01075867335175218) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1953125) - actor/grad_norm:np.float64(0.1485450118780136) - perf/mfu/actor:np.float64(0.026130653216077507) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.94947814941406) - actor/lr:np.float64(1e-05) - training/global_step:61 - training/epoch:1 - critic/score/mean:0.69140625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.69140625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0012518118601292372 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0012518118601292372 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:177.8671875 - response_length/max:577.0 - response_length/min:100.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:177.8671875 - response_length_non_aborted/max:577.0 - response_length_non_aborted/min:100.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:262.03125 - prompt_length/max:462.0 - prompt_length/min:177.0 - 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:0.00013461336493492126 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1704022362828255) - timing_s/agent_loop/generate_sequences/max:np.float64(4.18418750166893) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.9195357262869948) -
|
||
|
|
step:62 - global_seqlen/min:12623 - global_seqlen/max:16755 - global_seqlen/minmax_diff:4132 - global_seqlen/balanced_min:14504 - global_seqlen/balanced_max:14505 - global_seqlen/mean:14504.875 - actor/entropy:0.0662456825375557 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.30675792694091797 - training/rollout_probs_diff_mean:0.0012798206880688667 - training/rollout_probs_diff_std:0.007177107036113739 - training/rollout_actor_probs_pearson_corr:0.9985625147819519 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.8671875 - self_distillation/correct_sample_fraction:0.7109375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8671875 - rollout_corr/rollout_is_mean:0.999970555305481 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.7275270223617554 - rollout_corr/rollout_is_min:0.6085533499717712 - rollout_corr/rollout_is_std:0.018406324088573456 - rollout_corr/rollout_is_eff_sample_size:0.9996612028122431 - rollout_corr/rollout_is_seq_mean:1.0000133514404297 - rollout_corr/rollout_is_seq_std:0.0014781567733734846 - rollout_corr/rollout_is_seq_max:1.006226897239685 - rollout_corr/rollout_is_seq_min:0.9960687756538391 - rollout_corr/rollout_is_seq_max_deviation:0.006226897239685059 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.057463594712317) - rollout_corr/training_log_ppl:np.float64(0.05329444925558846) - rollout_corr/kl:np.float64(0.0001371526236845222) - rollout_corr/k3_kl:np.float64(0.00015063608808168225) - rollout_corr/rollout_ppl:np.float64(1.0573155363090336) - rollout_corr/rollout_log_ppl:np.float64(0.05315729657385759) - rollout_corr/log_ppl_diff:np.float64(0.0001371526817308677) - rollout_corr/log_ppl_abs_diff:np.float64(0.0009879193268886866) - rollout_corr/log_ppl_diff_max:np.float64(0.004697732627391815) - rollout_corr/log_ppl_diff_min:np.float64(-0.005460292100906372) - rollout_corr/ppl_ratio:np.float64(1.0001382497139275) - rollout_corr/chi2_token:np.float64(0.00032917875796556473) - rollout_corr/chi2_seq:np.float64(0.05339051899500191) - actor/pg_loss:np.float64(0.016300965010216828) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1328125) - actor/grad_norm:np.float64(0.1949380487203598) - perf/mfu/actor:np.float64(0.026599336089190795) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.62451553344727) - actor/lr:np.float64(1e-05) - training/global_step:62 - training/epoch:1 - critic/score/mean:0.7109375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7109375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0022602775134146214 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0022602775134146214 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:154.02734375 - response_length/max:451.0 - response_length/min:87.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:154.02734375 - response_length_non_aborted/max:451.0 - response_length_non_aborted/min:87.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:299.25 - prompt_length/max:662.0 - prompt_length/min:171.0 - 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:0.00013100355863571167 - timing_s/agent_loop/generate_sequences/min:np.float64(0.7049283906817436) - timing_s/agent_loop/generate_sequences/max:np.float64(3.4258072432130575) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.5876764539134456) - timing_s/agent_loop/tool_cal
|
||
|
|
step:63 - global_seqlen/min:13601 - global_seqlen/max:16717 - global_seqlen/minmax_diff:3116 - global_seqlen/balanced_min:15023 - global_seqlen/balanced_max:15025 - global_seqlen/mean:15024.0 - actor/entropy:0.10139713436365128 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.6351487040519714 - training/rollout_probs_diff_mean:0.0017283782362937927 - training/rollout_probs_diff_std:0.008464012295007706 - training/rollout_actor_probs_pearson_corr:0.9986895322799683 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.77734375 - self_distillation/correct_sample_fraction:0.58984375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.77734375 - rollout_corr/rollout_is_mean:1.000060796737671 - rollout_corr/rollout_is_ratio_fraction_high:4.262574657332152e-05 - rollout_corr/rollout_is_ratio_fraction_low:4.262574657332152e-05 - rollout_corr/rollout_is_max:2.134403705596924 - rollout_corr/rollout_is_min:0.22313041985034943 - rollout_corr/rollout_is_std:0.022056588903069496 - rollout_corr/rollout_is_eff_sample_size:0.999513981636632 - rollout_corr/rollout_is_seq_mean:1.0000665187835693 - rollout_corr/rollout_is_seq_std:0.001767620095051825 - rollout_corr/rollout_is_seq_max:1.009155035018921 - rollout_corr/rollout_is_seq_min:0.9960353970527649 - rollout_corr/rollout_is_seq_max_deviation:0.009155035018920898 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0906312982551754) - rollout_corr/training_log_ppl:np.float64(0.0831477354204253) - rollout_corr/kl:np.float64(0.00016323397627759917) - rollout_corr/k3_kl:np.float64(0.00022979614414300915) - rollout_corr/rollout_ppl:np.float64(1.090445393230766) - rollout_corr/rollout_log_ppl:np.float64(0.08298450080314979) - rollout_corr/log_ppl_diff:np.float64(0.00016323461727552058) - rollout_corr/log_ppl_abs_diff:np.float64(0.001246794446842614) - rollout_corr/log_ppl_diff_max:np.float64(0.005231320858001709) - rollout_corr/log_ppl_diff_min:np.float64(-0.006495680660009384) - rollout_corr/ppl_ratio:np.float64(1.0001647449098527) - rollout_corr/chi2_token:np.float64(0.0006031591910868883) - rollout_corr/chi2_seq:np.float64(0.16091938386671245) - actor/pg_loss:np.float64(0.017376098405740947) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.22265625) - actor/grad_norm:np.float64(0.17663174867630005) - perf/mfu/actor:np.float64(0.028010437778440596) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.17296600341797) - actor/lr:np.float64(1e-05) - training/global_step:63 - training/epoch:1 - critic/score/mean:0.58984375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.58984375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.007936380803585052 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.007936380803585052 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:183.28125 - response_length/max:581.0 - response_length/min:85.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:183.28125 - response_length_non_aborted/max:581.0 - response_length_non_aborted/min:85.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:286.21875 - prompt_length/max:532.0 - prompt_length/min:169.0 - 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:0.0001032017171382904 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0583508759737015) - timing_s/agent_loop/generate_sequences/max:np.float64(4.4486562460660934) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.98884266270033
|
||
|
|
step:64 - global_seqlen/min:11229 - global_seqlen/max:19243 - global_seqlen/minmax_diff:8014 - global_seqlen/balanced_min:15197 - global_seqlen/balanced_max:15276 - global_seqlen/mean:15207.0 - actor/entropy:0.06869005411863327 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3016374111175537 - training/rollout_probs_diff_mean:0.0014628589851781726 - training/rollout_probs_diff_std:0.008576503023505211 - training/rollout_actor_probs_pearson_corr:0.9980633854866028 - self_distillation/success_group_fraction:0.96875 - self_distillation/success_sample_fraction:0.96484375 - self_distillation/correct_sample_fraction:0.84375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.96484375 - rollout_corr/rollout_is_mean:0.9999962449073792 - rollout_corr/rollout_is_ratio_fraction_high:2.140777542081196e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00010703887528507039 - rollout_corr/rollout_is_max:2.513392686843872 - rollout_corr/rollout_is_min:0.4068211317062378 - rollout_corr/rollout_is_std:0.023297207430005074 - rollout_corr/rollout_is_eff_sample_size:0.9994575345331753 - rollout_corr/rollout_is_seq_mean:0.9999815225601196 - rollout_corr/rollout_is_seq_std:0.0019230464240536094 - rollout_corr/rollout_is_seq_max:1.0084824562072754 - rollout_corr/rollout_is_seq_min:0.9929937124252319 - rollout_corr/rollout_is_seq_max_deviation:0.00848245620727539 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0565863237716258) - rollout_corr/training_log_ppl:np.float64(0.05248735924487846) - rollout_corr/kl:np.float64(0.00030224999491512694) - rollout_corr/k3_kl:np.float64(0.0002837222421087544) - rollout_corr/rollout_ppl:np.float64(1.0562526867724955) - rollout_corr/rollout_log_ppl:np.float64(0.05218510861199377) - rollout_corr/log_ppl_diff:np.float64(0.000302250632884693) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012763329428793213) - rollout_corr/log_ppl_diff_max:np.float64(0.008936308324337006) - rollout_corr/log_ppl_diff_min:np.float64(-0.005065102130174637) - rollout_corr/ppl_ratio:np.float64(1.0003041788004339) - rollout_corr/chi2_token:np.float64(0.0005445803981274366) - rollout_corr/chi2_seq:np.float64(0.08466157596558332) - actor/pg_loss:np.float64(0.018894623313844505) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.03515625) - actor/grad_norm:np.float64(0.23259714245796204) - perf/mfu/actor:np.float64(0.02806062551050121) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(147.94005966186523) - actor/lr:np.float64(1e-05) - training/global_step:64 - training/epoch:1 - critic/score/mean:0.84375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.84375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.005651652812957764 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.005651652812957764 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:182.46875 - response_length/max:794.0 - response_length/min:88.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:182.46875 - response_length_non_aborted/max:794.0 - response_length_non_aborted/min:88.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:292.75 - prompt_length/max:495.0 - prompt_length/min:172.0 - 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:0.00010216794908046722 - timing_s/agent_loop/generate_sequences/min:np.float64(0.993412047624588) - timing_s/agent_loop/generate_sequences/max:np.float64(5.269853789359331) - timing_s/agent_loop/generate_sequences/mean:np.float64(2.0068536385442712) - tim
|
||
|
|
step:65 - global_seqlen/min:12358 - global_seqlen/max:18479 - global_seqlen/minmax_diff:6121 - global_seqlen/balanced_min:15052 - global_seqlen/balanced_max:15148 - global_seqlen/mean:15067.0 - actor/entropy:0.04862136393785477 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.19437427818775177 - training/rollout_probs_diff_mean:0.0011476365616545081 - training/rollout_probs_diff_std:0.007192918099462986 - training/rollout_actor_probs_pearson_corr:0.998099684715271 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - self_distillation/correct_sample_fraction:0.7578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8359375 - rollout_corr/rollout_is_mean:1.000012755393982 - rollout_corr/rollout_is_ratio_fraction_high:2.086114909616299e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.00010430574184283614 - rollout_corr/rollout_is_max:2.06850004196167 - rollout_corr/rollout_is_min:0.4432307779788971 - rollout_corr/rollout_is_std:0.019248440861701965 - rollout_corr/rollout_is_eff_sample_size:0.9996296347493171 - rollout_corr/rollout_is_seq_mean:1.0000423192977905 - rollout_corr/rollout_is_seq_std:0.0015931164380162954 - rollout_corr/rollout_is_seq_max:1.0080982446670532 - rollout_corr/rollout_is_seq_min:0.9924854040145874 - rollout_corr/rollout_is_seq_max_deviation:0.008098244667053223 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0366348451934755) - rollout_corr/training_log_ppl:np.float64(0.03463995937056552) - rollout_corr/kl:np.float64(0.00015224430039761927) - rollout_corr/k3_kl:np.float64(0.0001945765113010242) - rollout_corr/rollout_ppl:np.float64(1.0364593034610152) - rollout_corr/rollout_log_ppl:np.float64(0.03448771477081891) - rollout_corr/log_ppl_diff:np.float64(0.0001522445997466093) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010212431220679719) - rollout_corr/log_ppl_diff_max:np.float64(0.0086936354637146) - rollout_corr/log_ppl_diff_min:np.float64(-0.005649242550134659) - rollout_corr/ppl_ratio:np.float64(1.0001536114141345) - rollout_corr/chi2_token:np.float64(0.0004651371855288744) - rollout_corr/chi2_seq:np.float64(0.1004136900883168) - actor/pg_loss:np.float64(0.007932114620849973) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1640625) - actor/grad_norm:np.float64(0.20579096674919128) - perf/mfu/actor:np.float64(0.027797081377724797) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(147.72283601760864) - actor/lr:np.float64(1e-05) - training/global_step:65 - training/epoch:1 - critic/score/mean:0.7578125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7578125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0019635555800050497 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0019635555800050497 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:187.25 - response_length/max:1054.0 - response_length/min:105.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:187.25 - response_length_non_aborted/max:1054.0 - response_length_non_aborted/min:105.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:283.59375 - prompt_length/max:818.0 - prompt_length/min:165.0 - 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:0.00021395832300186157 - timing_s/agent_loop/generate_sequences/min:np.float64(1.3624698352068663) - timing_s/agent_loop/generate_sequences/max:np.float64(6.4785280879586935) - timing_s/agent_loop/generate_sequences/mean:np.float64(2.138984315584821)
|
||
|
|
step:66 - global_seqlen/min:13333 - global_seqlen/max:17466 - global_seqlen/minmax_diff:4133 - global_seqlen/balanced_min:15616 - global_seqlen/balanced_max:15647 - global_seqlen/mean:15624.625 - actor/entropy:0.07274921238422394 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.39762523770332336 - training/rollout_probs_diff_mean:0.0013556178892031312 - training/rollout_probs_diff_std:0.0072161005809903145 - training/rollout_actor_probs_pearson_corr:0.9987087249755859 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.77734375 - self_distillation/correct_sample_fraction:0.59375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.77734375 - rollout_corr/rollout_is_mean:1.000118613243103 - rollout_corr/rollout_is_ratio_fraction_high:3.945317803299986e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.2949154376983643 - rollout_corr/rollout_is_min:0.5686209797859192 - rollout_corr/rollout_is_std:0.020531050860881805 - rollout_corr/rollout_is_eff_sample_size:0.9995786535598055 - rollout_corr/rollout_is_seq_mean:1.0000622272491455 - rollout_corr/rollout_is_seq_std:0.0017089492175728083 - rollout_corr/rollout_is_seq_max:1.0099276304244995 - rollout_corr/rollout_is_seq_min:0.9940470457077026 - rollout_corr/rollout_is_seq_max_deviation:0.009927630424499512 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0622092648409307) - rollout_corr/training_log_ppl:np.float64(0.05809684956898309) - rollout_corr/kl:np.float64(0.00014326969245193055) - rollout_corr/k3_kl:np.float64(0.00020553644062880718) - rollout_corr/rollout_ppl:np.float64(1.0620549879968166) - rollout_corr/rollout_log_ppl:np.float64(0.05795357976206361) - rollout_corr/log_ppl_diff:np.float64(0.0001432698069194771) - rollout_corr/log_ppl_abs_diff:np.float64(0.0009604893442727302) - rollout_corr/log_ppl_diff_max:np.float64(0.006864520721137524) - rollout_corr/log_ppl_diff_min:np.float64(-0.006961427628993988) - rollout_corr/ppl_ratio:np.float64(1.000144557794556) - rollout_corr/chi2_token:np.float64(0.0005759422201663256) - rollout_corr/chi2_seq:np.float64(0.16348062525503337) - actor/pg_loss:np.float64(0.015369333535062424) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.22265625) - actor/grad_norm:np.float64(0.22882021963596344) - perf/mfu/actor:np.float64(0.029006857824807292) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(147.98411178588867) - actor/lr:np.float64(1e-05) - training/global_step:66 - training/epoch:1 - 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:-0.002093484392389655 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.002093484392389655 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:198.01953125 - response_length/max:871.0 - response_length/min:85.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:198.01953125 - response_length_non_aborted/max:871.0 - response_length_non_aborted/min:85.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:290.25 - prompt_length/max:617.0 - prompt_length/min:184.0 - 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:8.31354409456253e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0428329799324274) - timing_s/agent_loop/generate_sequences/max:np.float64(5.64964609593153) - timing_s/agent_loop/generate_sequences/mean:np.float64(2.1026814351862413) - timing_s/ag
|
||
|
|
step:67 - global_seqlen/min:10760 - global_seqlen/max:17016 - global_seqlen/minmax_diff:6256 - global_seqlen/balanced_min:14319 - global_seqlen/balanced_max:14323 - global_seqlen/mean:14321.5 - actor/entropy:0.04746159166097641 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23051795363426208 - training/rollout_probs_diff_mean:0.0010486101964488626 - training/rollout_probs_diff_std:0.007005594205111265 - training/rollout_actor_probs_pearson_corr:0.9982449412345886 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.76953125 - self_distillation/correct_sample_fraction:0.59765625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.76953125 - rollout_corr/rollout_is_mean:1.0000702142715454 - rollout_corr/rollout_is_ratio_fraction_high:5.0175614887848496e-05 - rollout_corr/rollout_is_ratio_fraction_low:7.526342233177274e-05 - rollout_corr/rollout_is_max:2.0498805046081543 - rollout_corr/rollout_is_min:0.428154319524765 - rollout_corr/rollout_is_std:0.020623741671442986 - rollout_corr/rollout_is_eff_sample_size:0.9995748420910143 - rollout_corr/rollout_is_seq_mean:1.0000723600387573 - rollout_corr/rollout_is_seq_std:0.0016946522518992424 - rollout_corr/rollout_is_seq_max:1.0085837841033936 - rollout_corr/rollout_is_seq_min:0.9943454265594482 - rollout_corr/rollout_is_seq_max_deviation:0.008583784103393555 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.038079034537077) - rollout_corr/training_log_ppl:np.float64(0.036100168052712434) - rollout_corr/kl:np.float64(0.00011544184226575105) - rollout_corr/k3_kl:np.float64(0.00018780643174386939) - rollout_corr/rollout_ppl:np.float64(1.037946602795273) - rollout_corr/rollout_log_ppl:np.float64(0.035984725436321696) - rollout_corr/log_ppl_diff:np.float64(0.00011544261639073738) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010177288714317) - rollout_corr/log_ppl_diff_max:np.float64(0.007095633074641228) - rollout_corr/log_ppl_diff_min:np.float64(-0.006181836128234863) - rollout_corr/ppl_ratio:np.float64(1.0001168274320662) - rollout_corr/chi2_token:np.float64(0.0005425922572612762) - rollout_corr/chi2_seq:np.float64(0.1015120605006814) - actor/pg_loss:np.float64(0.012161348286099383) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.23046875) - actor/grad_norm:np.float64(0.1856652945280075) - perf/mfu/actor:np.float64(0.026166298878302647) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(147.85752487182617) - actor/lr:np.float64(1e-05) - training/global_step:67 - 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:-0.0038823382928967476 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0038823382928967476 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:155.703125 - response_length/max:653.0 - response_length/min:90.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:155.703125 - response_length_non_aborted/max:653.0 - response_length_non_aborted/min:90.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:291.84375 - prompt_length/max:576.0 - prompt_length/min:168.0 - 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:5.5110082030296326e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1322892885655165) - timing_s/agent_loop/generate_sequences/max:np.float64(4.47765926271677) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.814805
|
||
|
|
step:68 - global_seqlen/min:11895 - global_seqlen/max:19288 - global_seqlen/minmax_diff:7393 - global_seqlen/balanced_min:15029 - global_seqlen/balanced_max:15031 - global_seqlen/mean:15030.375 - actor/entropy:0.054700303822755814 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2310505509376526 - training/rollout_probs_diff_mean:0.0010520927608013153 - training/rollout_probs_diff_std:0.006294276565313339 - training/rollout_actor_probs_pearson_corr:0.9987224340438843 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.765625 - self_distillation/correct_sample_fraction:0.609375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.765625 - rollout_corr/rollout_is_mean:1.0000325441360474 - rollout_corr/rollout_is_ratio_fraction_high:5.057273665443063e-05 - rollout_corr/rollout_is_ratio_fraction_low:5.057273665443063e-05 - rollout_corr/rollout_is_max:2.0968377590179443 - rollout_corr/rollout_is_min:0.4065501391887665 - rollout_corr/rollout_is_std:0.017810557037591934 - rollout_corr/rollout_is_eff_sample_size:0.9996830038082445 - rollout_corr/rollout_is_seq_mean:1.0000410079956055 - rollout_corr/rollout_is_seq_std:0.001466538175009191 - rollout_corr/rollout_is_seq_max:1.0071442127227783 - rollout_corr/rollout_is_seq_min:0.9932287931442261 - rollout_corr/rollout_is_seq_max_deviation:0.00714421272277832 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0410470603965223) - rollout_corr/training_log_ppl:np.float64(0.038657502603683724) - rollout_corr/kl:np.float64(9.617471553734214e-05) - rollout_corr/k3_kl:np.float64(0.00013721287146261198) - rollout_corr/rollout_ppl:np.float64(1.0409434824250638) - rollout_corr/rollout_log_ppl:np.float64(0.03856132756040864) - rollout_corr/log_ppl_diff:np.float64(9.617504327508186e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.0008498393303497664) - rollout_corr/log_ppl_diff_max:np.float64(0.009416066110134125) - rollout_corr/log_ppl_diff_min:np.float64(-0.005098434165120125) - rollout_corr/ppl_ratio:np.float64(1.0000972892157733) - rollout_corr/chi2_token:np.float64(0.0003639685455709696) - rollout_corr/chi2_seq:np.float64(0.08358758222311735) - actor/pg_loss:np.float64(0.008533639485253985) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.234375) - actor/grad_norm:np.float64(0.17609483003616333) - perf/mfu/actor:np.float64(0.02794686230135153) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(147.89475631713867) - actor/lr:np.float64(1e-05) - training/global_step:68 - training/epoch:1 - critic/score/mean:0.609375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.609375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0014634485123679042 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.0014634485123679042 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:154.48046875 - response_length/max:599.0 - response_length/min:87.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:154.48046875 - response_length_non_aborted/max:599.0 - response_length_non_aborted/min:87.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:315.21875 - prompt_length/max:704.0 - prompt_length/min:167.0 - 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:0.00013556890189647675 - timing_s/agent_loop/generate_sequences/min:np.float64(1.094132350757718) - timing_s/agent_loop/generate_sequences/max:np.float64(4.072329668328166) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.7255080375107
|
||
|
|
step:69 - global_seqlen/min:12317 - global_seqlen/max:22993 - global_seqlen/minmax_diff:10676 - global_seqlen/balanced_min:15501 - global_seqlen/balanced_max:15724 - global_seqlen/mean:15690.25 - actor/entropy:0.060017649084329605 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2773001194000244 - training/rollout_probs_diff_mean:0.0012825635494664311 - training/rollout_probs_diff_std:0.007520930841565132 - training/rollout_actor_probs_pearson_corr:0.9982960820198059 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.7734375 - self_distillation/correct_sample_fraction:0.60546875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7734375 - rollout_corr/rollout_is_mean:1.0001111030578613 - rollout_corr/rollout_is_ratio_fraction_high:2.1640336854034103e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.6848840713500977 - rollout_corr/rollout_is_min:0.6432498693466187 - rollout_corr/rollout_is_std:0.01939343474805355 - rollout_corr/rollout_is_eff_sample_size:0.9996241552131722 - rollout_corr/rollout_is_seq_mean:1.0001531839370728 - rollout_corr/rollout_is_seq_std:0.001562311314046383 - rollout_corr/rollout_is_seq_max:1.013192892074585 - rollout_corr/rollout_is_seq_min:0.9963599443435669 - rollout_corr/rollout_is_seq_max_deviation:0.013192892074584961 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.042771234177053) - rollout_corr/training_log_ppl:np.float64(0.04025752223027723) - rollout_corr/kl:np.float64(2.0769094894179663e-05) - rollout_corr/k3_kl:np.float64(0.00017395197071124607) - rollout_corr/rollout_ppl:np.float64(1.042743084486574) - rollout_corr/rollout_log_ppl:np.float64(0.04023675218580891) - rollout_corr/log_ppl_diff:np.float64(2.0770044468321203e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.0009357155959186514) - rollout_corr/log_ppl_diff_max:np.float64(0.004242569208145142) - rollout_corr/log_ppl_diff_min:np.float64(-0.007939537987112999) - rollout_corr/ppl_ratio:np.float64(1.0000217508058995) - rollout_corr/chi2_token:np.float64(0.0007122822571545839) - rollout_corr/chi2_seq:np.float64(0.1271292066667229) - actor/pg_loss:np.float64(0.01221263497283509) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.2265625) - actor/grad_norm:np.float64(0.19111807644367218) - perf/mfu/actor:np.float64(0.02864372518669452) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(147.77561950683594) - actor/lr:np.float64(1e-05) - training/global_step:69 - training/epoch:1 - 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:0.00021910842042416334 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00021910842042416334 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:180.5078125 - response_length/max:1081.0 - response_length/min:84.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:180.5078125 - response_length_non_aborted/max:1081.0 - response_length_non_aborted/min:84.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:309.8125 - prompt_length/max:597.0 - prompt_length/min:168.0 - 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:0.00014573149383068085 - timing_s/agent_loop/generate_sequences/min:np.float64(0.993093540892005) - timing_s/agent_loop/generate_sequences/max:np.float64(6.451122162863612) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.905593388932175) - 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': 70}
|
||
|
|
validation generation end
|
||
|
|
[prompt] system
|
||
|
|
|
||
|
|
Given a question and four options, please select the right answer. Respond in the following format:
|
||
|
|
<reasoning>
|
||
|
|
...
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
...
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
user
|
||
|
|
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||
|
|
|
||
|
|
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||
|
|
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||
|
|
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||
|
|
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||
|
|
Please reason step by step.
|
||
|
|
assistant
|
||
|
|
<think>
|
||
|
|
|
||
|
|
</think>
|
||
|
|
|
||
|
|
|
||
|
|
[response] <reasoning>
|
||
|
|
The product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" contains a nitrogen-containing group (N), a chlorine atom (Cl), and a fluorine atom (F). Option A contains a nitrogen-containing group (N), a chlorine atom (Cl), and a fluorine atom (F), which matches the product's structure. The other options either lack the nitrogen-containing group or do not match the product's structure. Therefore, the correct answer is Option A.
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
[ground_truth] A
|
||
|
|
[score] 1.0
|
||
|
|
[acc] 1.0
|
||
|
|
[pred] A
|
||
|
|
[incorrect_format] 1
|
||
|
|
[feedback]
|
||
|
|
local_global_step_folder: /workspace/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-SDPO_TR-Qwen-Qwen3-4B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_70
|
||
|
|
step:70 - global_seqlen/min:10503 - global_seqlen/max:17198 - global_seqlen/minmax_diff:6695 - global_seqlen/balanced_min:14864 - global_seqlen/balanced_max:15530 - global_seqlen/mean:15075.25 - actor/entropy:0.049284253269433975 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.29323524236679077 - training/rollout_probs_diff_mean:0.001126911141909659 - training/rollout_probs_diff_std:0.007642928510904312 - training/rollout_actor_probs_pearson_corr:0.9979100823402405 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.828125 - self_distillation/correct_sample_fraction:0.65625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.828125 - rollout_corr/rollout_is_mean:0.9999964833259583 - rollout_corr/rollout_is_ratio_fraction_high:6.778435636078939e-05 - rollout_corr/rollout_is_ratio_fraction_low:6.778435636078939e-05 - rollout_corr/rollout_is_max:2.0903313159942627 - rollout_corr/rollout_is_min:0.4566020965576172 - rollout_corr/rollout_is_std:0.020981894806027412 - rollout_corr/rollout_is_eff_sample_size:0.9995600729239899 - rollout_corr/rollout_is_seq_mean:1.000002145767212 - rollout_corr/rollout_is_seq_std:0.0017738351598381996 - rollout_corr/rollout_is_seq_max:1.0079503059387207 - rollout_corr/rollout_is_seq_min:0.994708776473999 - rollout_corr/rollout_is_seq_max_deviation:0.007950305938720703 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0394724453799427) - rollout_corr/training_log_ppl:np.float64(0.037417455858438586) - rollout_corr/kl:np.float64(0.00022468566553501476) - rollout_corr/k3_kl:np.float64(0.0002268687162885225) - rollout_corr/rollout_ppl:np.float64(1.0392303792759776) - rollout_corr/rollout_log_ppl:np.float64(0.03719276980552699) - rollout_corr/log_ppl_diff:np.float64(0.0002246860529115935) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010704495566056238) - rollout_corr/log_ppl_diff_max:np.float64(0.0076381489634513855) - rollout_corr/log_ppl_diff_min:np.float64(-0.0054670460522174835) - rollout_corr/ppl_ratio:np.float64(1.0002261234913021) - rollout_corr/chi2_token:np.float64(0.0004954265896230936) - rollout_corr/chi2_seq:np.float64(0.05916072824038565) - actor/pg_loss:np.float64(0.013035437431557995) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.171875) - actor/grad_norm:np.float64(0.2884591519832611) - perf/mfu/actor:np.float64(0.027443551933430947) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.19314193725586) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.6827380952380953) - val-aux/sciknoweval/reward/std@16:np.float64(0.06520315552662033) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7081333333333333) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.05844871318325079) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6572714285714285) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.0512738099386681) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.6828333333333334) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.06482636784662292) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7326714285714285) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.047778680098007176) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6394190476190477) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.03680110141309759) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.6814714285714285) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.05070243225622103) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7522714285714286) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.03427190427518333) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6256857
|
||
|
|
step:71 - global_seqlen/min:11474 - global_seqlen/max:18344 - global_seqlen/minmax_diff:6870 - global_seqlen/balanced_min:15452 - global_seqlen/balanced_max:15535 - global_seqlen/mean:15465.75 - actor/entropy:0.048588912934064865 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4863182306289673 - training/rollout_probs_diff_mean:0.000986974104307592 - training/rollout_probs_diff_std:0.006218041758984327 - training/rollout_actor_probs_pearson_corr:0.9985578060150146 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71484375 - self_distillation/correct_sample_fraction:0.59375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.71484375 - rollout_corr/rollout_is_mean:0.9999036192893982 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:6.476963608292863e-05 - rollout_corr/rollout_is_max:1.5526795387268066 - rollout_corr/rollout_is_min:0.36988571286201477 - rollout_corr/rollout_is_std:0.01609477587044239 - rollout_corr/rollout_is_eff_sample_size:0.9997409061515363 - rollout_corr/rollout_is_seq_mean:0.999914824962616 - rollout_corr/rollout_is_seq_std:0.001312144217081368 - rollout_corr/rollout_is_seq_max:1.0045524835586548 - rollout_corr/rollout_is_seq_min:0.9943608045578003 - rollout_corr/rollout_is_seq_max_deviation:0.005639195442199707 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0364314783364534) - rollout_corr/training_log_ppl:np.float64(0.034763937364914455) - rollout_corr/kl:np.float64(0.00022432199393659924) - rollout_corr/k3_kl:np.float64(0.00013917502211987652) - rollout_corr/rollout_ppl:np.float64(1.036194120068103) - rollout_corr/rollout_log_ppl:np.float64(0.034539614922746154) - rollout_corr/log_ppl_diff:np.float64(0.000224322442168301) - rollout_corr/log_ppl_abs_diff:np.float64(0.0008982900929943582) - rollout_corr/log_ppl_diff_max:np.float64(0.00732392817735672) - rollout_corr/log_ppl_diff_min:np.float64(-0.004114896059036255) - rollout_corr/ppl_ratio:np.float64(1.0002253497950733) - rollout_corr/chi2_token:np.float64(8.037476800382137e-05) - rollout_corr/chi2_seq:np.float64(0.030314008239656687) - actor/pg_loss:np.float64(0.008997618729701173) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.28515625) - actor/grad_norm:np.float64(0.16296449303627014) - perf/mfu/actor:np.float64(0.028347251708969765) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.80670166015625) - actor/lr:np.float64(1e-05) - training/global_step:71 - training/epoch:1 - 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:-0.0037782287690788507 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0037782287690788507 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:180.9296875 - response_length/max:1151.0 - response_length/min:85.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:180.9296875 - response_length_non_aborted/max:1151.0 - response_length_non_aborted/min:85.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:302.375 - prompt_length/max:666.0 - prompt_length/min:169.0 - 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:9.109824895858765e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0407183561474085) - timing_s/agent_loop/generate_sequences/max:np.float64(6.6910761427134275) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.8566879731442896) - timing_s
|
||
|
|
step:72 - global_seqlen/min:10873 - global_seqlen/max:15786 - global_seqlen/minmax_diff:4913 - global_seqlen/balanced_min:13562 - global_seqlen/balanced_max:13563 - global_seqlen/mean:13562.25 - actor/entropy:0.03674636036157608 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.30163803696632385 - training/rollout_probs_diff_mean:0.000952803180553019 - training/rollout_probs_diff_std:0.007416649255901575 - training/rollout_actor_probs_pearson_corr:0.997348964214325 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - self_distillation/correct_sample_fraction:0.70703125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83203125 - rollout_corr/rollout_is_mean:0.9999909996986389 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:5.703530405298807e-05 - rollout_corr/rollout_is_max:1.798954725265503 - rollout_corr/rollout_is_min:0.3409341275691986 - rollout_corr/rollout_is_std:0.017625533044338226 - rollout_corr/rollout_is_eff_sample_size:0.9996894370707265 - rollout_corr/rollout_is_seq_mean:1.0000380277633667 - rollout_corr/rollout_is_seq_std:0.0016533611342310905 - rollout_corr/rollout_is_seq_max:1.0073695182800293 - rollout_corr/rollout_is_seq_min:0.993415117263794 - rollout_corr/rollout_is_seq_max_deviation:0.007369518280029297 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0312063470482826) - rollout_corr/training_log_ppl:np.float64(0.029597907576885518) - rollout_corr/kl:np.float64(0.00012548301245030036) - rollout_corr/k3_kl:np.float64(0.0001635165864383077) - rollout_corr/rollout_ppl:np.float64(1.0310773248784244) - rollout_corr/rollout_log_ppl:np.float64(0.02947242450879628) - rollout_corr/log_ppl_diff:np.float64(0.00012548306808923826) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010587770540695374) - rollout_corr/log_ppl_diff_max:np.float64(0.009746916592121124) - rollout_corr/log_ppl_diff_min:np.float64(-0.005916215479373932) - rollout_corr/ppl_ratio:np.float64(1.0001269716303796) - rollout_corr/chi2_token:np.float64(0.00038446555845439434) - rollout_corr/chi2_seq:np.float64(0.06596083450131118) - actor/pg_loss:np.float64(0.010816079263030076) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.16796875) - actor/grad_norm:np.float64(0.2558634281158447) - perf/mfu/actor:np.float64(0.025002151647709546) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.53983306884766) - actor/lr:np.float64(1e-05) - training/global_step:72 - training/epoch:1 - 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:0.0009339530952274799 - critic/advantages/max:0.875 - critic/advantages/min:-0.625 - critic/returns/mean:0.0009339530952274799 - critic/returns/max:0.875 - critic/returns/min:-0.625 - response_length/mean:136.9765625 - response_length/max:382.0 - response_length/min:93.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:136.9765625 - response_length_non_aborted/max:382.0 - response_length_non_aborted/min:93.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:286.84375 - prompt_length/max:620.0 - prompt_length/min:172.0 - 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:4.579685628414154e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0573269743472338) - timing_s/agent_loop/generate_sequences/max:np.float64(3.01920011267066) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.5792225186014548) - tim
|
||
|
|
step:73 - global_seqlen/min:12868 - global_seqlen/max:16390 - global_seqlen/minmax_diff:3522 - global_seqlen/balanced_min:14207 - global_seqlen/balanced_max:14415 - global_seqlen/mean:14233.25 - actor/entropy:0.05116261914372444 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24135515093803406 - training/rollout_probs_diff_mean:0.0010690698400139809 - training/rollout_probs_diff_std:0.006795805878937244 - training/rollout_actor_probs_pearson_corr:0.9983893036842346 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80078125 - self_distillation/correct_sample_fraction:0.61328125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.80078125 - rollout_corr/rollout_is_mean:1.0000085830688477 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.3736054572509602e-05 - rollout_corr/rollout_is_max:1.5512640476226807 - rollout_corr/rollout_is_min:0.40586087107658386 - rollout_corr/rollout_is_std:0.01709681749343872 - rollout_corr/rollout_is_eff_sample_size:0.9997077842370332 - rollout_corr/rollout_is_seq_mean:1.000005841255188 - rollout_corr/rollout_is_seq_std:0.0014473996125161648 - rollout_corr/rollout_is_seq_max:1.0049885511398315 - rollout_corr/rollout_is_seq_min:0.9945465326309204 - rollout_corr/rollout_is_seq_max_deviation:0.00545346736907959 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.043588503729552) - rollout_corr/training_log_ppl:np.float64(0.041208510853721236) - rollout_corr/kl:np.float64(0.00014402579568205398) - rollout_corr/k3_kl:np.float64(0.00014985975238270216) - rollout_corr/rollout_ppl:np.float64(1.043436171952635) - rollout_corr/rollout_log_ppl:np.float64(0.04106448414984243) - rollout_corr/log_ppl_diff:np.float64(0.0001440267038788079) - rollout_corr/log_ppl_abs_diff:np.float64(0.0009985639923115741) - rollout_corr/log_ppl_diff_max:np.float64(0.0068887826055288315) - rollout_corr/log_ppl_diff_min:np.float64(-0.00456034392118454) - rollout_corr/ppl_ratio:np.float64(1.0001451829448342) - rollout_corr/chi2_token:np.float64(0.00029682903550565243) - rollout_corr/chi2_seq:np.float64(0.050184973049908876) - actor/pg_loss:np.float64(0.010764306774603938) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.19921875) - actor/grad_norm:np.float64(0.2189781665802002) - perf/mfu/actor:np.float64(0.026349392448466562) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.35952758789062) - actor/lr:np.float64(1e-05) - training/global_step:73 - 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:-0.010595181956887245 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.010595181956887245 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:164.5703125 - response_length/max:811.0 - response_length/min:86.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:164.5703125 - response_length_non_aborted/max:811.0 - response_length_non_aborted/min:86.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:280.21875 - prompt_length/max:557.0 - prompt_length/min:170.0 - 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:0.00010772049427032471 - timing_s/agent_loop/generate_sequences/min:np.float64(0.8573889192193747) - timing_s/agent_loop/generate_sequences/max:np.float64(5.1699669398367405) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.746066306695866) -
|
||
|
|
step:74 - global_seqlen/min:11397 - global_seqlen/max:16658 - global_seqlen/minmax_diff:5261 - global_seqlen/balanced_min:13498 - global_seqlen/balanced_max:13500 - global_seqlen/mean:13499.0 - actor/entropy:0.051137976348400116 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1533067226409912 - training/rollout_probs_diff_mean:0.0009283441468141973 - training/rollout_probs_diff_std:0.0055831256322562695 - training/rollout_actor_probs_pearson_corr:0.9989195466041565 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80859375 - self_distillation/correct_sample_fraction:0.73046875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.80859375 - rollout_corr/rollout_is_mean:1.0000120401382446 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6356412172317505 - rollout_corr/rollout_is_min:0.55051189661026 - rollout_corr/rollout_is_std:0.015889784321188927 - rollout_corr/rollout_is_eff_sample_size:0.9997475784574541 - rollout_corr/rollout_is_seq_mean:1.0000572204589844 - rollout_corr/rollout_is_seq_std:0.001312208827584982 - rollout_corr/rollout_is_seq_max:1.0048962831497192 - rollout_corr/rollout_is_seq_min:0.9930908679962158 - rollout_corr/rollout_is_seq_max_deviation:0.00690913200378418 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0422634701244533) - rollout_corr/training_log_ppl:np.float64(0.039729325241637525) - rollout_corr/kl:np.float64(5.678095236544278e-05) - rollout_corr/k3_kl:np.float64(0.000114079510565551) - rollout_corr/rollout_ppl:np.float64(1.0421917419880629) - rollout_corr/rollout_log_ppl:np.float64(0.03967254392695452) - rollout_corr/log_ppl_diff:np.float64(5.678131468300762e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.000874444027644472) - rollout_corr/log_ppl_diff_max:np.float64(0.00868067517876625) - rollout_corr/log_ppl_diff_min:np.float64(-0.004446707665920258) - rollout_corr/ppl_ratio:np.float64(1.0000577236060053) - rollout_corr/chi2_token:np.float64(0.0003412545192986727) - rollout_corr/chi2_seq:np.float64(0.04243674594908953) - actor/pg_loss:np.float64(0.011527882362230457) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.19140625) - actor/grad_norm:np.float64(0.21671552956104279) - perf/mfu/actor:np.float64(0.024905560319006804) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.24897003173828) - actor/lr:np.float64(1e-05) - training/global_step:74 - training/epoch:1 - critic/score/mean:0.73046875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.73046875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00389473675750196 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.00389473675750196 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:148.4375 - response_length/max:477.0 - response_length/min:85.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:148.4375 - response_length_non_aborted/max:477.0 - response_length_non_aborted/min:85.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:273.40625 - prompt_length/max:665.0 - prompt_length/min:162.0 - 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:0.00010783970355987549 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9932483080774546) - timing_s/agent_loop/generate_sequences/max:np.float64(3.5065592229366302) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.6443017373021576) - timing_s/agent_loop/tool_calls/min:n
|
||
|
|
step:75 - global_seqlen/min:11106 - global_seqlen/max:18229 - global_seqlen/minmax_diff:7123 - global_seqlen/balanced_min:14844 - global_seqlen/balanced_max:14850 - global_seqlen/mean:14848.375 - actor/entropy:0.028360191732645035 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3014865517616272 - training/rollout_probs_diff_mean:0.0009677539928816259 - training/rollout_probs_diff_std:0.008439409546554089 - training/rollout_actor_probs_pearson_corr:0.9956839084625244 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80859375 - self_distillation/correct_sample_fraction:0.71484375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.80859375 - rollout_corr/rollout_is_mean:0.9999974370002747 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.00010376942373113707 - rollout_corr/rollout_is_max:1.9569276571273804 - rollout_corr/rollout_is_min:0.44322967529296875 - rollout_corr/rollout_is_std:0.020469993352890015 - rollout_corr/rollout_is_eff_sample_size:0.9995810357425605 - rollout_corr/rollout_is_seq_mean:1.0000098943710327 - rollout_corr/rollout_is_seq_std:0.001647818717174232 - rollout_corr/rollout_is_seq_max:1.0071107149124146 - rollout_corr/rollout_is_seq_min:0.9918871521949768 - rollout_corr/rollout_is_seq_max_deviation:0.008112847805023193 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0217003584839404) - rollout_corr/training_log_ppl:np.float64(0.020927514192749186) - rollout_corr/kl:np.float64(0.0002125317863110876) - rollout_corr/k3_kl:np.float64(0.00022249929528938097) - rollout_corr/rollout_ppl:np.float64(1.021478638984263) - rollout_corr/rollout_log_ppl:np.float64(0.020714982148412275) - rollout_corr/log_ppl_diff:np.float64(0.00021253204433691053) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011063803567807895) - rollout_corr/log_ppl_diff_max:np.float64(0.011423623189330101) - rollout_corr/log_ppl_diff_min:np.float64(-0.004935091361403465) - rollout_corr/ppl_ratio:np.float64(1.0002141240984201) - rollout_corr/chi2_token:np.float64(0.0004600733518600464) - rollout_corr/chi2_seq:np.float64(0.045901605393737555) - actor/pg_loss:np.float64(0.0076410405750948485) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.19140625) - actor/grad_norm:np.float64(0.2190471887588501) - perf/mfu/actor:np.float64(0.027442665995749377) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.1978302001953) - actor/lr:np.float64(1e-05) - training/global_step:75 - training/epoch:1 - critic/score/mean:0.71484375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.71484375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:7.134147745091468e-05 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:7.134147745091468e-05 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:150.57421875 - response_length/max:686.0 - response_length/min:88.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:150.57421875 - response_length_non_aborted/max:686.0 - response_length_non_aborted/min:88.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:313.4375 - prompt_length/max:608.0 - prompt_length/min:169.0 - 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:2.9299408197402954e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0843685306608677) - timing_s/agent_loop/generate_sequences/max:np.float64(4.162685789167881) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.722378140169894
|
||
|
|
step:76 - global_seqlen/min:13749 - global_seqlen/max:17028 - global_seqlen/minmax_diff:3279 - global_seqlen/balanced_min:15054 - global_seqlen/balanced_max:15055 - global_seqlen/mean:15054.625 - actor/entropy:0.0637604296207428 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3535170555114746 - training/rollout_probs_diff_mean:0.0011947648599743843 - training/rollout_probs_diff_std:0.007098553702235222 - training/rollout_actor_probs_pearson_corr:0.9986104965209961 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.7421875 - self_distillation/correct_sample_fraction:0.5625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7421875 - rollout_corr/rollout_is_mean:1.0000269412994385 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:4.228240504744463e-05 - rollout_corr/rollout_is_max:1.936368465423584 - rollout_corr/rollout_is_min:0.3715561330318451 - rollout_corr/rollout_is_std:0.018253492191433907 - rollout_corr/rollout_is_eff_sample_size:0.9996669210142809 - rollout_corr/rollout_is_seq_mean:1.0000336170196533 - rollout_corr/rollout_is_seq_std:0.0015557054430246353 - rollout_corr/rollout_is_seq_max:1.006518006324768 - rollout_corr/rollout_is_seq_min:0.9941936135292053 - rollout_corr/rollout_is_seq_max_deviation:0.006518006324768066 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0450231824070215) - rollout_corr/training_log_ppl:np.float64(0.04246747880779367) - rollout_corr/kl:np.float64(0.00014067312969157086) - rollout_corr/k3_kl:np.float64(0.00017436211087324413) - rollout_corr/rollout_ppl:np.float64(1.0448668636381626) - rollout_corr/rollout_log_ppl:np.float64(0.04232680522505916) - rollout_corr/log_ppl_diff:np.float64(0.00014067358273450736) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010677200666009412) - rollout_corr/log_ppl_diff_max:np.float64(0.007057949900627136) - rollout_corr/log_ppl_diff_min:np.float64(-0.005887016654014587) - rollout_corr/ppl_ratio:np.float64(1.0001420124899596) - rollout_corr/chi2_token:np.float64(0.00040210457518696785) - rollout_corr/chi2_seq:np.float64(0.06287222681567073) - actor/pg_loss:np.float64(0.009993368453478979) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.2578125) - actor/grad_norm:np.float64(0.2138492614030838) - perf/mfu/actor:np.float64(0.02780775031245608) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.2149429321289) - actor/lr:np.float64(1e-05) - training/global_step:76 - training/epoch:1 - critic/score/mean:0.5625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.005383078474551439 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.005383078474551439 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:184.76953125 - response_length/max:593.0 - response_length/min:87.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:184.76953125 - response_length_non_aborted/max:593.0 - response_length_non_aborted/min:87.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:285.6875 - prompt_length/max:526.0 - prompt_length/min:182.0 - 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:0.00015115924179553986 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9865694157779217) - timing_s/agent_loop/generate_sequences/max:np.float64(4.408676564693451) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.9008617323852377) - timing_s/agent_loop/t
|
||
|
|
step:77 - global_seqlen/min:13001 - global_seqlen/max:17194 - global_seqlen/minmax_diff:4193 - global_seqlen/balanced_min:14792 - global_seqlen/balanced_max:14951 - global_seqlen/mean:14817.0 - actor/entropy:0.03764856234192848 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23105892539024353 - training/rollout_probs_diff_mean:0.000981298042461276 - training/rollout_probs_diff_std:0.007331465370953083 - training/rollout_actor_probs_pearson_corr:0.9976370930671692 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.828125 - self_distillation/correct_sample_fraction:0.63671875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.828125 - rollout_corr/rollout_is_mean:0.9999348521232605 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.8513847589492798 - rollout_corr/rollout_is_min:0.5378821492195129 - rollout_corr/rollout_is_std:0.018528969958424568 - rollout_corr/rollout_is_eff_sample_size:0.9996567950761533 - rollout_corr/rollout_is_seq_mean:0.9999136924743652 - rollout_corr/rollout_is_seq_std:0.0014632911188527942 - rollout_corr/rollout_is_seq_max:1.0068610906600952 - rollout_corr/rollout_is_seq_min:0.9950417280197144 - rollout_corr/rollout_is_seq_max_deviation:0.006861090660095215 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0321186697110534) - rollout_corr/training_log_ppl:np.float64(0.030197202008565682) - rollout_corr/kl:np.float64(0.00023827231725626064) - rollout_corr/k3_kl:np.float64(0.00015201022574169373) - rollout_corr/rollout_ppl:np.float64(1.0318722985684872) - rollout_corr/rollout_log_ppl:np.float64(0.02995892934859512) - rollout_corr/log_ppl_diff:np.float64(0.00023827265997056202) - rollout_corr/log_ppl_abs_diff:np.float64(0.000967306014189262) - rollout_corr/log_ppl_diff_max:np.float64(0.0061843376606702805) - rollout_corr/log_ppl_diff_min:np.float64(-0.004977274686098099) - rollout_corr/ppl_ratio:np.float64(1.000239381333813) - rollout_corr/chi2_token:np.float64(0.0001320859882980585) - rollout_corr/chi2_seq:np.float64(0.07500208495184779) - actor/pg_loss:np.float64(0.009187513203812614) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.171875) - actor/grad_norm:np.float64(0.24896493554115295) - perf/mfu/actor:np.float64(0.02751012093943027) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.1029815673828) - actor/lr:np.float64(1e-05) - training/global_step:77 - training/epoch:1 - critic/score/mean:0.63671875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.63671875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.007862687110900879 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.007862687110900879 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:145.9375 - response_length/max:922.0 - response_length/min:95.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:145.9375 - response_length_non_aborted/max:922.0 - response_length_non_aborted/min:95.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:317.09375 - prompt_length/max:716.0 - prompt_length/min:173.0 - 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:0.00015098601579666138 - timing_s/agent_loop/generate_sequences/min:np.float64(0.7721793595701456) - timing_s/agent_loop/generate_sequences/max:np.float64(5.443623935803771) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.4915188666127506) - timing_s/agent_loop/tool_calls/
|
||
|
|
step:78 - global_seqlen/min:11907 - global_seqlen/max:15199 - global_seqlen/minmax_diff:3292 - global_seqlen/balanced_min:13946 - global_seqlen/balanced_max:13947 - global_seqlen/mean:13946.5 - actor/entropy:0.04153583571314812 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1943935751914978 - training/rollout_probs_diff_mean:0.0008782729273661971 - training/rollout_probs_diff_std:0.0059511493891477585 - training/rollout_actor_probs_pearson_corr:0.9984843730926514 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8046875 - self_distillation/correct_sample_fraction:0.69140625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8046875 - rollout_corr/rollout_is_mean:1.000045895576477 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.4997500077006407e-05 - rollout_corr/rollout_is_max:1.6279478073120117 - rollout_corr/rollout_is_min:0.4838121831417084 - rollout_corr/rollout_is_std:0.01630815491080284 - rollout_corr/rollout_is_eff_sample_size:0.9997341147887595 - rollout_corr/rollout_is_seq_mean:1.0000261068344116 - rollout_corr/rollout_is_seq_std:0.0012171355774626136 - rollout_corr/rollout_is_seq_max:1.0046321153640747 - rollout_corr/rollout_is_seq_min:0.9955993890762329 - rollout_corr/rollout_is_seq_max_deviation:0.004632115364074707 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0314596453681588) - rollout_corr/training_log_ppl:np.float64(0.029678920399192066) - rollout_corr/kl:np.float64(9.43284891232743e-05) - rollout_corr/k3_kl:np.float64(0.0001203459440544544) - rollout_corr/rollout_ppl:np.float64(1.0313554625026882) - rollout_corr/rollout_log_ppl:np.float64(0.029584591509888014) - rollout_corr/log_ppl_diff:np.float64(9.43288893040517e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.0007712379218673959) - rollout_corr/log_ppl_diff_max:np.float64(0.005080167204141617) - rollout_corr/log_ppl_diff_min:np.float64(-0.004145577549934387) - rollout_corr/ppl_ratio:np.float64(1.000095043098554) - rollout_corr/chi2_token:np.float64(0.0003042914904654026) - rollout_corr/chi2_seq:np.float64(0.036600027000531554) - actor/pg_loss:np.float64(0.007485388522098901) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1953125) - actor/grad_norm:np.float64(0.22008594870567322) - perf/mfu/actor:np.float64(0.02548106706915905) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.09624481201172) - actor/lr:np.float64(1e-05) - training/global_step:78 - training/epoch:1 - critic/score/mean:0.69140625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.69140625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0042245774529874325 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0042245774529874325 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:156.265625 - response_length/max:473.0 - response_length/min:88.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:156.265625 - response_length_non_aborted/max:473.0 - response_length_non_aborted/min:88.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:279.5625 - prompt_length/max:449.0 - prompt_length/min:170.0 - 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:0.00013442710041999817 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9204160179942846) - timing_s/agent_loop/generate_sequences/max:np.float64(3.4967967849224806) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.7237196209971444) - timing_s/
|
||
|
|
step:79 - global_seqlen/min:10923 - global_seqlen/max:15918 - global_seqlen/minmax_diff:4995 - global_seqlen/balanced_min:14186 - global_seqlen/balanced_max:14189 - global_seqlen/mean:14187.375 - actor/entropy:0.06894775480031967 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.13839662075042725 - training/rollout_probs_diff_mean:0.0012521574972197413 - training/rollout_probs_diff_std:0.0063478765077888966 - training/rollout_actor_probs_pearson_corr:0.9989535212516785 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - self_distillation/correct_sample_fraction:0.63671875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8359375 - rollout_corr/rollout_is_mean:1.0000087022781372 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5303720235824585 - rollout_corr/rollout_is_min:0.5504809021949768 - rollout_corr/rollout_is_std:0.01740090735256672 - rollout_corr/rollout_is_eff_sample_size:0.999697180922422 - rollout_corr/rollout_is_seq_mean:1.0000083446502686 - rollout_corr/rollout_is_seq_std:0.001208899891935289 - rollout_corr/rollout_is_seq_max:1.0040807723999023 - rollout_corr/rollout_is_seq_min:0.9961971044540405 - rollout_corr/rollout_is_seq_max_deviation:0.004080772399902344 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0459575485438108) - rollout_corr/training_log_ppl:np.float64(0.043042610666361725) - rollout_corr/kl:np.float64(0.00011928708112129982) - rollout_corr/k3_kl:np.float64(0.00012756668218838474) - rollout_corr/rollout_ppl:np.float64(1.0458236997947097) - rollout_corr/rollout_log_ppl:np.float64(0.042923322981778256) - rollout_corr/log_ppl_diff:np.float64(0.00011928768458346894) - rollout_corr/log_ppl_abs_diff:np.float64(0.0008542470420707104) - rollout_corr/log_ppl_diff_max:np.float64(0.00455290824174881) - rollout_corr/log_ppl_diff_min:np.float64(-0.003551371395587921) - rollout_corr/ppl_ratio:np.float64(1.0001200516708195) - rollout_corr/chi2_token:np.float64(0.00026788911782205105) - rollout_corr/chi2_seq:np.float64(0.034723672084510326) - actor/pg_loss:np.float64(0.010854204153702085) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1640625) - actor/grad_norm:np.float64(0.2237369418144226) - perf/mfu/actor:np.float64(0.02618372089395636) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.1860008239746) - actor/lr:np.float64(1e-05) - training/global_step:79 - training/epoch:1 - critic/score/mean:0.63671875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.63671875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0035693298559635878 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0035693298559635878 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:179.48046875 - response_length/max:687.0 - response_length/min:88.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:179.48046875 - response_length_non_aborted/max:687.0 - response_length_non_aborted/min:88.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:263.875 - prompt_length/max:766.0 - prompt_length/min:165.0 - 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:0.00011689774692058563 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0910113006830215) - timing_s/agent_loop/generate_sequences/max:np.float64(4.717157164588571) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.9333629141910933) - timing_s/agent_lo
|
||
|
|
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 80}
|
||
|
|
validation generation end
|
||
|
|
[prompt] system
|
||
|
|
|
||
|
|
Given a question and four options, please select the right answer. Respond in the following format:
|
||
|
|
<reasoning>
|
||
|
|
...
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
...
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
user
|
||
|
|
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||
|
|
|
||
|
|
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||
|
|
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||
|
|
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||
|
|
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||
|
|
Please reason step by step.
|
||
|
|
assistant
|
||
|
|
<think>
|
||
|
|
|
||
|
|
</think>
|
||
|
|
|
||
|
|
|
||
|
|
[response] <reasoning>
|
||
|
|
The product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" contains a nitrogen-containing group (N), a carboxylic acid group (C(=O)O), and a fluorine atom (F). Option C contains a carboxylic acid group (C(=O)O), a nitrogen-containing group (N), and matches the product's structure by containing a fluorine atom (F) and a hydroxyl group (OH). The other options either lack the carboxylic acid group or do not match the product's structure. Therefore, the correct answer is Option C.
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
C
|
||
|
|
</answer>
|
||
|
|
[ground_truth] A
|
||
|
|
[score] 0.0
|
||
|
|
[acc] 0.0
|
||
|
|
[pred] C
|
||
|
|
[incorrect_format] 1
|
||
|
|
[feedback]
|
||
|
|
local_global_step_folder: /workspace/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-SDPO_TR-Qwen-Qwen3-4B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_80
|
||
|
|
step:80 - global_seqlen/min:11504 - global_seqlen/max:18081 - global_seqlen/minmax_diff:6577 - global_seqlen/balanced_min:14928 - global_seqlen/balanced_max:14961 - global_seqlen/mean:14955.25 - actor/entropy:0.05248134583234787 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.21512264013290405 - training/rollout_probs_diff_mean:0.0012333473423495889 - training/rollout_probs_diff_std:0.007220245897769928 - training/rollout_actor_probs_pearson_corr:0.9982454180717468 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.87109375 - self_distillation/correct_sample_fraction:0.7109375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.87109375 - rollout_corr/rollout_is_mean:0.9998933672904968 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:5.0841426855186e-05 - rollout_corr/rollout_is_max:1.6334035396575928 - rollout_corr/rollout_is_min:0.42125990986824036 - rollout_corr/rollout_is_std:0.018967699259519577 - rollout_corr/rollout_is_eff_sample_size:0.9996401175075921 - rollout_corr/rollout_is_seq_mean:0.9998857975006104 - rollout_corr/rollout_is_seq_std:0.0016545089893043041 - rollout_corr/rollout_is_seq_max:1.0078845024108887 - rollout_corr/rollout_is_seq_min:0.9949550032615662 - rollout_corr/rollout_is_seq_max_deviation:0.007884502410888672 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.044125270564109) - rollout_corr/training_log_ppl:np.float64(0.041402834057748805) - rollout_corr/kl:np.float64(0.00029025448099773143) - rollout_corr/k3_kl:np.float64(0.0001760211884106605) - rollout_corr/rollout_ppl:np.float64(1.043800718151033) - rollout_corr/rollout_log_ppl:np.float64(0.04111257939223378) - rollout_corr/log_ppl_diff:np.float64(0.00029025466551502177) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012071309597558866) - rollout_corr/log_ppl_diff_max:np.float64(0.005534961819648743) - rollout_corr/log_ppl_diff_min:np.float64(-0.006361238658428192) - rollout_corr/ppl_ratio:np.float64(1.0002916881348938) - rollout_corr/chi2_token:np.float64(0.00011730869300663471) - rollout_corr/chi2_seq:np.float64(0.03241542098112404) - actor/pg_loss:np.float64(0.01455320394586579) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.12890625) - actor/grad_norm:np.float64(0.29284343123435974) - perf/mfu/actor:np.float64(0.027665724040623355) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.24424362182617) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7023809523809523) - val-aux/sciknoweval/reward/std@16:np.float64(0.07939465803043781) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7358619047619048) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.06898693487351874) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6689285714285714) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.0665987146693429) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7027238095238095) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.07936131476944205) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7635904761904762) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.05060553230792266) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6424428571428572) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.04611964946365383) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7055809523809524) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.066899507622478) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7832) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.03212659649492731) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6246142857142857) - val-aux/sciknow
|
||
|
|
step:81 - global_seqlen/min:10370 - global_seqlen/max:17008 - global_seqlen/minmax_diff:6638 - global_seqlen/balanced_min:14090 - global_seqlen/balanced_max:14093 - global_seqlen/mean:14092.0 - actor/entropy:0.03968024626374245 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24491968750953674 - training/rollout_probs_diff_mean:0.0010003945790231228 - training/rollout_probs_diff_std:0.006792832165956497 - training/rollout_actor_probs_pearson_corr:0.9979662895202637 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - self_distillation/correct_sample_fraction:0.609375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83203125 - rollout_corr/rollout_is_mean:0.9999866485595703 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.00014726673543918878 - rollout_corr/rollout_is_max:1.64872407913208 - rollout_corr/rollout_is_min:0.4280904233455658 - rollout_corr/rollout_is_std:0.018435446545481682 - rollout_corr/rollout_is_eff_sample_size:0.9996602497849302 - rollout_corr/rollout_is_seq_mean:1.0000017881393433 - rollout_corr/rollout_is_seq_std:0.0015761813847348094 - rollout_corr/rollout_is_seq_max:1.0064371824264526 - rollout_corr/rollout_is_seq_min:0.9935445785522461 - rollout_corr/rollout_is_seq_max_deviation:0.006455421447753906 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0351794604212046) - rollout_corr/training_log_ppl:np.float64(0.03341220649645038) - rollout_corr/kl:np.float64(0.00016929561834833073) - rollout_corr/k3_kl:np.float64(0.00017109771060365713) - rollout_corr/rollout_ppl:np.float64(1.0349895739927888) - rollout_corr/rollout_log_ppl:np.float64(0.03324291037876037) - rollout_corr/log_ppl_diff:np.float64(0.00016929611769000985) - rollout_corr/log_ppl_abs_diff:np.float64(0.001038110969261652) - rollout_corr/log_ppl_diff_max:np.float64(0.009015612304210663) - rollout_corr/log_ppl_diff_min:np.float64(-0.005220435559749603) - rollout_corr/ppl_ratio:np.float64(1.0001707125920802) - rollout_corr/chi2_token:np.float64(0.0003324218560010195) - rollout_corr/chi2_seq:np.float64(0.04573003971017897) - actor/pg_loss:np.float64(0.011603191023851878) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.16796875) - actor/grad_norm:np.float64(0.24654333293437958) - perf/mfu/actor:np.float64(0.026028145972960923) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(149.31386947631836) - actor/lr:np.float64(1e-05) - training/global_step:81 - training/epoch:1 - critic/score/mean:0.609375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.609375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0007031986606307328 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0007031986606307328 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:132.625 - response_length/max:514.0 - response_length/min:81.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:132.625 - response_length_non_aborted/max:514.0 - response_length_non_aborted/min:81.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:307.75 - prompt_length/max:735.0 - prompt_length/min:170.0 - 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:7.819943130016327e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(0.6045786067843437) - timing_s/agent_loop/generate_sequences/max:np.float64(3.653476554900408) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.3476513583955239) - timing_s/agent_loo
|
||
|
|
step:82 - global_seqlen/min:11137 - global_seqlen/max:15847 - global_seqlen/minmax_diff:4710 - global_seqlen/balanced_min:13173 - global_seqlen/balanced_max:13602 - global_seqlen/mean:13241.875 - actor/entropy:0.060025569051504135 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.19035351276397705 - training/rollout_probs_diff_mean:0.0011168333003297448 - training/rollout_probs_diff_std:0.006102167069911957 - training/rollout_actor_probs_pearson_corr:0.9989132285118103 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.8984375 - self_distillation/correct_sample_fraction:0.67578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8984375 - rollout_corr/rollout_is_mean:0.999947726726532 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5513248443603516 - rollout_corr/rollout_is_min:0.6109191179275513 - rollout_corr/rollout_is_std:0.01643557846546173 - rollout_corr/rollout_is_eff_sample_size:0.9997298255545458 - rollout_corr/rollout_is_seq_mean:0.9999595284461975 - rollout_corr/rollout_is_seq_std:0.0012216182658448815 - rollout_corr/rollout_is_seq_max:1.0048490762710571 - rollout_corr/rollout_is_seq_min:0.9947028756141663 - rollout_corr/rollout_is_seq_max_deviation:0.00529712438583374 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0472495881840587) - rollout_corr/training_log_ppl:np.float64(0.043586462446910446) - rollout_corr/kl:np.float64(0.0001512962158181086) - rollout_corr/k3_kl:np.float64(0.00011082207649528255) - rollout_corr/rollout_ppl:np.float64(1.0470793913118541) - rollout_corr/rollout_log_ppl:np.float64(0.0434351656612435) - rollout_corr/log_ppl_diff:np.float64(0.00015129678566694338) - rollout_corr/log_ppl_abs_diff:np.float64(0.0008437764319566554) - rollout_corr/log_ppl_diff_max:np.float64(0.005687326192855835) - rollout_corr/log_ppl_diff_min:np.float64(-0.0035227984189987183) - rollout_corr/ppl_ratio:np.float64(1.0001520542427897) - rollout_corr/chi2_token:np.float64(0.00014312099665403366) - rollout_corr/chi2_seq:np.float64(0.013327852822840214) - actor/pg_loss:np.float64(0.014179444573983346) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1015625) - actor/grad_norm:np.float64(0.2800471782684326) - perf/mfu/actor:np.float64(0.024405619523858633) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(149.01367568969727) - actor/lr:np.float64(1e-05) - training/global_step:82 - 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:-0.012409777380526066 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.012409777380526066 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:147.74609375 - response_length/max:974.0 - response_length/min:89.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:147.74609375 - response_length_non_aborted/max:974.0 - response_length_non_aborted/min:89.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:266.0625 - prompt_length/max:408.0 - prompt_length/min:163.0 - 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:0.00011133402585983276 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0631630178540945) - timing_s/agent_loop/generate_sequences/max:np.float64(5.774274609982967) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.6591488730409765) - timing_s/agent_loop
|
||
|
|
step:83 - global_seqlen/min:11128 - global_seqlen/max:15122 - global_seqlen/minmax_diff:3994 - global_seqlen/balanced_min:13601 - global_seqlen/balanced_max:13803 - global_seqlen/mean:13627.0 - actor/entropy:0.044485900551080704 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23965704441070557 - training/rollout_probs_diff_mean:0.0009909354848787189 - training/rollout_probs_diff_std:0.006712916307151318 - training/rollout_actor_probs_pearson_corr:0.9981244802474976 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.78125 - self_distillation/correct_sample_fraction:0.640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.78125 - rollout_corr/rollout_is_mean:1.0000256299972534 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.755704641342163 - rollout_corr/rollout_is_min:0.5292616486549377 - rollout_corr/rollout_is_std:0.016995415091514587 - rollout_corr/rollout_is_eff_sample_size:0.9997112392982459 - rollout_corr/rollout_is_seq_mean:1.0000783205032349 - rollout_corr/rollout_is_seq_std:0.0012930911034345627 - rollout_corr/rollout_is_seq_max:1.0052177906036377 - rollout_corr/rollout_is_seq_min:0.9961187839508057 - rollout_corr/rollout_is_seq_max_deviation:0.005217790603637695 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0283324904739857) - rollout_corr/training_log_ppl:np.float64(0.026819557405588057) - rollout_corr/kl:np.float64(4.604050838519225e-05) - rollout_corr/k3_kl:np.float64(0.0001243426268259651) - rollout_corr/rollout_ppl:np.float64(1.028277720324695) - rollout_corr/rollout_log_ppl:np.float64(0.02677351630618574) - rollout_corr/log_ppl_diff:np.float64(4.604109940231638e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.0008100297492248387) - rollout_corr/log_ppl_diff_max:np.float64(0.004717933014035225) - rollout_corr/log_ppl_diff_min:np.float64(-0.003781719133257866) - rollout_corr/ppl_ratio:np.float64(1.0000469228252769) - rollout_corr/chi2_token:np.float64(0.0004020256455987692) - rollout_corr/chi2_seq:np.float64(0.037706532049924135) - actor/pg_loss:np.float64(0.010695430000623674) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.21875) - actor/grad_norm:np.float64(0.20858627557754517) - perf/mfu/actor:np.float64(0.025191763488066923) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.8725471496582) - actor/lr:np.float64(1e-05) - training/global_step:83 - training/epoch:1 - critic/score/mean:0.640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.007154726888984442 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.007154726888984442 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:126.9375 - response_length/max:892.0 - response_length/min:78.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:126.9375 - response_length_non_aborted/max:892.0 - response_length_non_aborted/min:78.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:298.90625 - prompt_length/max:563.0 - prompt_length/min:170.0 - 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:0.00015988945960998535 - timing_s/agent_loop/generate_sequences/min:np.float64(0.8830814864486456) - timing_s/agent_loop/generate_sequences/max:np.float64(5.12624941393733) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.4224173862530733) - timing_s/agent_loop/tool_calls/min:np.float64
|
||
|
|
step:84 - global_seqlen/min:10525 - global_seqlen/max:15608 - global_seqlen/minmax_diff:5083 - global_seqlen/balanced_min:12838 - global_seqlen/balanced_max:12840 - global_seqlen/mean:12839.625 - actor/entropy:0.06778481602668762 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12655651569366455 - training/rollout_probs_diff_mean:0.0010602419497445226 - training/rollout_probs_diff_std:0.0057013193145394325 - training/rollout_actor_probs_pearson_corr:0.9991278648376465 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.90234375 - self_distillation/correct_sample_fraction:0.7578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.90234375 - rollout_corr/rollout_is_mean:0.9999590516090393 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6338132619857788 - rollout_corr/rollout_is_min:0.6070662140846252 - rollout_corr/rollout_is_std:0.016988398507237434 - rollout_corr/rollout_is_eff_sample_size:0.9997113584387137 - rollout_corr/rollout_is_seq_mean:0.9999632239341736 - rollout_corr/rollout_is_seq_std:0.001300561474636197 - rollout_corr/rollout_is_seq_max:1.0062448978424072 - rollout_corr/rollout_is_seq_min:0.9953770041465759 - rollout_corr/rollout_is_seq_max_deviation:0.0062448978424072266 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0418651765212417) - rollout_corr/training_log_ppl:np.float64(0.03784611283361983) - rollout_corr/kl:np.float64(0.00015446774743299585) - rollout_corr/k3_kl:np.float64(0.00011775112265250656) - rollout_corr/rollout_ppl:np.float64(1.0416851043701172) - rollout_corr/rollout_log_ppl:np.float64(0.03769164434780237) - rollout_corr/log_ppl_diff:np.float64(0.00015446848581746053) - rollout_corr/log_ppl_abs_diff:np.float64(0.0008566886906180571) - rollout_corr/log_ppl_diff_max:np.float64(0.005237758159637451) - rollout_corr/log_ppl_diff_min:np.float64(-0.004913561046123505) - rollout_corr/ppl_ratio:np.float64(1.0001553527545184) - rollout_corr/chi2_token:np.float64(0.00016302778385579586) - rollout_corr/chi2_seq:np.float64(0.08548461622558534) - actor/pg_loss:np.float64(0.01143554385011214) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.09765625) - actor/grad_norm:np.float64(0.18208560347557068) - perf/mfu/actor:np.float64(0.023830850888686333) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.48444747924805) - actor/lr:np.float64(1e-05) - training/global_step:84 - training/epoch:1 - critic/score/mean:0.7578125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7578125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0001791163522284478 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0001791163522284478 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:147.20703125 - response_length/max:611.0 - response_length/min:84.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:147.20703125 - response_length_non_aborted/max:611.0 - response_length_non_aborted/min:84.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:254.03125 - prompt_length/max:466.0 - prompt_length/min:163.0 - 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:0.00013000331819057465 - timing_s/agent_loop/generate_sequences/min:np.float64(1.095025546848774) - timing_s/agent_loop/generate_sequences/max:np.float64(3.965017654001713) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.7116082004431519) - timing_s/agent_
|
||
|
|
step:85 - global_seqlen/min:11648 - global_seqlen/max:17427 - global_seqlen/minmax_diff:5779 - global_seqlen/balanced_min:13791 - global_seqlen/balanced_max:13794 - global_seqlen/mean:13793.125 - actor/entropy:0.02245117723941803 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2093091607093811 - training/rollout_probs_diff_mean:0.000610601797234267 - training/rollout_probs_diff_std:0.005804179701954126 - training/rollout_actor_probs_pearson_corr:0.997347354888916 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.7578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8125 - rollout_corr/rollout_is_mean:1.0001749992370605 - rollout_corr/rollout_is_ratio_fraction_high:2.9768107197014615e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:4.472691535949707 - rollout_corr/rollout_is_min:0.6783106923103333 - rollout_corr/rollout_is_std:0.014197989366948605 - rollout_corr/rollout_is_eff_sample_size:0.9997985768801061 - rollout_corr/rollout_is_seq_mean:1.0002259016036987 - rollout_corr/rollout_is_seq_std:0.002043855143710971 - rollout_corr/rollout_is_seq_max:1.0259480476379395 - rollout_corr/rollout_is_seq_min:0.9973135590553284 - rollout_corr/rollout_is_seq_max_deviation:0.025948047637939453 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0188493309542537) - rollout_corr/training_log_ppl:np.float64(0.01781918815561312) - rollout_corr/kl:np.float64(-9.15504629754732e-05) - rollout_corr/k3_kl:np.float64(0.0001342924706074105) - rollout_corr/rollout_ppl:np.float64(1.0189437484368682) - rollout_corr/rollout_log_ppl:np.float64(0.017910738053615205) - rollout_corr/log_ppl_diff:np.float64(-9.154989641046996e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.0007111164925390767) - rollout_corr/log_ppl_diff_max:np.float64(0.0032405853271484375) - rollout_corr/log_ppl_diff_min:np.float64(-0.011211381293833256) - rollout_corr/ppl_ratio:np.float64(0.9999094093218446) - rollout_corr/chi2_token:np.float64(0.0009676266927272081) - rollout_corr/chi2_seq:np.float64(0.1281680171377957) - actor/pg_loss:np.float64(0.006235062068046915) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.29819807410240173) - perf/mfu/actor:np.float64(0.025649074187588758) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.30879592895508) - actor/lr:np.float64(1e-05) - training/global_step:85 - training/epoch:1 - critic/score/mean:0.7578125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7578125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.004658708814531565 - critic/advantages/max:0.5 - critic/advantages/min:-0.875 - critic/returns/mean:0.004658708814531565 - critic/returns/max:0.5 - critic/returns/min:-0.875 - response_length/mean:131.22265625 - response_length/max:518.0 - response_length/min:91.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:131.22265625 - response_length_non_aborted/max:518.0 - response_length_non_aborted/min:91.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:299.8125 - prompt_length/max:706.0 - prompt_length/min:172.0 - 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:0.0001498386263847351 - timing_s/agent_loop/generate_sequences/min:np.float64(1.122318483889103) - timing_s/agent_loop/generate_sequences/max:np.float64(3.4682942163199186) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.63217447714851) - timing_s/agent_loop/tool_
|
||
|
|
step:86 - global_seqlen/min:11663 - global_seqlen/max:14542 - global_seqlen/minmax_diff:2879 - global_seqlen/balanced_min:12749 - global_seqlen/balanced_max:12751 - global_seqlen/mean:12750.375 - actor/entropy:0.041824180632829666 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.15314745903015137 - training/rollout_probs_diff_mean:0.0009020075085572898 - training/rollout_probs_diff_std:0.006250815466046333 - training/rollout_actor_probs_pearson_corr:0.9983953237533569 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.79296875 - self_distillation/correct_sample_fraction:0.5546875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.79296875 - rollout_corr/rollout_is_mean:1.0000126361846924 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.562320590019226 - rollout_corr/rollout_is_min:0.6136078238487244 - rollout_corr/rollout_is_std:0.01628987118601799 - rollout_corr/rollout_is_eff_sample_size:0.9997347105186445 - rollout_corr/rollout_is_seq_mean:1.0000067949295044 - rollout_corr/rollout_is_seq_std:0.001423215726390481 - rollout_corr/rollout_is_seq_max:1.0053277015686035 - rollout_corr/rollout_is_seq_min:0.9946157336235046 - rollout_corr/rollout_is_seq_max_deviation:0.005384266376495361 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0339017608202994) - rollout_corr/training_log_ppl:np.float64(0.030575152453025112) - rollout_corr/kl:np.float64(0.0001128499749730727) - rollout_corr/k3_kl:np.float64(0.00011968861869084309) - rollout_corr/rollout_ppl:np.float64(1.0337728704325855) - rollout_corr/rollout_log_ppl:np.float64(0.03046230203989353) - rollout_corr/log_ppl_diff:np.float64(0.00011285041313158217) - rollout_corr/log_ppl_abs_diff:np.float64(0.0009649223407564023) - rollout_corr/log_ppl_diff_max:np.float64(0.005895189940929413) - rollout_corr/log_ppl_diff_min:np.float64(-0.005070731043815613) - rollout_corr/ppl_ratio:np.float64(1.0001138679217547) - rollout_corr/chi2_token:np.float64(0.00025752536021173) - rollout_corr/chi2_seq:np.float64(0.038065674947574735) - actor/pg_loss:np.float64(0.01089239947489773) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.20703125) - actor/grad_norm:np.float64(0.23640994727611542) - perf/mfu/actor:np.float64(0.023879704816425578) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.29499053955078) - actor/lr:np.float64(1e-05) - training/global_step:86 - training/epoch:1 - critic/score/mean:0.5546875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5546875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00030246813548728824 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00030246813548728824 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:121.07421875 - response_length/max:527.0 - response_length/min:85.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:121.07421875 - response_length_non_aborted/max:527.0 - response_length_non_aborted/min:85.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:277.375 - prompt_length/max:503.0 - prompt_length/min:170.0 - 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:7.187016308307648e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(0.936371335759759) - timing_s/agent_loop/generate_sequences/max:np.float64(3.4768210630863905) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.3832009471370839) - timing_s/agent_loop/t
|
||
|
|
step:87 - global_seqlen/min:11582 - global_seqlen/max:15930 - global_seqlen/minmax_diff:4348 - global_seqlen/balanced_min:14004 - global_seqlen/balanced_max:14016 - global_seqlen/mean:14006.875 - actor/entropy:0.05480833724141121 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.16897732019424438 - training/rollout_probs_diff_mean:0.000986600760370493 - training/rollout_probs_diff_std:0.006059554405510426 - training/rollout_actor_probs_pearson_corr:0.9988378882408142 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - self_distillation/correct_sample_fraction:0.734375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8359375 - rollout_corr/rollout_is_mean:1.00006103515625 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6415419578552246 - rollout_corr/rollout_is_min:0.5123928189277649 - rollout_corr/rollout_is_std:0.01711772195994854 - rollout_corr/rollout_is_eff_sample_size:0.9997071885392379 - rollout_corr/rollout_is_seq_mean:1.0000572204589844 - rollout_corr/rollout_is_seq_std:0.0013290804345160723 - rollout_corr/rollout_is_seq_max:1.0064915418624878 - rollout_corr/rollout_is_seq_min:0.9957578182220459 - rollout_corr/rollout_is_seq_max_deviation:0.006491541862487793 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0299364058300853) - rollout_corr/training_log_ppl:np.float64(0.027746889040997758) - rollout_corr/kl:np.float64(6.69919317624057e-05) - rollout_corr/k3_kl:np.float64(0.00012424909971345786) - rollout_corr/rollout_ppl:np.float64(1.029863527044654) - rollout_corr/rollout_log_ppl:np.float64(0.027679896902441214) - rollout_corr/log_ppl_diff:np.float64(6.699213855654307e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.0007840513266614835) - rollout_corr/log_ppl_diff_max:np.float64(0.004926964640617371) - rollout_corr/log_ppl_diff_min:np.float64(-0.005129293538630009) - rollout_corr/ppl_ratio:np.float64(1.000067861750722) - rollout_corr/chi2_token:np.float64(0.0003655164036899805) - rollout_corr/chi2_seq:np.float64(0.03480148012749851) - actor/pg_loss:np.float64(0.008134817650386594) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1640625) - actor/grad_norm:np.float64(0.20570968091487885) - perf/mfu/actor:np.float64(0.0257927784692671) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.2779312133789) - actor/lr:np.float64(1e-05) - training/global_step:87 - training/epoch:1 - critic/score/mean:0.734375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.734375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.001170534174889326 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.001170534174889326 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:142.24609375 - response_length/max:718.0 - response_length/min:85.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:142.24609375 - response_length_non_aborted/max:718.0 - response_length_non_aborted/min:85.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:295.46875 - prompt_length/max:845.0 - prompt_length/min:171.0 - 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:4.02890145778656e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0555304419249296) - timing_s/agent_loop/generate_sequences/max:np.float64(4.630197988823056) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.6122335814943654) - timing_s/agent_loop/tool_calls/mi
|
||
|
|
step:88 - global_seqlen/min:11451 - global_seqlen/max:13471 - global_seqlen/minmax_diff:2020 - global_seqlen/balanced_min:12430 - global_seqlen/balanced_max:12431 - global_seqlen/mean:12430.375 - actor/entropy:0.022946609184145927 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24135535955429077 - training/rollout_probs_diff_mean:0.0006310467724688351 - training/rollout_probs_diff_std:0.0057634930126369 - training/rollout_actor_probs_pearson_corr:0.9972387552261353 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.76953125 - self_distillation/correct_sample_fraction:0.6640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.76953125 - rollout_corr/rollout_is_mean:1.0002304315567017 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.7523047924041748 - rollout_corr/rollout_is_min:0.6783539652824402 - rollout_corr/rollout_is_std:0.014798224903643131 - rollout_corr/rollout_is_eff_sample_size:0.9997811796372159 - rollout_corr/rollout_is_seq_mean:1.000230312347412 - rollout_corr/rollout_is_seq_std:0.0013730740174651146 - rollout_corr/rollout_is_seq_max:1.0076873302459717 - rollout_corr/rollout_is_seq_min:0.996023952960968 - rollout_corr/rollout_is_seq_max_deviation:0.00768733024597168 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0201888671144843) - rollout_corr/training_log_ppl:np.float64(0.019152402668737523) - rollout_corr/kl:np.float64(-0.0001377914098066313) - rollout_corr/k3_kl:np.float64(9.254714846145201e-05) - rollout_corr/rollout_ppl:np.float64(1.0203322544693947) - rollout_corr/rollout_log_ppl:np.float64(0.01929019407469923) - rollout_corr/log_ppl_diff:np.float64(-0.00013779140596170691) - rollout_corr/log_ppl_abs_diff:np.float64(0.0007093784771541323) - rollout_corr/log_ppl_diff_max:np.float64(0.004439342767000198) - rollout_corr/log_ppl_diff_min:np.float64(-0.006012720987200737) - rollout_corr/ppl_ratio:np.float64(0.9998630278278142) - rollout_corr/chi2_token:np.float64(0.0006705890409648418) - rollout_corr/chi2_seq:np.float64(0.09148641373030841) - actor/pg_loss:np.float64(0.008129281004016775) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.23046875) - actor/grad_norm:np.float64(0.23788434267044067) - perf/mfu/actor:np.float64(0.02319708061673284) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.29092407226562) - actor/lr:np.float64(1e-05) - training/global_step:88 - training/epoch:1 - critic/score/mean:0.6640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.000375485687982291 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.000375485687982291 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:119.63671875 - response_length/max:229.0 - response_length/min:78.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:119.63671875 - response_length_non_aborted/max:229.0 - response_length_non_aborted/min:78.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:268.8125 - prompt_length/max:482.0 - prompt_length/min:166.0 - 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:0.00010538101196289062 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9504817668348551) - timing_s/agent_loop/generate_sequences/max:np.float64(2.05767609924078) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.471321147022536) - timing_s/agent_loop/to
|
||
|
|
step:89 - global_seqlen/min:11744 - global_seqlen/max:16035 - global_seqlen/minmax_diff:4291 - global_seqlen/balanced_min:13322 - global_seqlen/balanced_max:13324 - global_seqlen/mean:13323.25 - actor/entropy:0.04421846568584442 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12246036529541016 - training/rollout_probs_diff_mean:0.0008685207576490939 - training/rollout_probs_diff_std:0.005534641444683075 - training/rollout_actor_probs_pearson_corr:0.9987534284591675 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.77734375 - self_distillation/correct_sample_fraction:0.640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.77734375 - rollout_corr/rollout_is_mean:1.0000410079956055 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.609689474105835 - rollout_corr/rollout_is_min:0.7242475748062134 - rollout_corr/rollout_is_std:0.015513978898525238 - rollout_corr/rollout_is_eff_sample_size:0.9997593743590318 - rollout_corr/rollout_is_seq_mean:1.0000495910644531 - rollout_corr/rollout_is_seq_std:0.0013041149359196424 - rollout_corr/rollout_is_seq_max:1.005750060081482 - rollout_corr/rollout_is_seq_min:0.996555745601654 - rollout_corr/rollout_is_seq_max_deviation:0.005750060081481934 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0292462357319891) - rollout_corr/training_log_ppl:np.float64(0.02711499427778108) - rollout_corr/kl:np.float64(5.409235772918519e-05) - rollout_corr/k3_kl:np.float64(0.00010377083346918906) - rollout_corr/rollout_ppl:np.float64(1.0291879558935761) - rollout_corr/rollout_log_ppl:np.float64(0.027060901795294967) - rollout_corr/log_ppl_diff:np.float64(5.40924824861122e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.0007935035443153993) - rollout_corr/log_ppl_diff_max:np.float64(0.0037947893142700195) - rollout_corr/log_ppl_diff_min:np.float64(-0.004834167659282684) - rollout_corr/ppl_ratio:np.float64(1.0000548926182091) - rollout_corr/chi2_token:np.float64(0.0003208282869309187) - rollout_corr/chi2_seq:np.float64(0.026848674518987536) - actor/pg_loss:np.float64(0.008321232285965152) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.22265625) - actor/grad_norm:np.float64(0.1950538456439972) - perf/mfu/actor:np.float64(0.0248378100174118) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.520263671875) - actor/lr:np.float64(1e-05) - training/global_step:89 - training/epoch:1 - critic/score/mean:0.640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.005691050551831722 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.005691050551831722 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:125.0078125 - response_length/max:450.0 - response_length/min:71.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:125.0078125 - response_length_non_aborted/max:450.0 - response_length_non_aborted/min:71.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:291.34375 - prompt_length/max:720.0 - prompt_length/min:169.0 - 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:0.00013791397213935852 - timing_s/agent_loop/generate_sequences/min:np.float64(0.738397516310215) - timing_s/agent_loop/generate_sequences/max:np.float64(3.0790412425994873) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.4573000645978027) - timing_s/agent_loop/tool_calls/
|
||
|
|
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 90}
|
||
|
|
validation generation end
|
||
|
|
[prompt] system
|
||
|
|
|
||
|
|
Given a question and four options, please select the right answer. Respond in the following format:
|
||
|
|
<reasoning>
|
||
|
|
...
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
...
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
user
|
||
|
|
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||
|
|
|
||
|
|
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||
|
|
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||
|
|
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||
|
|
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||
|
|
Please reason step by step.
|
||
|
|
assistant
|
||
|
|
<think>
|
||
|
|
|
||
|
|
</think>
|
||
|
|
|
||
|
|
|
||
|
|
[response] <reasoning>
|
||
|
|
The product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" contains a carboxylic acid group (C(=O)), a nitrogen-containing group (N), and a complex hydrocarbon chain. Option A contains a carboxylic acid group (C(=O)), a nitrogen-containing group (N), and matches the product's structure, including the hydrocarbon chain and functional groups. The other options either lack the carboxylic acid group or do not match the product's structure. Therefore, the correct answer is Option A.
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
[ground_truth] A
|
||
|
|
[score] 1.0
|
||
|
|
[acc] 1.0
|
||
|
|
[pred] A
|
||
|
|
[incorrect_format] 1
|
||
|
|
[feedback]
|
||
|
|
local_global_step_folder: /workspace/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-SDPO_TR-Qwen-Qwen3-4B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_90
|
||
|
|
step:90 - global_seqlen/min:10800 - global_seqlen/max:16199 - global_seqlen/minmax_diff:5399 - global_seqlen/balanced_min:13405 - global_seqlen/balanced_max:13406 - global_seqlen/mean:13405.25 - actor/entropy:0.05048282444477081 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.17359209060668945 - training/rollout_probs_diff_mean:0.0007999222143553197 - training/rollout_probs_diff_std:0.005081967916339636 - training/rollout_actor_probs_pearson_corr:0.999090313911438 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.84375 - self_distillation/correct_sample_fraction:0.72265625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.84375 - rollout_corr/rollout_is_mean:0.9999803900718689 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:8.308868564199656e-05 - rollout_corr/rollout_is_max:1.5303757190704346 - rollout_corr/rollout_is_min:0.3545355796813965 - rollout_corr/rollout_is_std:0.01461584959179163 - rollout_corr/rollout_is_eff_sample_size:0.9997863034198247 - rollout_corr/rollout_is_seq_mean:0.9999797940254211 - rollout_corr/rollout_is_seq_std:0.0011293459683656693 - rollout_corr/rollout_is_seq_max:1.0052525997161865 - rollout_corr/rollout_is_seq_min:0.99466872215271 - rollout_corr/rollout_is_seq_max_deviation:0.005331277847290039 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0296389972791076) - rollout_corr/training_log_ppl:np.float64(0.027549039331802305) - rollout_corr/kl:np.float64(0.0001258116808598997) - rollout_corr/k3_kl:np.float64(0.0001056011147461686) - rollout_corr/rollout_ppl:np.float64(1.0295025557279587) - rollout_corr/rollout_log_ppl:np.float64(0.02742322766897587) - rollout_corr/log_ppl_diff:np.float64(0.00012581166282643608) - rollout_corr/log_ppl_abs_diff:np.float64(0.0006251908834542519) - rollout_corr/log_ppl_diff_max:np.float64(0.008351578377187252) - rollout_corr/log_ppl_diff_min:np.float64(-0.004474513232707977) - rollout_corr/ppl_ratio:np.float64(1.0001267069019377) - rollout_corr/chi2_token:np.float64(0.00013855332508683205) - rollout_corr/chi2_seq:np.float64(0.02339452737942338) - actor/pg_loss:np.float64(0.006540415857301696) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.15625) - actor/grad_norm:np.float64(0.13174369931221008) - perf/mfu/actor:np.float64(0.024985257488655037) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.45252227783203) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.743452380952381) - val-aux/sciknoweval/reward/std@16:np.float64(0.06569448383701967) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.768552380952381) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.053744888849360184) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.717795238095238) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.05690907553745164) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7433666666666666) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.06575666613585864) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7893428571428571) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.041637420806750265) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6963285714285714) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.046257777187404994) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7448380952380952) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.04959794568559073) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.8064333333333333) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.030793787147274983) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6779190476190476) - val-au
|
||
|
|
step:91 - global_seqlen/min:10989 - global_seqlen/max:21502 - global_seqlen/minmax_diff:10513 - global_seqlen/balanced_min:13549 - global_seqlen/balanced_max:20576 - global_seqlen/mean:14474.0 - actor/entropy:0.03857384994626045 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.15483945608139038 - training/rollout_probs_diff_mean:0.0006573514547199011 - training/rollout_probs_diff_std:0.00461356807500124 - training/rollout_actor_probs_pearson_corr:0.9990202784538269 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.86328125 - self_distillation/correct_sample_fraction:0.671875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.86328125 - rollout_corr/rollout_is_mean:0.999982476234436 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.543942928314209 - rollout_corr/rollout_is_min:0.6783113479614258 - rollout_corr/rollout_is_std:0.01197534054517746 - rollout_corr/rollout_is_eff_sample_size:0.9998566117877983 - rollout_corr/rollout_is_seq_mean:0.9999902248382568 - rollout_corr/rollout_is_seq_std:0.0010964501416310668 - rollout_corr/rollout_is_seq_max:1.0052000284194946 - rollout_corr/rollout_is_seq_min:0.9965869784355164 - rollout_corr/rollout_is_seq_max_deviation:0.005200028419494629 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0374901527538896) - rollout_corr/training_log_ppl:np.float64(0.03470212036074827) - rollout_corr/kl:np.float64(9.124622281975192e-05) - rollout_corr/k3_kl:np.float64(8.150314501727436e-05) - rollout_corr/rollout_ppl:np.float64(1.037398514803499) - rollout_corr/rollout_log_ppl:np.float64(0.034610873819161725) - rollout_corr/log_ppl_diff:np.float64(9.124654158654266e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.0006715026422909887) - rollout_corr/log_ppl_diff_max:np.float64(0.0038138628005981445) - rollout_corr/log_ppl_diff_min:np.float64(-0.0043762922286987305) - rollout_corr/ppl_ratio:np.float64(1.0000918528530747) - rollout_corr/chi2_token:np.float64(0.00014221644960343838) - rollout_corr/chi2_seq:np.float64(0.01914523052982986) - actor/pg_loss:np.float64(0.011343885561558409) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.13671875) - actor/grad_norm:np.float64(0.17565734684467316) - perf/mfu/actor:np.float64(0.025028156618550214) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.79314422607422) - actor/lr:np.float64(1e-05) - training/global_step:91 - training/epoch:1 - critic/score/mean:0.671875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.671875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.17195726931095123 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.17195726931095123 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:166.1875 - response_length/max:8192.0 - response_length/min:87.0 - response_length/clip_ratio:0.00390625 - response_length_non_aborted/mean:166.1875 - response_length_non_aborted/max:8192.0 - response_length_non_aborted/min:87.0 - response_length_non_aborted/clip_ratio:0.00390625 - response/aborted_ratio:0.0 - prompt_length/mean:286.125 - prompt_length/max:1009.0 - prompt_length/min:176.0 - 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:0.00010056421160697937 - timing_s/agent_loop/generate_sequences/min:np.float64(0.9996694829314947) - timing_s/agent_loop/generate_sequences/max:np.float64(49.42658263258636) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.7606404674661462) - timing_s/agent_loop/
|
||
|
|
step:92 - global_seqlen/min:12185 - global_seqlen/max:14909 - global_seqlen/minmax_diff:2724 - global_seqlen/balanced_min:13566 - global_seqlen/balanced_max:13567 - global_seqlen/mean:13566.75 - actor/entropy:0.05854354798793793 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.15478312969207764 - training/rollout_probs_diff_mean:0.0010605284478515387 - training/rollout_probs_diff_std:0.00595305860042572 - training/rollout_actor_probs_pearson_corr:0.9988794326782227 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.8671875 - self_distillation/correct_sample_fraction:0.7265625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8671875 - rollout_corr/rollout_is_mean:1.0000475645065308 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.62340247631073 - rollout_corr/rollout_is_min:0.6301739811897278 - rollout_corr/rollout_is_std:0.015799501910805702 - rollout_corr/rollout_is_eff_sample_size:0.9997503188946146 - rollout_corr/rollout_is_seq_mean:1.0000717639923096 - rollout_corr/rollout_is_seq_std:0.0012398355174809694 - rollout_corr/rollout_is_seq_max:1.006272315979004 - rollout_corr/rollout_is_seq_min:0.9960307478904724 - rollout_corr/rollout_is_seq_max_deviation:0.006272315979003906 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0421050656586885) - rollout_corr/training_log_ppl:np.float64(0.039705921768018015) - rollout_corr/kl:np.float64(4.001860810198821e-05) - rollout_corr/k3_kl:np.float64(0.00011182411940677608) - rollout_corr/rollout_ppl:np.float64(1.0420579640194774) - rollout_corr/rollout_log_ppl:np.float64(0.03966590254384528) - rollout_corr/log_ppl_diff:np.float64(4.001922417273818e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.0008390767857235915) - rollout_corr/log_ppl_diff_max:np.float64(0.004487637430429459) - rollout_corr/log_ppl_diff_min:np.float64(-0.0055781155824661255) - rollout_corr/ppl_ratio:np.float64(1.0000407760962844) - rollout_corr/chi2_token:np.float64(0.00036986544728279114) - rollout_corr/chi2_seq:np.float64(0.03695979272015393) - actor/pg_loss:np.float64(0.013767448267131499) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1328125) - actor/grad_norm:np.float64(0.22888551652431488) - perf/mfu/actor:np.float64(0.025295786362716385) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.95596313476562) - actor/lr:np.float64(1e-05) - training/global_step:92 - 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:-0.001878542359918356 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.001878542359918356 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:136.4609375 - response_length/max:508.0 - response_length/min:86.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:136.4609375 - response_length_non_aborted/max:508.0 - response_length_non_aborted/min:86.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:287.5 - prompt_length/max:436.0 - prompt_length/min:172.0 - 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:0.00015094876289367676 - timing_s/agent_loop/generate_sequences/min:np.float64(1.052999148145318) - timing_s/agent_loop/generate_sequences/max:np.float64(3.5217525959014893) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.5543789482326247) - timing_s/agent_loop/tool_calls/m
|
||
|
|
step:93 - global_seqlen/min:10727 - global_seqlen/max:17360 - global_seqlen/minmax_diff:6633 - global_seqlen/balanced_min:13189 - global_seqlen/balanced_max:13190 - global_seqlen/mean:13189.125 - actor/entropy:0.04647291824221611 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.857110857963562 - training/rollout_probs_diff_mean:0.0011152068618685007 - training/rollout_probs_diff_std:0.01373054925352335 - training/rollout_actor_probs_pearson_corr:0.9932172894477844 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.87109375 - self_distillation/correct_sample_fraction:0.7890625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.87109375 - rollout_corr/rollout_is_mean:0.9998798966407776 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0002144944592146203 - rollout_corr/rollout_is_max:1.868226408958435 - rollout_corr/rollout_is_min:0.052375778555870056 - rollout_corr/rollout_is_std:0.01766606606543064 - rollout_corr/rollout_is_eff_sample_size:0.9996880074496861 - rollout_corr/rollout_is_seq_mean:0.9998594522476196 - rollout_corr/rollout_is_seq_std:0.0016694950172677636 - rollout_corr/rollout_is_seq_max:1.0071744918823242 - rollout_corr/rollout_is_seq_min:0.9942285418510437 - rollout_corr/rollout_is_seq_max_deviation:0.007174491882324219 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.042677128687501) - rollout_corr/training_log_ppl:np.float64(0.03964809522994983) - rollout_corr/kl:np.float64(0.0006370269320958855) - rollout_corr/k3_kl:np.float64(0.0004965091741022312) - rollout_corr/rollout_ppl:np.float64(1.0420256052166224) - rollout_corr/rollout_log_ppl:np.float64(0.03901106807026622) - rollout_corr/log_ppl_diff:np.float64(0.0006370271587741172) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013677216842609141) - rollout_corr/log_ppl_diff_max:np.float64(0.01775827631354332) - rollout_corr/log_ppl_diff_min:np.float64(-0.0053708478808403015) - rollout_corr/ppl_ratio:np.float64(1.0006428451742977) - rollout_corr/chi2_token:np.float64(0.00013257237151265144) - rollout_corr/chi2_seq:np.float64(0.04067836725153029) - actor/pg_loss:np.float64(0.01683109411840178) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.12890625) - actor/grad_norm:np.float64(0.21249641478061676) - perf/mfu/actor:np.float64(0.02462064372088703) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.76744842529297) - actor/lr:np.float64(1e-05) - training/global_step:93 - 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:-0.005090891849249601 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.005090891849249601 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:145.69140625 - response_length/max:493.0 - response_length/min:99.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:145.69140625 - response_length_non_aborted/max:493.0 - response_length_non_aborted/min:99.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:266.46875 - prompt_length/max:709.0 - prompt_length/min:167.0 - 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:0.00016945414245128632 - timing_s/agent_loop/generate_sequences/min:np.float64(1.058131204918027) - timing_s/agent_loop/generate_sequences/max:np.float64(3.6493652053177357) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.6524944337506895) - timing_
|
||
|
|
step:94 - global_seqlen/min:11893 - global_seqlen/max:14952 - global_seqlen/minmax_diff:3059 - global_seqlen/balanced_min:12688 - global_seqlen/balanced_max:12691 - global_seqlen/mean:12690.125 - actor/entropy:0.05989615246653557 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1846369206905365 - training/rollout_probs_diff_mean:0.0010570499580353498 - training/rollout_probs_diff_std:0.006358841434121132 - training/rollout_actor_probs_pearson_corr:0.9988254308700562 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.93359375 - self_distillation/correct_sample_fraction:0.828125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.93359375 - rollout_corr/rollout_is_mean:0.9998717904090881 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:5.434635022538714e-05 - rollout_corr/rollout_is_max:1.421715497970581 - rollout_corr/rollout_is_min:0.47367003560066223 - rollout_corr/rollout_is_std:0.016109582036733627 - rollout_corr/rollout_is_eff_sample_size:0.9997405487090898 - rollout_corr/rollout_is_seq_mean:0.9998012781143188 - rollout_corr/rollout_is_seq_std:0.0014130610506981611 - rollout_corr/rollout_is_seq_max:1.0083123445510864 - rollout_corr/rollout_is_seq_min:0.9934720993041992 - rollout_corr/rollout_is_seq_max_deviation:0.008312344551086426 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0487608988769352) - rollout_corr/training_log_ppl:np.float64(0.044996698657683964) - rollout_corr/kl:np.float64(0.0003258662981671656) - rollout_corr/k3_kl:np.float64(0.0001271579087226359) - rollout_corr/rollout_ppl:np.float64(1.048406885471195) - rollout_corr/rollout_log_ppl:np.float64(0.04467083194712984) - rollout_corr/log_ppl_diff:np.float64(0.00032586671055412353) - rollout_corr/log_ppl_abs_diff:np.float64(0.0008906177959744355) - rollout_corr/log_ppl_diff_max:np.float64(0.007141053676605225) - rollout_corr/log_ppl_diff_min:np.float64(-0.006822258234024048) - rollout_corr/ppl_ratio:np.float64(1.0003270325250924) - rollout_corr/chi2_token:np.float64(-0.00016086967661976814) - rollout_corr/chi2_seq:np.float64(0.0033831519540399313) - actor/pg_loss:np.float64(0.014782077601310561) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.06640625) - actor/grad_norm:np.float64(0.2000967264175415) - perf/mfu/actor:np.float64(0.02368019906297236) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.7580337524414) - actor/lr:np.float64(1e-05) - training/global_step:94 - training/epoch:1 - 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:0.00041778755257837474 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00041778755257837474 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:143.75390625 - response_length/max:514.0 - response_length/min:93.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:143.75390625 - response_length_non_aborted/max:514.0 - response_length_non_aborted/min:93.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:252.8125 - prompt_length/max:384.0 - prompt_length/min:163.0 - 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:9.22325998544693e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1856862846761942) - timing_s/agent_loop/generate_sequences/max:np.float64(3.5140521097928286) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.6755123658149387) - tim
|
||
|
|
step:95 - global_seqlen/min:11330 - global_seqlen/max:16113 - global_seqlen/minmax_diff:4783 - global_seqlen/balanced_min:13170 - global_seqlen/balanced_max:13173 - global_seqlen/mean:13171.625 - actor/entropy:0.07470375299453735 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.19308215379714966 - training/rollout_probs_diff_mean:0.001270041218958795 - training/rollout_probs_diff_std:0.006702691316604614 - training/rollout_actor_probs_pearson_corr:0.998897910118103 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83984375 - self_distillation/correct_sample_fraction:0.7578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83984375 - rollout_corr/rollout_is_mean:0.9999818205833435 - rollout_corr/rollout_is_ratio_fraction_high:2.6317868105252273e-05 - rollout_corr/rollout_is_ratio_fraction_low:2.6317868105252273e-05 - rollout_corr/rollout_is_max:2.1670584678649902 - rollout_corr/rollout_is_min:0.2820621430873871 - rollout_corr/rollout_is_std:0.018377156928181648 - rollout_corr/rollout_is_eff_sample_size:0.9996621558413732 - rollout_corr/rollout_is_seq_mean:1.0000029802322388 - rollout_corr/rollout_is_seq_std:0.0015528063522651792 - rollout_corr/rollout_is_seq_max:1.0093673467636108 - rollout_corr/rollout_is_seq_min:0.9953069090843201 - rollout_corr/rollout_is_seq_max_deviation:0.00936734676361084 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.054156034719199) - rollout_corr/training_log_ppl:np.float64(0.049783984127770964) - rollout_corr/kl:np.float64(0.0001599594839012397) - rollout_corr/k3_kl:np.float64(0.00016291694669845437) - rollout_corr/rollout_ppl:np.float64(1.0539761930704117) - rollout_corr/rollout_log_ppl:np.float64(0.049624024586222504) - rollout_corr/log_ppl_diff:np.float64(0.00015995954154846004) - rollout_corr/log_ppl_abs_diff:np.float64(0.0010271767426388578) - rollout_corr/log_ppl_diff_max:np.float64(0.007684396579861641) - rollout_corr/log_ppl_diff_min:np.float64(-0.0061374930664896965) - rollout_corr/ppl_ratio:np.float64(1.000161218456924) - rollout_corr/chi2_token:np.float64(0.00033229216933250427) - rollout_corr/chi2_seq:np.float64(0.04464193223975599) - actor/pg_loss:np.float64(0.01729701646929982) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.16015625) - actor/grad_norm:np.float64(0.30493301153182983) - perf/mfu/actor:np.float64(0.024376464688528394) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.7802848815918) - actor/lr:np.float64(1e-05) - training/global_step:95 - training/epoch:1 - critic/score/mean:0.7578125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7578125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.003750296076759696 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.003750296076759696 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:148.42578125 - response_length/max:696.0 - response_length/min:91.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:148.42578125 - response_length_non_aborted/max:696.0 - response_length_non_aborted/min:91.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:263.1875 - prompt_length/max:646.0 - prompt_length/min:159.0 - 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:0.00014052912592887878 - timing_s/agent_loop/generate_sequences/min:np.float64(0.6790105514228344) - timing_s/agent_loop/generate_sequences/max:np.float64(4.464070187881589) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.46
|
||
|
|
step:96 - global_seqlen/min:11105 - global_seqlen/max:24436 - global_seqlen/minmax_diff:13331 - global_seqlen/balanced_min:13637 - global_seqlen/balanced_max:21185 - global_seqlen/mean:14582.5 - actor/entropy:0.058466240763664246 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1222105622291565 - training/rollout_probs_diff_mean:0.0009500277228653431 - training/rollout_probs_diff_std:0.005283925216645002 - training/rollout_actor_probs_pearson_corr:0.9991226196289062 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - self_distillation/correct_sample_fraction:0.671875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8359375 - rollout_corr/rollout_is_mean:0.999849796295166 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6028850078582764 - rollout_corr/rollout_is_min:0.6215155720710754 - rollout_corr/rollout_is_std:0.01449093408882618 - rollout_corr/rollout_is_eff_sample_size:0.9997899973421911 - rollout_corr/rollout_is_seq_mean:0.9998643398284912 - rollout_corr/rollout_is_seq_std:0.0010207281447947025 - rollout_corr/rollout_is_seq_max:1.0033302307128906 - rollout_corr/rollout_is_seq_min:0.9968570470809937 - rollout_corr/rollout_is_seq_max_deviation:0.003330230712890625 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0414116936735809) - rollout_corr/training_log_ppl:np.float64(0.03881705878433195) - rollout_corr/kl:np.float64(0.00023286626369850572) - rollout_corr/k3_kl:np.float64(9.725179917913023e-05) - rollout_corr/rollout_ppl:np.float64(1.0411630477756262) - rollout_corr/rollout_log_ppl:np.float64(0.03858419217473852) - rollout_corr/log_ppl_diff:np.float64(0.00023286660959342953) - rollout_corr/log_ppl_abs_diff:np.float64(0.0007177864165726078) - rollout_corr/log_ppl_diff_max:np.float64(0.003934789448976517) - rollout_corr/log_ppl_diff_min:np.float64(-0.003107704222202301) - rollout_corr/ppl_ratio:np.float64(1.0002334534656256) - rollout_corr/chi2_token:np.float64(-8.24218150228262e-05) - rollout_corr/chi2_seq:np.float64(-0.023605454713106155) - actor/pg_loss:np.float64(0.01246099645813814) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1640625) - actor/grad_norm:np.float64(0.18334656953811646) - perf/mfu/actor:np.float64(0.02569388592093621) - perf/max_memory_allocated_gb:np.float64(137.98988819122314) - perf/max_memory_reserved_gb:np.float64(142.8515625) - perf/cpu_memory_used_gb:np.float64(148.66317749023438) - actor/lr:np.float64(1e-05) - training/global_step:96 - training/epoch:1 - critic/score/mean:0.671875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.671875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.14236456155776978 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.14236456155776978 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:189.046875 - response_length/max:8192.0 - response_length/min:89.0 - response_length/clip_ratio:0.00390625 - response_length_non_aborted/mean:189.046875 - response_length_non_aborted/max:8192.0 - response_length_non_aborted/min:89.0 - response_length_non_aborted/clip_ratio:0.00390625 - response/aborted_ratio:0.0 - prompt_length/mean:266.65625 - prompt_length/max:456.0 - prompt_length/min:163.0 - 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:0.00012750737369060516 - timing_s/agent_loop/generate_sequences/min:np.float64(0.6305286791175604) - timing_s/agent_loop/generate_sequences/max:np.float64(46.55937415920198) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.532111374915985) - timing_s/agent_l
|
||
|
|
step:97 - global_seqlen/min:10947 - global_seqlen/max:26499 - global_seqlen/minmax_diff:15552 - global_seqlen/balanced_min:14157 - global_seqlen/balanced_max:21436 - global_seqlen/mean:16011.875 - actor/entropy:0.03894832357764244 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.14815488457679749 - training/rollout_probs_diff_mean:0.000644878251478076 - training/rollout_probs_diff_std:0.004595402628183365 - training/rollout_actor_probs_pearson_corr:0.9990875720977783 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8125 - self_distillation/correct_sample_fraction:0.6640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8125 - rollout_corr/rollout_is_mean:0.9999856352806091 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.617956519126892 - rollout_corr/rollout_is_min:0.6251919865608215 - rollout_corr/rollout_is_std:0.012909476645290852 - rollout_corr/rollout_is_eff_sample_size:0.9998331348432564 - rollout_corr/rollout_is_seq_mean:1.0000520944595337 - rollout_corr/rollout_is_seq_std:0.0010946537367999554 - rollout_corr/rollout_is_seq_max:1.0051692724227905 - rollout_corr/rollout_is_seq_min:0.9956714510917664 - rollout_corr/rollout_is_seq_max_deviation:0.005169272422790527 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0276606953702867) - rollout_corr/training_log_ppl:np.float64(0.02557707243022378) - rollout_corr/kl:np.float64(2.8707819488715813e-05) - rollout_corr/k3_kl:np.float64(8.088493853453094e-05) - rollout_corr/rollout_ppl:np.float64(1.0276206359267235) - rollout_corr/rollout_log_ppl:np.float64(0.025548364405963753) - rollout_corr/log_ppl_diff:np.float64(2.8708024260026832e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.000592460996074351) - rollout_corr/log_ppl_diff_max:np.float64(0.004720468074083328) - rollout_corr/log_ppl_diff_min:np.float64(-0.004352850839495659) - rollout_corr/ppl_ratio:np.float64(1.0000292733311653) - rollout_corr/chi2_token:np.float64(0.00027486286126077175) - rollout_corr/chi2_seq:np.float64(0.029070035088807344) - actor/pg_loss:np.float64(0.009027956928882475) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1875) - actor/grad_norm:np.float64(0.21440507471561432) - perf/mfu/actor:np.float64(0.028015629234339888) - perf/max_memory_allocated_gb:np.float64(138.14347791671753) - perf/max_memory_reserved_gb:np.float64(142.96875) - perf/cpu_memory_used_gb:np.float64(148.72295761108398) - actor/lr:np.float64(1e-05) - training/global_step:97 - training/epoch:1 - critic/score/mean:0.6640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.16916854679584503 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.16916854679584503 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:205.96484375 - response_length/max:8192.0 - response_length/min:86.0 - response_length/clip_ratio:0.0078125 - response_length_non_aborted/mean:205.96484375 - response_length_non_aborted/max:8192.0 - response_length_non_aborted/min:86.0 - response_length_non_aborted/clip_ratio:0.0078125 - response/aborted_ratio:0.0 - prompt_length/mean:294.40625 - prompt_length/max:900.0 - prompt_length/min:172.0 - 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:5.6277960538864136e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0710464902222157) - timing_s/agent_loop/generate_sequences/max:np.float64(50.111243879422545) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.9627161290554795) - timing_s/agent_loop
|
||
|
|
step:98 - global_seqlen/min:10943 - global_seqlen/max:18555 - global_seqlen/minmax_diff:7612 - global_seqlen/balanced_min:13886 - global_seqlen/balanced_max:21393 - global_seqlen/mean:14825.75 - actor/entropy:0.03268497809767723 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12233138084411621 - training/rollout_probs_diff_mean:0.0006073372205719352 - training/rollout_probs_diff_std:0.004555620718747377 - training/rollout_actor_probs_pearson_corr:0.9988921880722046 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.78125 - self_distillation/correct_sample_fraction:0.70703125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.78125 - rollout_corr/rollout_is_mean:0.9999187588691711 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.373380084463861e-05 - rollout_corr/rollout_is_max:1.3645859956741333 - rollout_corr/rollout_is_min:0.2873389720916748 - rollout_corr/rollout_is_std:0.012591598555445671 - rollout_corr/rollout_is_eff_sample_size:0.99984135760701 - rollout_corr/rollout_is_seq_mean:0.9999216794967651 - rollout_corr/rollout_is_seq_std:0.0010262156138196588 - rollout_corr/rollout_is_seq_max:1.0038503408432007 - rollout_corr/rollout_is_seq_min:0.9934375882148743 - rollout_corr/rollout_is_seq_max_deviation:0.006562411785125732 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0289837447926402) - rollout_corr/training_log_ppl:np.float64(0.02731430772119836) - rollout_corr/kl:np.float64(0.0001697021603940807) - rollout_corr/k3_kl:np.float64(9.143156329386218e-05) - rollout_corr/rollout_ppl:np.float64(1.0287985960021615) - rollout_corr/rollout_log_ppl:np.float64(0.02714460548383002) - rollout_corr/log_ppl_diff:np.float64(0.000169702237368341) - rollout_corr/log_ppl_abs_diff:np.float64(0.0005865523027104302) - rollout_corr/log_ppl_diff_max:np.float64(0.01138085126876831) - rollout_corr/log_ppl_diff_min:np.float64(-0.003382183611392975) - rollout_corr/ppl_ratio:np.float64(1.0001704336609691) - rollout_corr/chi2_token:np.float64(6.326939910650253e-06) - rollout_corr/chi2_seq:np.float64(-0.0013431634288281202) - actor/pg_loss:np.float64(0.009801261201893396) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.21875) - actor/grad_norm:np.float64(0.13937202095985413) - perf/mfu/actor:np.float64(0.026039030983783996) - perf/max_memory_allocated_gb:np.float64(138.14347791671753) - perf/max_memory_reserved_gb:np.float64(142.96875) - perf/cpu_memory_used_gb:np.float64(148.59864807128906) - actor/lr:np.float64(1e-05) - training/global_step:98 - training/epoch:1 - 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:-0.0011718564201146364 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0011718564201146364 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:164.5859375 - response_length/max:8192.0 - response_length/min:93.0 - response_length/clip_ratio:0.00390625 - response_length_non_aborted/mean:164.5859375 - response_length_non_aborted/max:8192.0 - response_length_non_aborted/min:93.0 - response_length_non_aborted/clip_ratio:0.00390625 - response/aborted_ratio:0.0 - prompt_length/mean:298.71875 - prompt_length/max:631.0 - prompt_length/min:166.0 - 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:0.00010897591710090637 - timing_s/agent_loop/generate_sequences/min:np.float64(0.6693439912050962) - timing_s/agent_loop/generate_sequences/max:np.float64(49.06370754353702) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.5679938126195339
|
||
|
|
step:99 - global_seqlen/min:11943 - global_seqlen/max:18408 - global_seqlen/minmax_diff:6465 - global_seqlen/balanced_min:14424 - global_seqlen/balanced_max:14426 - global_seqlen/mean:14425.0 - actor/entropy:0.04450289160013199 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12590056657791138 - training/rollout_probs_diff_mean:0.0008235365967266262 - training/rollout_probs_diff_std:0.00543550867587328 - training/rollout_actor_probs_pearson_corr:0.9988293051719666 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - self_distillation/correct_sample_fraction:0.68359375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8359375 - rollout_corr/rollout_is_mean:1.0000438690185547 - rollout_corr/rollout_is_ratio_fraction_high:5.64461515750736e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.5544400215148926 - rollout_corr/rollout_is_min:0.6363856196403503 - rollout_corr/rollout_is_std:0.015613551251590252 - rollout_corr/rollout_is_eff_sample_size:0.9997561572673596 - rollout_corr/rollout_is_seq_mean:1.0000444650650024 - rollout_corr/rollout_is_seq_std:0.0014577171532437205 - rollout_corr/rollout_is_seq_max:1.0118895769119263 - rollout_corr/rollout_is_seq_min:0.9962204694747925 - rollout_corr/rollout_is_seq_max_deviation:0.01188957691192627 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0298282350413501) - rollout_corr/training_log_ppl:np.float64(0.027603220796322603) - rollout_corr/kl:np.float64(6.449486278770067e-05) - rollout_corr/k3_kl:np.float64(0.00010897722906148127) - rollout_corr/rollout_ppl:np.float64(1.0297575378790498) - rollout_corr/rollout_log_ppl:np.float64(0.02753872562553994) - rollout_corr/log_ppl_diff:np.float64(6.449517078266354e-05) - rollout_corr/log_ppl_abs_diff:np.float64(0.0006765530931360786) - rollout_corr/log_ppl_diff_max:np.float64(0.004024788737297058) - rollout_corr/log_ppl_diff_min:np.float64(-0.007068036124110222) - rollout_corr/ppl_ratio:np.float64(1.0000652414746583) - rollout_corr/chi2_token:np.float64(0.0003731292672455311) - rollout_corr/chi2_seq:np.float64(0.0740649092476815) - actor/pg_loss:np.float64(0.007869484574584362) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1640625) - actor/grad_norm:np.float64(0.15554626286029816) - perf/mfu/actor:np.float64(0.026712613432520186) - perf/max_memory_allocated_gb:np.float64(138.14347791671753) - perf/max_memory_reserved_gb:np.float64(142.96875) - perf/cpu_memory_used_gb:np.float64(148.56956481933594) - actor/lr:np.float64(1e-05) - training/global_step:99 - training/epoch:1 - critic/score/mean:0.68359375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.68359375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0012135922443121672 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0012135922443121672 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:138.40625 - response_length/max:485.0 - response_length/min:87.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:138.40625 - response_length_non_aborted/max:485.0 - response_length_non_aborted/min:87.0 - response_length_non_aborted/clip_ratio:0.0 - response/aborted_ratio:0.0 - prompt_length/mean:312.375 - prompt_length/max:836.0 - prompt_length/min:170.0 - 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:0.00010474026203155518 - timing_s/agent_loop/generate_sequences/min:np.float64(0.6377026215195656) - timing_s/agent_loop/generate_sequences/max:np.float64(3.4349332191050053) - timing_s/agent_loop/generate_sequences/mean:np.float64(1.4134128085934208) - timing_s/ag
|
||
|
|
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 100}
|
||
|
|
validation generation end
|
||
|
|
[prompt] system
|
||
|
|
|
||
|
|
Given a question and four options, please select the right answer. Respond in the following format:
|
||
|
|
<reasoning>
|
||
|
|
...
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
...
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
For the answer, only output the letter corresponding to the correct option (A, B, C, or D), and nothing else. Do not restate the answer text. For example, if the answer is "A", just output:
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
|
||
|
|
user
|
||
|
|
Which of the following molecules is the only correct reactant used in the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" ?
|
||
|
|
|
||
|
|
A: O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1
|
||
|
|
B: O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1
|
||
|
|
C: O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1
|
||
|
|
D: O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F
|
||
|
|
Please reason step by step.
|
||
|
|
assistant
|
||
|
|
<think>
|
||
|
|
|
||
|
|
</think>
|
||
|
|
|
||
|
|
|
||
|
|
[response] <reasoning>
|
||
|
|
The product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" contains a nitrogen (N), a hydrocarbon chain, and a complex hydrocarbon chain with multiple functional groups. Option A contains a carboxylic acid group (C(=O)), a nitrogen (N), and matches the product's structure, including the hydrocarbon chain and the complex hydrocarbon chain with multiple functional groups. The other options either lack the carboxylic acid group or do not match the product's structure. Therefore, the correct answer is Option A.
|
||
|
|
</reasoning>
|
||
|
|
<answer>
|
||
|
|
A
|
||
|
|
</answer>
|
||
|
|
[ground_truth] A
|
||
|
|
[score] 1.0
|
||
|
|
[acc] 1.0
|
||
|
|
[pred] A
|
||
|
|
[incorrect_format] 1
|
||
|
|
[feedback]
|
||
|
|
local_global_step_folder: /workspace/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-SDPO_TR-Qwen-Qwen3-4B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_100
|
||
|
|
step:100 - global_seqlen/min:11694 - global_seqlen/max:32179 - global_seqlen/minmax_diff:20485 - global_seqlen/balanced_min:15112 - global_seqlen/balanced_max:22508 - global_seqlen/mean:16992.375 - actor/entropy:0.04114176332950592 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23105812072753906 - training/rollout_probs_diff_mean:0.000801601039711386 - training/rollout_probs_diff_std:0.005446349270641804 - training/rollout_actor_probs_pearson_corr:0.9987240433692932 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.8671875 - self_distillation/correct_sample_fraction:0.7265625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8671875 - rollout_corr/rollout_is_mean:1.0000271797180176 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.601178526878357 - rollout_corr/rollout_is_min:0.620167076587677 - rollout_corr/rollout_is_std:0.014016308821737766 - rollout_corr/rollout_is_eff_sample_size:0.9998034625161065 - rollout_corr/rollout_is_seq_mean:0.9999521374702454 - rollout_corr/rollout_is_seq_std:0.0012089080410078168 - rollout_corr/rollout_is_seq_max:1.0051069259643555 - rollout_corr/rollout_is_seq_min:0.9954840540885925 - rollout_corr/rollout_is_seq_max_deviation:0.005106925964355469 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.0374733386561275) - rollout_corr/training_log_ppl:np.float64(0.03557254603532556) - rollout_corr/kl:np.float64(0.00015642605447674374) - rollout_corr/k3_kl:np.float64(0.0001086825423562062) - rollout_corr/rollout_ppl:np.float64(1.0373146124184132) - rollout_corr/rollout_log_ppl:np.float64(0.035416119580048644) - rollout_corr/log_ppl_diff:np.float64(0.00015642645527691457) - rollout_corr/log_ppl_abs_diff:np.float64(0.0008119762612466275) - rollout_corr/log_ppl_diff_max:np.float64(0.004914548248052597) - rollout_corr/log_ppl_diff_min:np.float64(-0.0037901997566223145) - rollout_corr/ppl_ratio:np.float64(1.0001571618486196) - rollout_corr/chi2_token:np.float64(0.00012734439224004745) - rollout_corr/chi2_seq:np.float64(0.08643001760356128) - actor/pg_loss:np.float64(0.012388480715372907) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1328125) - actor/grad_norm:np.float64(0.1816544234752655) - perf/mfu/actor:np.float64(0.03008261803486565) - perf/max_memory_allocated_gb:np.float64(138.14347791671753) - perf/max_memory_reserved_gb:np.float64(142.96875) - perf/cpu_memory_used_gb:np.float64(148.59503936767578) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7491071428571429) - val-aux/sciknoweval/reward/std@16:np.float64(0.053113895809534975) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.771904761904762) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.0438263744703264) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.7262095238095239) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.046990424566282364) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7489761904761905) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.05319068044795249) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7891285714285714) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.029400016901477092) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.706704761904762) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.03441618216443395) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7514238095238096) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.04465209687151754) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7998666666666667) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.017334238666592075) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6937857142857142) - val-aux/sciknoweval
|
||
|
|
("Final validation metrics: {'val-aux/sciknoweval/reward/mean@16': "
|
||
|
|
"np.float64(0.7491071428571429), 'val-aux/sciknoweval/reward/std@16': "
|
||
|
|
"np.float64(0.053113895809534975), 'val-aux/sciknoweval/reward/best@2/mean': "
|
||
|
|
"np.float64(0.771904761904762), 'val-aux/sciknoweval/reward/best@2/std': "
|
||
|
|
"np.float64(0.0438263744703264), 'val-aux/sciknoweval/reward/worst@2/mean': "
|
||
|
|
"np.float64(0.7262095238095239), 'val-aux/sciknoweval/reward/worst@2/std': "
|
||
|
|
"np.float64(0.046990424566282364), 'val-aux/sciknoweval/reward/maj@2/mean': "
|
||
|
|
"np.float64(0.7489761904761905), 'val-aux/sciknoweval/reward/maj@2/std': "
|
||
|
|
"np.float64(0.05319068044795249), 'val-aux/sciknoweval/reward/best@4/mean': "
|
||
|
|
"np.float64(0.7891285714285714), 'val-aux/sciknoweval/reward/best@4/std': "
|
||
|
|
"np.float64(0.029400016901477092), 'val-aux/sciknoweval/reward/worst@4/mean': "
|
||
|
|
"np.float64(0.706704761904762), 'val-aux/sciknoweval/reward/worst@4/std': "
|
||
|
|
"np.float64(0.03441618216443395), 'val-aux/sciknoweval/reward/maj@4/mean': "
|
||
|
|
"np.float64(0.7514238095238096), 'val-aux/sciknoweval/reward/maj@4/std': "
|
||
|
|
"np.float64(0.04465209687151754), 'val-aux/sciknoweval/reward/best@8/mean': "
|
||
|
|
"np.float64(0.7998666666666667), 'val-aux/sciknoweval/reward/best@8/std': "
|
||
|
|
"np.float64(0.017334238666592075), 'val-aux/sciknoweval/reward/worst@8/mean': "
|
||
|
|
"np.float64(0.6937857142857142), 'val-aux/sciknoweval/reward/worst@8/std': "
|
||
|
|
"np.float64(0.02011337982070672), 'val-aux/sciknoweval/reward/maj@8/mean': "
|
||
|
|
"np.float64(0.7533142857142858), 'val-aux/sciknoweval/reward/maj@8/std': "
|
||
|
|
"np.float64(0.03611226257903519), 'val-aux/sciknoweval/reward/best@16/mean': "
|
||
|
|
"np.float64(0.8060142857142858), 'val-aux/sciknoweval/reward/best@16/std': "
|
||
|
|
'np.float64(0.008797318267434817), '
|
||
|
|
"'val-aux/sciknoweval/reward/worst@16/mean': np.float64(0.6871380952380951), "
|
||
|
|
"'val-aux/sciknoweval/reward/worst@16/std': np.float64(0.010972447210233164), "
|
||
|
|
"'val-aux/sciknoweval/reward/maj@16/mean': np.float64(0.7554761904761905), "
|
||
|
|
"'val-aux/sciknoweval/reward/maj@16/std': np.float64(0.028181641738688423), "
|
||
|
|
"'val-aux/sciknoweval/score/mean@16': np.float64(0.7491071428571429), "
|
||
|
|
"'val-aux/sciknoweval/score/std@16': np.float64(0.053113895809534975), "
|
||
|
|
"'val-aux/sciknoweval/score/best@2/mean': np.float64(0.771904761904762), "
|
||
|
|
"'val-aux/sciknoweval/score/best@2/std': np.float64(0.0438263744703264), "
|
||
|
|
"'val-aux/sciknoweval/score/worst@2/mean': np.float64(0.7262095238095239), "
|
||
|
|
"'val-aux/sciknoweval/score/worst@2/std': np.float64(0.046990424566282364), "
|
||
|
|
"'val-aux/sciknoweval/score/maj@2/mean': np.float64(0.7489761904761905), "
|
||
|
|
"'val-aux/sciknoweval/score/maj@2/std': np.float64(0.05319068044795249), "
|
||
|
|
"'val-aux/sciknoweval/score/best@4/mean': np.float64(0.7891285714285714), "
|
||
|
|
"'val-aux/sciknoweval/score/best@4/std': np.float64(0.029400016901477092), "
|
||
|
|
"'val-aux/sciknoweval/score/worst@4/mean': np.float64(0.706704761904762), "
|
||
|
|
"'val-aux/sciknoweval/score/worst@4/std': np.float64(0.03441618216443395), "
|
||
|
|
"'val-aux/sciknoweval/score/maj@4/mean': np.float64(0.7514238095238096), "
|
||
|
|
"'val-aux/sciknoweval/score/maj@4/std': np.float64(0.04465209687151754), "
|
||
|
|
"'val-aux/sciknoweval/score/best@8/mean': np.float64(0.7998666666666667), "
|
||
|
|
"'val-aux/sciknoweval/score/best@8/std': np.float64(0.017334238666592075), "
|
||
|
|
"'val-aux/sciknoweval/score/worst@8/mean': np.float64(0.6937857142857142), "
|
||
|
|
"'val-aux/sciknoweval/score/worst@8/std': np.float64(0.02011337982070672), "
|
||
|
|
"'val-aux/sciknoweval/score/maj@8/mean': np.float64(0.7533142857142858), "
|
||
|
|
"'val-aux/sciknoweval/score/maj@8/std': np.float64(0.03611226257903519), "
|
||
|
|
"'val-aux/sciknoweval/score/best@16/mean': np.float64(0.8060142857142858), "
|
||
|
|
"'val-aux/sciknoweval/score/best@16/std': np.float64(0.008797318267434817), "
|
||
|
|
"'val-aux/sciknoweval/score/worst@16/mean': np.float64(0.6871380952380951), "
|
||
|
|
"'val-aux/sciknoweval/score/worst@16/std': np.float64(0.010972447210233164), "
|
||
|
|
"'val-aux/sciknoweval/score/maj@16/mean': np.float64(0.7554761904761905), "
|
||
|
|
"'val-aux/sciknoweval/score/maj@16/std': np.float64(0.028181641738688423), "
|
||
|
|
"'val-core/sciknoweval/acc/mean@16': np.float64(0.7491071428571429), "
|
||
|
|
"'val-aux/sciknoweval/acc/std@16': np.float64(0.053113895809534975), "
|
||
|
|
"'val-aux/sciknoweval/acc/best@2/mean': np.float64(0.771904761904762), "
|
||
|
|
"'val-aux/sciknoweval/acc/best@2/std': np.float64(0.0438263744703264), "
|
||
|
|
"'val-aux/sciknoweval/acc/worst@2/mean': np.float64(0.7262095238095239), "
|
||
|
|
"'val-aux/sciknoweval/acc/worst@2/std': np.float64(0.046990424566282364), "
|
||
|
|
"'val-aux/sciknoweval/acc/maj@2/mean': np.float64(0.7489761904761905), "
|
||
|
|
"'val-aux/sciknoweval/acc/maj@2/std': np.float64(0.05319068044795249), "
|
||
|
|
"'val-aux/sciknoweval/acc/best@4/mean': np.float64(0.7891285714285714), "
|
||
|
|
"'val-aux/sciknoweval/acc/best@4/std': np.float64(0.029400016901477092), "
|
||
|
|
"'val-aux/sciknoweval/acc/worst@4/mean': np.float64(0.706704761904762), "
|
||
|
|
"'val-aux/sciknoweval/acc/worst@4/std': np.float64(0.03441618216443395), "
|
||
|
|
"'val-aux/sciknoweval/acc/maj@4/mean': np.float64(0.7514238095238096), "
|
||
|
|
"'val-aux/sciknoweval/acc/maj@4/std': np.float64(0.04465209687151754), "
|
||
|
|
"'val-aux/sciknoweval/acc/best@8/mean': np.float64(0.7998666666666667), "
|
||
|
|
"'val-aux/sciknoweval/acc/best@8/std': np.float64(0.017334238666592075), "
|
||
|
|
"'val-aux/sciknoweval/acc/worst@8/mean': np.float64(0.6937857142857142), "
|
||
|
|
"'val-aux/sciknoweval/acc/worst@8/std': np.float64(0.02011337982070672), "
|
||
|
|
"'val-aux/sciknoweval/acc/maj@8/mean': np.float64(0.7533142857142858), "
|
||
|
|
"'val-aux/sciknoweval/acc/maj@8/std': np.float64(0.03611226257903519), "
|
||
|
|
"'val-core/sciknoweval/acc/best@16/mean': np.float64(0.8060142857142858), "
|
||
|
|
"'val-core/sciknoweval/acc/best@16/std': np.float64(0.008797318267434817), "
|
||
|
|
"'val-aux/sciknoweval/acc/worst@16/mean': np.float64(0.6871380952380951), "
|
||
|
|
"'val-aux/sciknoweval/acc/worst@16/std': np.float64(0.010972447210233164), "
|
||
|
|
"'val-core/sciknoweval/acc/maj@16/mean': np.float64(0.7554761904761905), "
|
||
|
|
"'val-core/sciknoweval/acc/maj@16/std': np.float64(0.028181641738688423), "
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/mean@16': "
|
||
|
|
'np.float64(0.999702380952381), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/std@16': "
|
||
|
|
'np.float64(0.0011526736149426837), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/best@2/mean': "
|
||
|
|
'np.float64(0.9999857142857144), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/best@2/std': "
|
||
|
|
'np.float64(0.0002604287412166728), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/worst@2/mean': "
|
||
|
|
'np.float64(0.9994238095238095), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/worst@2/std': "
|
||
|
|
'np.float64(0.00155298702746771), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/maj@2/mean': "
|
||
|
|
'np.float64(0.999652380952381), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/maj@2/std': "
|
||
|
|
'np.float64(0.001238745250613849), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/best@4/mean': np.float64(1.0), "
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/best@4/std': np.float64(0.0), "
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/worst@4/mean': "
|
||
|
|
'np.float64(0.9989190476190476), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/worst@4/std': "
|
||
|
|
'np.float64(0.0019947266079247076), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/maj@4/mean': "
|
||
|
|
'np.float64(0.9999333333333333), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/maj@4/std': "
|
||
|
|
'np.float64(0.0005594782149609339), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/best@8/mean': np.float64(1.0), "
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/best@8/std': np.float64(0.0), "
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/worst@8/mean': "
|
||
|
|
'np.float64(0.9981047619047619), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/worst@8/std': "
|
||
|
|
'np.float64(0.002330883067212054), '
|
||
|
|
"'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(0.9969142857142858), '
|
||
|
|
"'val-aux/sciknoweval/incorrect_format/worst@16/std': "
|
||
|
|
'np.float64(0.002274256999100845), '
|
||
|
|
"'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)}")
|