807 lines
433 KiB
Plaintext
807 lines
433 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-Qwen3-8B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/latest_checkpointed_iteration.txt
|
|
Training from scratch
|
|
Training Progress: 9%|▉ | 9/100 [07:44<1:07:18, 44.38s/it]
|
|
step:1 - global_seqlen/min:18264 - global_seqlen/max:25243 - global_seqlen/minmax_diff:6979 - global_seqlen/balanced_min:22594 - global_seqlen/balanced_max:22761 - global_seqlen/mean:22624.125 - actor/entropy:0.5361832976341248 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.21512234210968018 - training/rollout_probs_diff_mean:0.004742587450891733 - training/rollout_probs_diff_std:0.00959940068423748 - training/rollout_actor_probs_pearson_corr:0.9994576573371887 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.7734375 - 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:0.7734375 - rollout_corr/rollout_is_mean:1.0001872777938843 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.599510669708252 - rollout_corr/rollout_is_min:0.6091659069061279 - rollout_corr/rollout_is_std:0.03540123254060745 - rollout_corr/rollout_is_eff_sample_size:0.9987486781350137 - rollout_corr/rollout_is_seq_mean:1.0002036094665527 - rollout_corr/rollout_is_seq_std:0.0017308281967416406 - rollout_corr/rollout_is_seq_max:1.0046968460083008 - rollout_corr/rollout_is_seq_min:0.9944801330566406 - rollout_corr/rollout_is_seq_max_deviation:0.005519866943359375 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7683840207755566) - rollout_corr/training_log_ppl:np.float64(0.5554063443560153) - rollout_corr/kl:np.float64(0.00046359800026252174) - rollout_corr/k3_kl:np.float64(0.0006672355027603771) - rollout_corr/rollout_ppl:np.float64(1.767560938373208) - rollout_corr/rollout_log_ppl:np.float64(0.5549427493824624) - rollout_corr/log_ppl_diff:np.float64(0.0004635949735529721) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013674245565198362) - rollout_corr/log_ppl_diff_max:np.float64(0.006195336580276489) - rollout_corr/log_ppl_diff_min:np.float64(-0.0038204193115234375) - rollout_corr/ppl_ratio:np.float64(1.0004652007482946) - rollout_corr/chi2_token:np.float64(0.0017550343181937933) - rollout_corr/chi2_seq:np.float64(0.7015784336254001) - actor/pg_loss:np.float64(0.03477558112172119) - 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(0.38673874735832214) - perf/mfu/actor:np.float64(0.03807273644076217) - perf/max_memory_allocated_gb:np.float64(127.42017793655396) - perf/max_memory_reserved_gb:np.float64(140.970703125) - perf/cpu_memory_used_gb:np.float64(209.70529651641846) - actor/lr:np.float64(0.0) - training/global_step:1 - 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:7.612336503370898e-06 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:7.612336503370898e-06 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:449.00390625 - response_length/max:1567.0 - response_length/min:174.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:449.00390625 - response_length_non_aborted/max:1567.0 - response_length_non_aborted/min:174.0 - 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:0.00023845955729484558 - timing_s/agent_loop/generate_sequences/min:np.float64(12.658092947676778) - timing_s/agent_loop/generate_sequences/max:np.float64(26.287752015516162) - timing_s/agent_loop/generate_sequences/mean:np.float64(18.09391785164189) - timing_s/agent_loop/tool_calls/mi
|
|
step:2 - global_seqlen/min:18232 - global_seqlen/max:29203 - global_seqlen/minmax_diff:10971 - global_seqlen/balanced_min:24893 - global_seqlen/balanced_max:25032 - global_seqlen/mean:24924.375 - actor/entropy:0.5580510497093201 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1991278976202011 - training/rollout_probs_diff_mean:0.004670199006795883 - training/rollout_probs_diff_std:0.009616076946258545 - training/rollout_actor_probs_pearson_corr:0.9994807243347168 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.76171875 - self_distillation/correct_sample_fraction:0.40234375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.76171875 - rollout_corr/rollout_is_mean:0.9999762177467346 - rollout_corr/rollout_is_ratio_fraction_high:8.502605851390399e-06 - rollout_corr/rollout_is_ratio_fraction_low:2.5507817554171197e-05 - rollout_corr/rollout_is_max:2.457724094390869 - rollout_corr/rollout_is_min:0.4281544089317322 - rollout_corr/rollout_is_std:0.03536585718393326 - rollout_corr/rollout_is_eff_sample_size:0.9987505807168433 - rollout_corr/rollout_is_seq_mean:0.9999352097511292 - rollout_corr/rollout_is_seq_std:0.0017190161161124706 - rollout_corr/rollout_is_seq_max:1.0061914920806885 - rollout_corr/rollout_is_seq_min:0.9952245950698853 - rollout_corr/rollout_is_seq_max_deviation:0.0061914920806884766 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.8340089046396315) - rollout_corr/training_log_ppl:np.float64(0.5909716563182883) - rollout_corr/kl:np.float64(0.0007465766247918282) - rollout_corr/k3_kl:np.float64(0.0006817502627995964) - rollout_corr/rollout_ppl:np.float64(1.832611644640565) - rollout_corr/rollout_log_ppl:np.float64(0.590225073392503) - rollout_corr/log_ppl_diff:np.float64(0.0007465829257853329) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014643348404206336) - rollout_corr/log_ppl_diff_max:np.float64(0.005464375019073486) - rollout_corr/log_ppl_diff_min:np.float64(-0.005249321460723877) - rollout_corr/ppl_ratio:np.float64(1.0007483414374292) - rollout_corr/chi2_token:np.float64(0.0012394185177981853) - rollout_corr/chi2_seq:np.float64(0.7276081086602062) - actor/pg_loss:np.float64(0.024596926054073265) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.23828125) - actor/grad_norm:np.float64(0.43673500418663025) - perf/mfu/actor:np.float64(0.05571106418206393) - perf/max_memory_allocated_gb:np.float64(132.06460809707642) - perf/max_memory_reserved_gb:np.float64(142.73828125) - perf/cpu_memory_used_gb:np.float64(199.32543182373047) - actor/lr:np.float64(1.0000000000000002e-06) - training/global_step:2 - training/epoch:0 - critic/score/mean:0.40234375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.40234375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.007014649920165539 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.007014649920165539 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:459.41796875 - response_length/max:1856.0 - response_length/min:147.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:459.41796875 - response_length_non_aborted/max:1856.0 - response_length_non_aborted/min:147.0 - 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:0.00013169273734092712 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9460461046546698) - timing_s/agent_loop/generate_sequences/max:np.float64(14.567798275500536) - timing_s/agent_loop/generate_sequences/mean:np
|
|
step:3 - global_seqlen/min:19687 - global_seqlen/max:26150 - global_seqlen/minmax_diff:6463 - global_seqlen/balanced_min:23902 - global_seqlen/balanced_max:24292 - global_seqlen/mean:23962.125 - actor/entropy:0.5360055565834045 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3522404730319977 - training/rollout_probs_diff_mean:0.004774059634655714 - training/rollout_probs_diff_std:0.009763491339981556 - training/rollout_actor_probs_pearson_corr:0.9994365572929382 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.72265625 - 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:0.72265625 - rollout_corr/rollout_is_mean:0.9999971389770508 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6585496664047241 - rollout_corr/rollout_is_min:0.5071716904640198 - rollout_corr/rollout_is_std:0.03536417335271835 - rollout_corr/rollout_is_eff_sample_size:0.9987509374517434 - rollout_corr/rollout_is_seq_mean:0.9999326467514038 - rollout_corr/rollout_is_seq_std:0.001957420026883483 - rollout_corr/rollout_is_seq_max:1.0059623718261719 - rollout_corr/rollout_is_seq_min:0.9935569763183594 - rollout_corr/rollout_is_seq_max_deviation:0.006443023681640625 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.783161018975079) - rollout_corr/training_log_ppl:np.float64(0.5628575202426873) - rollout_corr/kl:np.float64(0.0007382586350743736) - rollout_corr/k3_kl:np.float64(0.0006708653914984097) - rollout_corr/rollout_ppl:np.float64(1.7817814219743013) - rollout_corr/rollout_log_ppl:np.float64(0.5621192620019428) - rollout_corr/log_ppl_diff:np.float64(0.0007382582407444715) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015515966806560755) - rollout_corr/log_ppl_diff_max:np.float64(0.007107377052307129) - rollout_corr/log_ppl_diff_min:np.float64(-0.005211293697357178) - rollout_corr/ppl_ratio:np.float64(1.0007404882926494) - rollout_corr/chi2_token:np.float64(0.0012027209158986807) - rollout_corr/chi2_seq:np.float64(1.2020828784443438) - actor/pg_loss:np.float64(0.02400384763313923) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.27734375) - actor/grad_norm:np.float64(0.37456464767456055) - perf/mfu/actor:np.float64(0.053126952495109604) - perf/max_memory_allocated_gb:np.float64(132.06460809707642) - perf/max_memory_reserved_gb:np.float64(142.828125) - perf/cpu_memory_used_gb:np.float64(195.88725662231445) - actor/lr:np.float64(2.0000000000000003e-06) - training/global_step:3 - 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:0.010932918637990952 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.010932918637990952 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:446.34765625 - response_length/max:1700.0 - response_length/min:125.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:446.34765625 - response_length_non_aborted/max:1700.0 - response_length_non_aborted/min:125.0 - 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:0.00017523393034934998 - timing_s/agent_loop/generate_sequences/min:np.float64(1.6542972903698683) - timing_s/agent_loop/generate_sequences/max:np.float64(13.097854308784008) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.294624231639318) - timing_s/agent_l
|
|
step:4 - global_seqlen/min:21661 - global_seqlen/max:25248 - global_seqlen/minmax_diff:3587 - global_seqlen/balanced_min:23262 - global_seqlen/balanced_max:23434 - global_seqlen/mean:23298.25 - actor/entropy:0.5366904735565186 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.3571394979953766 - training/rollout_probs_diff_mean:0.004756357055157423 - training/rollout_probs_diff_std:0.010087710805237293 - training/rollout_actor_probs_pearson_corr:0.999401330947876 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.76171875 - 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:0.76171875 - rollout_corr/rollout_is_mean:0.9999596476554871 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.7812286134576425e-05 - rollout_corr/rollout_is_max:1.7879111766815186 - rollout_corr/rollout_is_min:0.25212857127189636 - rollout_corr/rollout_is_std:0.03586458042263985 - rollout_corr/rollout_is_eff_sample_size:0.9987151463199753 - rollout_corr/rollout_is_seq_mean:0.9998703598976135 - rollout_corr/rollout_is_seq_std:0.001910206861793995 - rollout_corr/rollout_is_seq_max:1.0045236349105835 - rollout_corr/rollout_is_seq_min:0.993385374546051 - rollout_corr/rollout_is_seq_max_deviation:0.006614625453948975 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7895527575165033) - rollout_corr/training_log_ppl:np.float64(0.5654744401108474) - rollout_corr/kl:np.float64(0.0008311298718308713) - rollout_corr/k3_kl:np.float64(0.0007015044729428155) - rollout_corr/rollout_ppl:np.float64(1.7880147523246706) - rollout_corr/rollout_log_ppl:np.float64(0.5646433079382405) - rollout_corr/log_ppl_diff:np.float64(0.0008311321726068854) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016444014618173242) - rollout_corr/log_ppl_diff_max:np.float64(0.0075508952140808105) - rollout_corr/log_ppl_diff_min:np.float64(-0.0037292838096618652) - rollout_corr/ppl_ratio:np.float64(1.0008334126323462) - rollout_corr/chi2_token:np.float64(0.0011358766350895166) - rollout_corr/chi2_seq:np.float64(1.3665155000053346) - actor/pg_loss:np.float64(0.02486543896702642) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.23828125) - actor/grad_norm:np.float64(0.3548567295074463) - perf/mfu/actor:np.float64(0.052566123055033456) - perf/max_memory_allocated_gb:np.float64(132.06460809707642) - perf/max_memory_reserved_gb:np.float64(142.828125) - perf/cpu_memory_used_gb:np.float64(193.57017135620117) - actor/lr:np.float64(3e-06) - training/global_step:4 - 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:0.009384328499436378 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.009384328499436378 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:421.3515625 - response_length/max:1318.0 - response_length/min:137.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:421.3515625 - response_length_non_aborted/max:1318.0 - response_length_non_aborted/min:137.0 - 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:0.00012049823999404907 - timing_s/agent_loop/generate_sequences/min:np.float64(1.7902468871325254) - timing_s/agent_loop/generate_sequences/max:np.float64(10.965458948165178) - timing_s/agent_loop/generate_sequences/mean:np.float64(5.072465981749701) - timing_s/a
|
|
step:5 - global_seqlen/min:15085 - global_seqlen/max:25151 - global_seqlen/minmax_diff:10066 - global_seqlen/balanced_min:19102 - global_seqlen/balanced_max:26220 - global_seqlen/mean:19994.375 - actor/entropy:0.48705151677131653 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.19511598348617554 - training/rollout_probs_diff_mean:0.00506820622831583 - training/rollout_probs_diff_std:0.010638521984219551 - training/rollout_actor_probs_pearson_corr:0.9992846846580505 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8203125 - 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:0.8203125 - rollout_corr/rollout_is_mean:0.9995800256729126 - rollout_corr/rollout_is_ratio_fraction_high:1.1123594049422536e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.1123594049422536e-05 - rollout_corr/rollout_is_max:2.0103986263275146 - rollout_corr/rollout_is_min:0.47443971037864685 - rollout_corr/rollout_is_std:0.03754768520593643 - rollout_corr/rollout_is_eff_sample_size:0.9985910268552471 - rollout_corr/rollout_is_seq_mean:0.9995495676994324 - rollout_corr/rollout_is_seq_std:0.0024097918067127466 - rollout_corr/rollout_is_seq_max:1.0074304342269897 - rollout_corr/rollout_is_seq_min:0.9922919273376465 - rollout_corr/rollout_is_seq_max_deviation:0.007708072662353516 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7421766105107963) - rollout_corr/training_log_ppl:np.float64(0.5444306683421019) - rollout_corr/kl:np.float64(0.0012648270819290985) - rollout_corr/k3_kl:np.float64(0.0008143815021526279) - rollout_corr/rollout_ppl:np.float64(1.7398993847891688) - rollout_corr/rollout_log_ppl:np.float64(0.5431658360466827) - rollout_corr/log_ppl_diff:np.float64(0.0012648322954191826) - rollout_corr/log_ppl_abs_diff:np.float64(0.002076018681691494) - rollout_corr/log_ppl_diff_max:np.float64(0.00910496711730957) - rollout_corr/log_ppl_diff_min:np.float64(-0.006477653980255127) - rollout_corr/ppl_ratio:np.float64(1.0012685591354966) - rollout_corr/chi2_token:np.float64(0.0007255298551172018) - rollout_corr/chi2_seq:np.float64(0.14311646949499846) - actor/pg_loss:np.float64(0.03172439633999602) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1796875) - actor/grad_norm:np.float64(0.32942453026771545) - perf/mfu/actor:np.float64(0.04376792762511619) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(192.3722686767578) - 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:0.004998665302991867 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.004998665302991867 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:351.16796875 - response_length/max:8192.0 - response_length/min:78.0 - response_length/clip_ratio:0.00390625 - response_length_non_aborted/mean:351.16796875 - response_length_non_aborted/max:8192.0 - response_length_non_aborted/min:78.0 - response_length_non_aborted/clip_ratio:0.00390625 - 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:0.00011280179023742676 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0356590189039707) - timing_s/agent_loop/generate_sequences/max:np.float64(58.78317418880761) - timing_s/agent_loop/generate_sequences/m
|
|
step:6 - global_seqlen/min:14030 - global_seqlen/max:18222 - global_seqlen/minmax_diff:4192 - global_seqlen/balanced_min:16579 - global_seqlen/balanced_max:16612 - global_seqlen/mean:16584.75 - actor/entropy:0.48602938652038574 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.202138751745224 - training/rollout_probs_diff_mean:0.006041024345904589 - training/rollout_probs_diff_std:0.012362598441541195 - training/rollout_actor_probs_pearson_corr:0.9990302920341492 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.7109375 - self_distillation/correct_sample_fraction:0.43359375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.7109375 - rollout_corr/rollout_is_mean:1.000019907951355 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.8505375385284424 - rollout_corr/rollout_is_min:0.5421095490455627 - rollout_corr/rollout_is_std:0.04186709225177765 - rollout_corr/rollout_is_eff_sample_size:0.998250213725255 - rollout_corr/rollout_is_seq_mean:1.0000016689300537 - rollout_corr/rollout_is_seq_std:0.0029089609161019325 - rollout_corr/rollout_is_seq_max:1.009252905845642 - rollout_corr/rollout_is_seq_min:0.9885916709899902 - rollout_corr/rollout_is_seq_max_deviation:0.011408329010009766 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6425856784917414) - rollout_corr/training_log_ppl:np.float64(0.4853546608937904) - rollout_corr/kl:np.float64(0.0009002109859856944) - rollout_corr/k3_kl:np.float64(0.000901849874594518) - rollout_corr/rollout_ppl:np.float64(1.6410169932059944) - rollout_corr/rollout_log_ppl:np.float64(0.48445445025572553) - rollout_corr/log_ppl_diff:np.float64(0.0009002106380648911) - rollout_corr/log_ppl_abs_diff:np.float64(0.0022630870225839317) - rollout_corr/log_ppl_diff_max:np.float64(0.012603789567947388) - rollout_corr/log_ppl_diff_min:np.float64(-0.00795358419418335) - rollout_corr/ppl_ratio:np.float64(1.000904877204448) - rollout_corr/chi2_token:np.float64(0.001827634871006012) - rollout_corr/chi2_seq:np.float64(0.3302309727296233) - actor/pg_loss:np.float64(0.03024700687819859) - 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(0.400918573141098) - perf/mfu/actor:np.float64(0.038789031825860965) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(192.33726501464844) - actor/lr:np.float64(5e-06) - training/global_step:6 - training/epoch:0 - critic/score/mean:0.43359375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.43359375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.002675791969522834 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.002675791969522834 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:243.4296875 - response_length/max:770.0 - response_length/min:106.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:243.4296875 - response_length_non_aborted/max:770.0 - response_length_non_aborted/min:106.0 - 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:0.00013294443488121033 - timing_s/agent_loop/generate_sequences/min:np.float64(1.4515924919396639) - timing_s/agent_loop/generate_sequences/max:np.float64(6.349148690700531) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.0092627580088447) - timing_s/agent_loop/tool_calls/min:np
|
|
step:7 - global_seqlen/min:13868 - global_seqlen/max:19514 - global_seqlen/minmax_diff:5646 - global_seqlen/balanced_min:16944 - global_seqlen/balanced_max:16945 - global_seqlen/mean:16944.625 - actor/entropy:0.5125228762626648 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.33189502358436584 - training/rollout_probs_diff_mean:0.00589430658146739 - training/rollout_probs_diff_std:0.01203934382647276 - training/rollout_actor_probs_pearson_corr:0.9991406202316284 - self_distillation/success_group_fraction:0.53125 - self_distillation/success_sample_fraction:0.5234375 - self_distillation/correct_sample_fraction:0.31640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.5234375 - rollout_corr/rollout_is_mean:0.9998487830162048 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:4.872265344602056e-05 - rollout_corr/rollout_is_max:1.7312734127044678 - rollout_corr/rollout_is_min:0.25963935256004333 - rollout_corr/rollout_is_std:0.0410112664103508 - rollout_corr/rollout_is_eff_sample_size:0.9983204250147065 - rollout_corr/rollout_is_seq_mean:0.9998657703399658 - rollout_corr/rollout_is_seq_std:0.002865270711481571 - rollout_corr/rollout_is_seq_max:1.0084131956100464 - rollout_corr/rollout_is_seq_min:0.9913281798362732 - rollout_corr/rollout_is_seq_max_deviation:0.008671820163726807 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.7074019988067448) - rollout_corr/training_log_ppl:np.float64(0.5222927427967079) - rollout_corr/kl:np.float64(0.001015672506596843) - rollout_corr/k3_kl:np.float64(0.0008813646290946053) - rollout_corr/rollout_ppl:np.float64(1.7056407472118735) - rollout_corr/rollout_log_ppl:np.float64(0.5212770731304772) - rollout_corr/log_ppl_diff:np.float64(0.0010156696662306786) - rollout_corr/log_ppl_abs_diff:np.float64(0.0024303774116560817) - rollout_corr/log_ppl_diff_max:np.float64(0.009511113166809082) - rollout_corr/log_ppl_diff_min:np.float64(-0.0070722103118896484) - rollout_corr/ppl_ratio:np.float64(1.0010203716810793) - rollout_corr/chi2_token:np.float64(0.0014786615502089262) - rollout_corr/chi2_seq:np.float64(0.3757970060687512) - actor/pg_loss:np.float64(0.021978582000883762) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.4765625) - actor/grad_norm:np.float64(0.2870350778102875) - perf/mfu/actor:np.float64(0.03992257046207755) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(192.2972068786621) - actor/lr:np.float64(6e-06) - training/global_step:7 - training/epoch:0 - critic/score/mean:0.31640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.31640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.005625436548143625 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.005625436548143625 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:240.51953125 - response_length/max:503.0 - response_length/min:101.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:240.51953125 - response_length_non_aborted/max:503.0 - response_length_non_aborted/min:101.0 - 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:0.0001263655722141266 - timing_s/agent_loop/generate_sequences/min:np.float64(1.3591309823095798) - timing_s/agent_loop/generate_sequences/max:np.float64(4.831638224422932) - timing_s/agent_loop/generate_sequences/mean:np.float64(2.944226457235345) - timing_s/agent_loo
|
|
step:8 - global_seqlen/min:17020 - global_seqlen/max:23126 - global_seqlen/minmax_diff:6106 - global_seqlen/balanced_min:19001 - global_seqlen/balanced_max:19003 - global_seqlen/mean:19001.75 - actor/entropy:0.6210432648658752 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.19370514154434204 - training/rollout_probs_diff_mean:0.005355164408683777 - training/rollout_probs_diff_std:0.010168607346713543 - training/rollout_actor_probs_pearson_corr:0.9994337558746338 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.765625 - self_distillation/correct_sample_fraction:0.51953125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.765625 - rollout_corr/rollout_is_mean:0.9999929070472717 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.2965796486241743e-05 - rollout_corr/rollout_is_max:1.6241605281829834 - rollout_corr/rollout_is_min:0.4899890422821045 - rollout_corr/rollout_is_std:0.03794168308377266 - rollout_corr/rollout_is_eff_sample_size:0.9985624980061147 - rollout_corr/rollout_is_seq_mean:1.0000338554382324 - rollout_corr/rollout_is_seq_std:0.002568972995504737 - rollout_corr/rollout_is_seq_max:1.0126250982284546 - rollout_corr/rollout_is_seq_min:0.9924743175506592 - rollout_corr/rollout_is_seq_max_deviation:0.01262509822845459 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.929349189158529) - rollout_corr/training_log_ppl:np.float64(0.6381971901864745) - rollout_corr/kl:np.float64(0.0007086855434117467) - rollout_corr/k3_kl:np.float64(0.0007425311049473748) - rollout_corr/rollout_ppl:np.float64(1.928023083601147) - rollout_corr/rollout_log_ppl:np.float64(0.6374885056866333) - rollout_corr/log_ppl_diff:np.float64(0.0007086844998411834) - rollout_corr/log_ppl_abs_diff:np.float64(0.0020769419497810304) - rollout_corr/log_ppl_diff_max:np.float64(0.008784592151641846) - rollout_corr/log_ppl_diff_min:np.float64(-0.01150214672088623) - rollout_corr/ppl_ratio:np.float64(1.0007122184615582) - rollout_corr/chi2_token:np.float64(0.0015567950904369354) - rollout_corr/chi2_seq:np.float64(0.9907183169852942) - actor/pg_loss:np.float64(0.03114511050080182) - 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(0.3413833975791931) - perf/mfu/actor:np.float64(0.04412715206215493) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(192.4750714302063) - actor/lr:np.float64(7e-06) - training/global_step:8 - training/epoch:0 - critic/score/mean:0.51953125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.51953125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0018038663547486067 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0018038663547486067 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:301.2734375 - response_length/max:711.0 - response_length/min:158.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:301.2734375 - response_length_non_aborted/max:711.0 - response_length_non_aborted/min:158.0 - 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:0.00014099478721618652 - timing_s/agent_loop/generate_sequences/min:np.float64(2.074558747932315) - timing_s/agent_loop/generate_sequences/max:np.float64(6.326467616483569) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.7320152404354303) - timing_s/agent_loop/t
|
|
step:9 - global_seqlen/min:15744 - global_seqlen/max:23758 - global_seqlen/minmax_diff:8014 - global_seqlen/balanced_min:18742 - global_seqlen/balanced_max:18745 - global_seqlen/mean:18743.875 - actor/entropy:0.6213304400444031 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2981865406036377 - training/rollout_probs_diff_mean:0.005725470371544361 - training/rollout_probs_diff_std:0.010449985042214394 - training/rollout_actor_probs_pearson_corr:0.9993769526481628 - self_distillation/success_group_fraction:0.59375 - self_distillation/success_sample_fraction:0.5859375 - self_distillation/correct_sample_fraction:0.359375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.5859375 - rollout_corr/rollout_is_mean:0.9998330473899841 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.596377968788147 - rollout_corr/rollout_is_min:0.6214840412139893 - rollout_corr/rollout_is_std:0.03818284347653389 - rollout_corr/rollout_is_eff_sample_size:0.9985438362659167 - rollout_corr/rollout_is_seq_mean:0.999792754650116 - rollout_corr/rollout_is_seq_std:0.0024519991129636765 - rollout_corr/rollout_is_seq_max:1.0065456628799438 - rollout_corr/rollout_is_seq_min:0.992550253868103 - rollout_corr/rollout_is_seq_max_deviation:0.007449746131896973 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.8954706261865795) - rollout_corr/training_log_ppl:np.float64(0.6263229054748081) - rollout_corr/kl:np.float64(0.0009552467599878156) - rollout_corr/k3_kl:np.float64(0.0007480508141952669) - rollout_corr/rollout_ppl:np.float64(1.893649737816304) - rollout_corr/rollout_log_ppl:np.float64(0.6253676604246721) - rollout_corr/log_ppl_diff:np.float64(0.0009552450501359999) - rollout_corr/log_ppl_abs_diff:np.float64(0.0020716768340207636) - rollout_corr/log_ppl_diff_max:np.float64(0.008369922637939453) - rollout_corr/log_ppl_diff_min:np.float64(-0.005397498607635498) - rollout_corr/ppl_ratio:np.float64(1.0009586894884706) - rollout_corr/chi2_token:np.float64(0.0010812999680638313) - rollout_corr/chi2_seq:np.float64(0.4782353830523789) - actor/pg_loss:np.float64(0.026237479949486442) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.4140625) - actor/grad_norm:np.float64(0.24165703356266022) - perf/mfu/actor:np.float64(0.04349852775955479) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(192.48685836791992) - actor/lr:np.float64(8.000000000000001e-06) - training/global_step:9 - training/epoch:0 - critic/score/mean:0.359375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.359375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.005137684755027294 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.005137684755027294 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:296.90234375 - response_length/max:652.0 - response_length/min:158.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:296.90234375 - response_length_non_aborted/max:652.0 - response_length_non_aborted/min:158.0 - 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:0.00020686164498329163 - timing_s/agent_loop/generate_sequences/min:np.float64(2.118359662592411) - timing_s/agent_loop/generate_sequences/max:np.float64(6.14809369854629) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.723439998626418) - timing_s/agent_loop/tool_
|
|
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 target product, "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1", contains a nitrogen-containing ring with a chlorinated aromatic ring and a trifluoromethyl group.
|
|
|
|
- **Option A** includes a nitrogen-containing ring, a chlorinated aromatic ring, and a trifluoromethyl group, which matches the target product. It also has the correct functional groups and structure.
|
|
- **Option B** has a different arrangement of atoms and includes a chlorine atom in a position that does not match the target structure.
|
|
- **Option C** introduces an oxygen-containing group that is not present in the target molecule.
|
|
- **Option D** contains a double bond and a different carbon chain structure, which does not match the required trifluoromethyl group.
|
|
|
|
Therefore, **Option A** is the only correct reactant that can be used to synthesize the target product.
|
|
|
|
</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-Qwen3-8B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_10
|
|
step:10 - global_seqlen/min:17265 - global_seqlen/max:21645 - global_seqlen/minmax_diff:4380 - global_seqlen/balanced_min:18758 - global_seqlen/balanced_max:18760 - global_seqlen/mean:18758.625 - actor/entropy:0.5005174875259399 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.16887551546096802 - training/rollout_probs_diff_mean:0.005386246833950281 - training/rollout_probs_diff_std:0.010893335565924644 - training/rollout_actor_probs_pearson_corr:0.9992403388023376 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.734375 - self_distillation/correct_sample_fraction:0.48828125 - 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:1.0001847743988037 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.577920913696289 - rollout_corr/rollout_is_min:0.6317436695098877 - rollout_corr/rollout_is_std:0.03774796053767204 - rollout_corr/rollout_is_eff_sample_size:0.9985775943190434 - rollout_corr/rollout_is_seq_mean:1.0001989603042603 - rollout_corr/rollout_is_seq_std:0.0022370326332747936 - rollout_corr/rollout_is_seq_max:1.0071781873703003 - rollout_corr/rollout_is_seq_min:0.9931437969207764 - rollout_corr/rollout_is_seq_max_deviation:0.007178187370300293 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6738104745745659) - rollout_corr/training_log_ppl:np.float64(0.5069695328711532) - rollout_corr/kl:np.float64(0.0005306363165065164) - rollout_corr/k3_kl:np.float64(0.0007295863661056501) - rollout_corr/rollout_ppl:np.float64(1.6729822000488639) - rollout_corr/rollout_log_ppl:np.float64(0.506438895361498) - rollout_corr/log_ppl_diff:np.float64(0.0005306375096552074) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017152809887193143) - rollout_corr/log_ppl_diff_max:np.float64(0.007777601480484009) - rollout_corr/log_ppl_diff_min:np.float64(-0.006521284580230713) - rollout_corr/ppl_ratio:np.float64(1.000533226178959) - rollout_corr/chi2_token:np.float64(0.0018700843211263418) - rollout_corr/chi2_seq:np.float64(0.5648596533574164) - actor/pg_loss:np.float64(0.02884473373705987) - 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(0.261650025844574) - perf/mfu/actor:np.float64(0.04384994246511119) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(192.59494352340698) - actor/lr:np.float64(9e-06) - val-aux/sciknoweval/reward/mean@16:np.float64(0.5714285714285714) - val-aux/sciknoweval/reward/std@16:np.float64(0.2001270489443599) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.6555904761904762) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.16489550721346855) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.4885809523809524) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.1742465026614643) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.5725380952380953) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.19933740317084478) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7203952380952381) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.11649641179634473) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.4191) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.13202274301223454) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.5837380952380952) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.16435423379933606) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7632714285714286) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.07306074251331264) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.36913809523809527) - val-aux/sciknoweval/reward/worst@8/std:np.float64
|
|
step:11 - global_seqlen/min:16220 - global_seqlen/max:20861 - global_seqlen/minmax_diff:4641 - global_seqlen/balanced_min:18903 - global_seqlen/balanced_max:18906 - global_seqlen/mean:18904.5 - actor/entropy:0.4962438941001892 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18277812004089355 - training/rollout_probs_diff_mean:0.0054591624066233635 - training/rollout_probs_diff_std:0.010961337015032768 - training/rollout_actor_probs_pearson_corr:0.9992218613624573 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.8984375 - 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:0.8984375 - rollout_corr/rollout_is_mean:1.0002753734588623 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.540004849433899 - rollout_corr/rollout_is_min:0.6445291042327881 - rollout_corr/rollout_is_std:0.03751830756664276 - rollout_corr/rollout_is_eff_sample_size:0.9985951874486111 - rollout_corr/rollout_is_seq_mean:1.0002588033676147 - rollout_corr/rollout_is_seq_std:0.002207246609032154 - rollout_corr/rollout_is_seq_max:1.0072370767593384 - rollout_corr/rollout_is_seq_min:0.9937965273857117 - rollout_corr/rollout_is_seq_max_deviation:0.007237076759338379 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6612941497005522) - rollout_corr/training_log_ppl:np.float64(0.4997066975920461) - rollout_corr/kl:np.float64(0.000457240256257499) - rollout_corr/k3_kl:np.float64(0.0007160234944763033) - rollout_corr/rollout_ppl:np.float64(1.6605073073878884) - rollout_corr/rollout_log_ppl:np.float64(0.49924945569364354) - rollout_corr/log_ppl_diff:np.float64(0.0004572418984025717) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017398506170138717) - rollout_corr/log_ppl_diff_max:np.float64(0.006929993629455566) - rollout_corr/log_ppl_diff_min:np.float64(-0.006457328796386719) - rollout_corr/ppl_ratio:np.float64(1.000459767645225) - rollout_corr/chi2_token:np.float64(0.0019666727166622877) - rollout_corr/chi2_seq:np.float64(0.5841377004981041) - actor/pg_loss:np.float64(0.03643380916037131) - 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(0.3261227607727051) - perf/mfu/actor:np.float64(0.04426301137559212) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.3918800354004) - actor/lr:np.float64(1e-05) - training/global_step:11 - 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:-0.007205228321254253 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.007205228321254253 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:291.671875 - response_length/max:676.0 - response_length/min:135.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:291.671875 - response_length_non_aborted/max:676.0 - response_length_non_aborted/min:135.0 - 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:7.468834519386292e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.7408243250101805) - timing_s/agent_loop/generate_sequences/max:np.float64(6.118830543011427) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.6799432409607107) - timing_s/agent_loop/tool_calls/min:
|
|
step:12 - global_seqlen/min:15459 - global_seqlen/max:19913 - global_seqlen/minmax_diff:4454 - global_seqlen/balanced_min:17907 - global_seqlen/balanced_max:17913 - global_seqlen/mean:17911.875 - actor/entropy:0.46301960945129395 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18448376655578613 - training/rollout_probs_diff_mean:0.0045851110480725765 - training/rollout_probs_diff_std:0.00968915969133377 - training/rollout_actor_probs_pearson_corr:0.9994009137153625 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.76953125 - self_distillation/correct_sample_fraction:0.58203125 - 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:0.999979555606842 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.3302294973982498e-05 - rollout_corr/rollout_is_max:1.629321575164795 - rollout_corr/rollout_is_min:0.473160058259964 - rollout_corr/rollout_is_std:0.03289290890097618 - rollout_corr/rollout_is_eff_sample_size:0.998919225824644 - rollout_corr/rollout_is_seq_mean:0.9999831914901733 - rollout_corr/rollout_is_seq_std:0.0019613418262451887 - rollout_corr/rollout_is_seq_max:1.0053656101226807 - rollout_corr/rollout_is_seq_min:0.9946230053901672 - rollout_corr/rollout_is_seq_max_deviation:0.005376994609832764 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6116638104431331) - rollout_corr/training_log_ppl:np.float64(0.46986032044515014) - rollout_corr/kl:np.float64(0.0005791713123853848) - rollout_corr/k3_kl:np.float64(0.000562254604460577) - rollout_corr/rollout_ppl:np.float64(1.610707962885499) - rollout_corr/rollout_log_ppl:np.float64(0.4692811457789503) - rollout_corr/log_ppl_diff:np.float64(0.0005791746661998332) - rollout_corr/log_ppl_abs_diff:np.float64(0.001605078054126352) - rollout_corr/log_ppl_diff_max:np.float64(0.006518274545669556) - rollout_corr/log_ppl_diff_min:np.float64(-0.004486262798309326) - rollout_corr/ppl_ratio:np.float64(1.0005813005845994) - rollout_corr/chi2_token:np.float64(0.001095455139875412) - rollout_corr/chi2_seq:np.float64(0.2610923498868942) - actor/pg_loss:np.float64(0.030304332489322405) - 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(0.22565484046936035) - perf/mfu/actor:np.float64(0.0416924520274006) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.64941883087158) - actor/lr:np.float64(1e-05) - training/global_step:12 - training/epoch:0 - critic/score/mean:0.58203125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.58203125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.004950116388499737 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.004950116388499737 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:293.65234375 - response_length/max:827.0 - response_length/min:110.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:293.65234375 - response_length_non_aborted/max:827.0 - response_length_non_aborted/min:110.0 - 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:7.905997335910797e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.4059838727116585) - timing_s/agent_loop/generate_sequences/max:np.float64(7.031799431890249) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.644470164035738) - timing_s/age
|
|
step:13 - global_seqlen/min:15134 - global_seqlen/max:18319 - global_seqlen/minmax_diff:3185 - global_seqlen/balanced_min:17141 - global_seqlen/balanced_max:17143 - global_seqlen/mean:17141.875 - actor/entropy:0.4880797266960144 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12243753671646118 - training/rollout_probs_diff_mean:0.004677236080169678 - training/rollout_probs_diff_std:0.009508935734629631 - training/rollout_actor_probs_pearson_corr:0.9994279742240906 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - self_distillation/correct_sample_fraction:0.578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83203125 - rollout_corr/rollout_is_mean:1.0001262426376343 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4946610927581787 - rollout_corr/rollout_is_min:0.644141435623169 - rollout_corr/rollout_is_std:0.03319055587053299 - rollout_corr/rollout_is_eff_sample_size:0.9989000749597069 - rollout_corr/rollout_is_seq_mean:1.0002038478851318 - rollout_corr/rollout_is_seq_std:0.002288853982463479 - rollout_corr/rollout_is_seq_max:1.0071090459823608 - rollout_corr/rollout_is_seq_min:0.9938533306121826 - rollout_corr/rollout_is_seq_max_deviation:0.00710904598236084 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6584706637077034) - rollout_corr/training_log_ppl:np.float64(0.4972652036231011) - rollout_corr/kl:np.float64(0.0003569795799762687) - rollout_corr/k3_kl:np.float64(0.0005608202401390372) - rollout_corr/rollout_ppl:np.float64(1.6578380083665252) - rollout_corr/rollout_log_ppl:np.float64(0.4969082251191139) - rollout_corr/log_ppl_diff:np.float64(0.0003569785039871931) - rollout_corr/log_ppl_abs_diff:np.float64(0.0018185239750891924) - rollout_corr/log_ppl_diff_max:np.float64(0.0069852471351623535) - rollout_corr/log_ppl_diff_min:np.float64(-0.006149470806121826) - rollout_corr/ppl_ratio:np.float64(1.000359632074833) - rollout_corr/chi2_token:np.float64(0.0015383281279355288) - rollout_corr/chi2_seq:np.float64(0.5321485358290374) - actor/pg_loss:np.float64(0.029986228575580753) - 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(0.24110852181911469) - perf/mfu/actor:np.float64(0.04007115304788385) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.10686874389648) - actor/lr:np.float64(1e-05) - training/global_step:13 - training/epoch:0 - critic/score/mean:0.578125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.578125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.006019000895321369 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.006019000895321369 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:258.62109375 - response_length/max:563.0 - response_length/min:132.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:258.62109375 - response_length_non_aborted/max:563.0 - response_length_non_aborted/min:132.0 - 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:0.00010392069816589355 - timing_s/agent_loop/generate_sequences/min:np.float64(1.659997260197997) - timing_s/agent_loop/generate_sequences/max:np.float64(5.363083679229021) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.2231069085246418) - timing_s/agent_loop/tool_calls/
|
|
step:14 - global_seqlen/min:14966 - global_seqlen/max:20705 - global_seqlen/minmax_diff:5739 - global_seqlen/balanced_min:17221 - global_seqlen/balanced_max:17223 - global_seqlen/mean:17221.875 - actor/entropy:0.42735856771469116 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.16049106419086456 - training/rollout_probs_diff_mean:0.004373057745397091 - training/rollout_probs_diff_std:0.009555500000715256 - training/rollout_actor_probs_pearson_corr:0.999373197555542 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.828125 - 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:0.828125 - rollout_corr/rollout_is_mean:0.9998784065246582 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4041495323181152 - rollout_corr/rollout_is_min:0.5407115817070007 - rollout_corr/rollout_is_std:0.03213580697774887 - rollout_corr/rollout_is_eff_sample_size:0.9989682363510435 - rollout_corr/rollout_is_seq_mean:0.9999240040779114 - rollout_corr/rollout_is_seq_std:0.002165188081562519 - rollout_corr/rollout_is_seq_max:1.0071263313293457 - rollout_corr/rollout_is_seq_min:0.9942255616188049 - rollout_corr/rollout_is_seq_max_deviation:0.007126331329345703 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5679032318294048) - rollout_corr/training_log_ppl:np.float64(0.44162937096552923) - rollout_corr/kl:np.float64(0.0006079288402069949) - rollout_corr/k3_kl:np.float64(0.0005319095446338906) - rollout_corr/rollout_ppl:np.float64(1.566970007494092) - rollout_corr/rollout_log_ppl:np.float64(0.44102143921190873) - rollout_corr/log_ppl_diff:np.float64(0.0006079317536205053) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017368750413879752) - rollout_corr/log_ppl_diff_max:np.float64(0.006370842456817627) - rollout_corr/log_ppl_diff_min:np.float64(-0.006435096263885498) - rollout_corr/ppl_ratio:np.float64(1.0006104472558945) - rollout_corr/chi2_token:np.float64(0.0009174160659313202) - rollout_corr/chi2_seq:np.float64(0.3130092390347272) - actor/pg_loss:np.float64(0.027514628131029895) - 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(0.20976684987545013) - perf/mfu/actor:np.float64(0.04016782132727843) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.04870986938477) - actor/lr:np.float64(1e-05) - training/global_step:14 - 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:0.0002849475422408432 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0002849475422408432 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:270.74609375 - response_length/max:637.0 - response_length/min:111.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:270.74609375 - response_length_non_aborted/max:637.0 - response_length_non_aborted/min:111.0 - 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:0.0001306142657995224 - timing_s/agent_loop/generate_sequences/min:np.float64(1.479357348755002) - timing_s/agent_loop/generate_sequences/max:np.float64(5.726118391379714) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.341816912405193) - timing_s/agent_loop/tool_calls
|
|
step:15 - global_seqlen/min:16330 - global_seqlen/max:21530 - global_seqlen/minmax_diff:5200 - global_seqlen/balanced_min:18330 - global_seqlen/balanced_max:18332 - global_seqlen/mean:18331.375 - actor/entropy:0.4372391700744629 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.17282047867774963 - training/rollout_probs_diff_mean:0.004539350513368845 - training/rollout_probs_diff_std:0.009851260110735893 - training/rollout_actor_probs_pearson_corr:0.9993435144424438 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.76953125 - self_distillation/correct_sample_fraction:0.56640625 - 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:0.9998887777328491 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.8670971393585205 - rollout_corr/rollout_is_min:0.5850483179092407 - rollout_corr/rollout_is_std:0.03362765535712242 - rollout_corr/rollout_is_eff_sample_size:0.9988702201070337 - rollout_corr/rollout_is_seq_mean:0.9999094605445862 - rollout_corr/rollout_is_seq_std:0.0023757158778607845 - rollout_corr/rollout_is_seq_max:1.0065155029296875 - rollout_corr/rollout_is_seq_min:0.9888423681259155 - rollout_corr/rollout_is_seq_max_deviation:0.011157631874084473 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5889150914736092) - rollout_corr/training_log_ppl:np.float64(0.4546032711514272) - rollout_corr/kl:np.float64(0.0006808645128231205) - rollout_corr/k3_kl:np.float64(0.000590274933585988) - rollout_corr/rollout_ppl:np.float64(1.587844471912831) - rollout_corr/rollout_log_ppl:np.float64(0.4539224046166055) - rollout_corr/log_ppl_diff:np.float64(0.0006808665348216891) - rollout_corr/log_ppl_abs_diff:np.float64(0.001896374044008553) - rollout_corr/log_ppl_diff_max:np.float64(0.01238185167312622) - rollout_corr/log_ppl_diff_min:np.float64(-0.0057533979415893555) - rollout_corr/ppl_ratio:np.float64(1.0006839365232736) - rollout_corr/chi2_token:np.float64(0.0010036253370344639) - rollout_corr/chi2_seq:np.float64(0.3624662000220269) - actor/pg_loss:np.float64(0.026144841123823426) - 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(0.19042067229747772) - perf/mfu/actor:np.float64(0.0429849936546457) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.02049779891968) - actor/lr:np.float64(1e-05) - training/global_step:15 - training/epoch:0 - critic/score/mean:0.56640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.56640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.006459776312112808 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.006459776312112808 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:285.94921875 - response_length/max:689.0 - response_length/min:137.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:285.94921875 - response_length_non_aborted/max:689.0 - response_length_non_aborted/min:137.0 - 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:0.00017814897000789642 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8202488366514444) - timing_s/agent_loop/generate_sequences/max:np.float64(6.110847994685173) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.5281097801271244) - timing_s/agent_loop/tool_c
|
|
step:16 - global_seqlen/min:16303 - global_seqlen/max:23663 - global_seqlen/minmax_diff:7360 - global_seqlen/balanced_min:19396 - global_seqlen/balanced_max:19399 - global_seqlen/mean:19397.625 - actor/entropy:0.38260984420776367 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1588025689125061 - training/rollout_probs_diff_mean:0.004053414333611727 - training/rollout_probs_diff_std:0.009250919334590435 - training/rollout_actor_probs_pearson_corr:0.9993999004364014 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80078125 - self_distillation/correct_sample_fraction:0.55859375 - 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:1.000063419342041 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.666884422302246 - rollout_corr/rollout_is_min:0.5937062501907349 - rollout_corr/rollout_is_std:0.03158390894532204 - rollout_corr/rollout_is_eff_sample_size:0.9990035697379116 - rollout_corr/rollout_is_seq_mean:1.000007152557373 - rollout_corr/rollout_is_seq_std:0.002001494634896517 - rollout_corr/rollout_is_seq_max:1.0068124532699585 - rollout_corr/rollout_is_seq_min:0.9939054250717163 - rollout_corr/rollout_is_seq_max_deviation:0.006812453269958496 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4917708160355687) - rollout_corr/training_log_ppl:np.float64(0.3936915730009787) - rollout_corr/kl:np.float64(0.0005167508950127964) - rollout_corr/k3_kl:np.float64(0.0005238576729880151) - rollout_corr/rollout_ppl:np.float64(1.4909629775211215) - rollout_corr/rollout_log_ppl:np.float64(0.39317482186015695) - rollout_corr/log_ppl_diff:np.float64(0.0005167511408217251) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015984405181370676) - rollout_corr/log_ppl_diff_max:np.float64(0.006906956434249878) - rollout_corr/log_ppl_diff_min:np.float64(-0.00619131326675415) - rollout_corr/ppl_ratio:np.float64(1.0005188947543502) - rollout_corr/chi2_token:np.float64(0.0010713336523622274) - rollout_corr/chi2_seq:np.float64(0.598915500100702) - actor/pg_loss:np.float64(0.02560856024865643) - 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(0.17774124443531036) - perf/mfu/actor:np.float64(0.045247853376615466) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.07054138183594) - actor/lr:np.float64(1e-05) - training/global_step:16 - training/epoch:0 - critic/score/mean:0.55859375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.55859375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.004094302654266357 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.004094302654266357 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:325.45703125 - response_length/max:839.0 - response_length/min:158.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:325.45703125 - response_length_non_aborted/max:839.0 - response_length_non_aborted/min:158.0 - 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:0.0001201629638671875 - timing_s/agent_loop/generate_sequences/min:np.float64(2.2049541864544153) - timing_s/agent_loop/generate_sequences/max:np.float64(7.3498986549675465) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.0494183809059905) - timing_s/agent_loop/tool_c
|
|
step:17 - global_seqlen/min:19906 - global_seqlen/max:25091 - global_seqlen/minmax_diff:5185 - global_seqlen/balanced_min:22898 - global_seqlen/balanced_max:22904 - global_seqlen/mean:22902.125 - actor/entropy:0.4285140931606293 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.14667820930480957 - training/rollout_probs_diff_mean:0.0039658634923398495 - training/rollout_probs_diff_std:0.008768660016357899 - training/rollout_actor_probs_pearson_corr:0.9995031952857971 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.70703125 - self_distillation/correct_sample_fraction:0.48828125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.70703125 - rollout_corr/rollout_is_mean:1.000002384185791 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.9453516006469727 - rollout_corr/rollout_is_min:0.6484981775283813 - rollout_corr/rollout_is_std:0.0307520292699337 - rollout_corr/rollout_is_eff_sample_size:0.999055087201374 - rollout_corr/rollout_is_seq_mean:0.9999778270721436 - rollout_corr/rollout_is_seq_std:0.001909718499518931 - rollout_corr/rollout_is_seq_max:1.0059030055999756 - rollout_corr/rollout_is_seq_min:0.993180513381958 - rollout_corr/rollout_is_seq_max_deviation:0.006819486618041992 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6043823445215821) - rollout_corr/training_log_ppl:np.float64(0.46223142312373966) - rollout_corr/kl:np.float64(0.0005413741727636201) - rollout_corr/k3_kl:np.float64(0.0005192171933572354) - rollout_corr/rollout_ppl:np.float64(1.6034233663231134) - rollout_corr/rollout_log_ppl:np.float64(0.4616900450200774) - rollout_corr/log_ppl_diff:np.float64(0.0005413781036622822) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015439948183484375) - rollout_corr/log_ppl_diff_max:np.float64(0.00736236572265625) - rollout_corr/log_ppl_diff_min:np.float64(-0.00518268346786499) - rollout_corr/ppl_ratio:np.float64(1.0005433990154415) - rollout_corr/chi2_token:np.float64(0.001000862568616867) - rollout_corr/chi2_seq:np.float64(0.42431468493305147) - actor/pg_loss:np.float64(0.021404292059742147) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.29296875) - actor/grad_norm:np.float64(0.15989063680171967) - perf/mfu/actor:np.float64(0.052148047530688954) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.12295150756836) - actor/lr:np.float64(1e-05) - training/global_step:17 - training/epoch:0 - critic/score/mean:0.48828125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.48828125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.005072104744613171 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.005072104744613171 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:382.47265625 - response_length/max:1067.0 - response_length/min:171.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:382.47265625 - response_length_non_aborted/max:1067.0 - response_length_non_aborted/min:171.0 - 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:0.00015934742987155914 - timing_s/agent_loop/generate_sequences/min:np.float64(2.380508380010724) - timing_s/agent_loop/generate_sequences/max:np.float64(9.026037689298391) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.485403979517287) - timing_s/agent_loop/tool_c
|
|
step:18 - global_seqlen/min:16995 - global_seqlen/max:21488 - global_seqlen/minmax_diff:4493 - global_seqlen/balanced_min:19202 - global_seqlen/balanced_max:19212 - global_seqlen/mean:19209.875 - actor/entropy:0.43559494614601135 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1548389196395874 - training/rollout_probs_diff_mean:0.003977622836828232 - training/rollout_probs_diff_std:0.008606972172856331 - training/rollout_actor_probs_pearson_corr:0.9995144605636597 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83984375 - 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:0.83984375 - rollout_corr/rollout_is_mean:0.9998852014541626 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5729793310165405 - rollout_corr/rollout_is_min:0.6484133005142212 - rollout_corr/rollout_is_std:0.03009967692196369 - rollout_corr/rollout_is_eff_sample_size:0.9990947104818377 - rollout_corr/rollout_is_seq_mean:0.9998518228530884 - rollout_corr/rollout_is_seq_std:0.001971276244148612 - rollout_corr/rollout_is_seq_max:1.0063395500183105 - rollout_corr/rollout_is_seq_min:0.9934338927268982 - rollout_corr/rollout_is_seq_max_deviation:0.006566107273101807 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6012631598860025) - rollout_corr/training_log_ppl:np.float64(0.4596485758665949) - rollout_corr/kl:np.float64(0.0006253383266443535) - rollout_corr/k3_kl:np.float64(0.0004771838254669092) - rollout_corr/rollout_ppl:np.float64(1.6002365187741816) - rollout_corr/rollout_log_ppl:np.float64(0.45902323676273227) - rollout_corr/log_ppl_diff:np.float64(0.0006253391038626432) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015947106294333935) - rollout_corr/log_ppl_diff_max:np.float64(0.007226765155792236) - rollout_corr/log_ppl_diff_min:np.float64(-0.005567431449890137) - rollout_corr/ppl_ratio:np.float64(1.0006274920888245) - rollout_corr/chi2_token:np.float64(0.0006632185541093349) - rollout_corr/chi2_seq:np.float64(0.283309651305899) - actor/pg_loss:np.float64(0.026730960482382216) - 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(0.19164638221263885) - perf/mfu/actor:np.float64(0.04450560627476201) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.05157089233398) - actor/lr:np.float64(1e-05) - training/global_step:18 - 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:-0.002379985526204109 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.002379985526204109 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:327.02734375 - response_length/max:1010.0 - response_length/min:139.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:327.02734375 - response_length_non_aborted/max:1010.0 - response_length_non_aborted/min:139.0 - 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:0.00017552450299263 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9803128950297832) - timing_s/agent_loop/generate_sequences/max:np.float64(8.442343609407544) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.9924488358883536) - timing_s/agent_loop/tool_calls/min:np
|
|
step:19 - global_seqlen/min:14562 - global_seqlen/max:27432 - global_seqlen/minmax_diff:12870 - global_seqlen/balanced_min:19678 - global_seqlen/balanced_max:19685 - global_seqlen/mean:19682.75 - actor/entropy:0.453940212726593 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4744124710559845 - training/rollout_probs_diff_mean:0.003921263385564089 - training/rollout_probs_diff_std:0.008538271300494671 - training/rollout_actor_probs_pearson_corr:0.9995570778846741 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - 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:0.8359375 - rollout_corr/rollout_is_mean:0.9999681115150452 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.2481589692470152e-05 - rollout_corr/rollout_is_max:1.5281609296798706 - rollout_corr/rollout_is_min:0.443146675825119 - rollout_corr/rollout_is_std:0.030551739037036896 - rollout_corr/rollout_is_eff_sample_size:0.9990673427151677 - rollout_corr/rollout_is_seq_mean:0.9999241828918457 - rollout_corr/rollout_is_seq_std:0.002501407405361533 - rollout_corr/rollout_is_seq_max:1.0178346633911133 - rollout_corr/rollout_is_seq_min:0.9788051247596741 - rollout_corr/rollout_is_seq_max_deviation:0.021194875240325928 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6223687655292451) - rollout_corr/training_log_ppl:np.float64(0.47468685437343083) - rollout_corr/kl:np.float64(0.0005873041097430587) - rollout_corr/k3_kl:np.float64(0.00051153602123577) - rollout_corr/rollout_ppl:np.float64(1.6214658892713487) - rollout_corr/rollout_log_ppl:np.float64(0.4740995457977988) - rollout_corr/log_ppl_diff:np.float64(0.0005873085756320506) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015665637620259076) - rollout_corr/log_ppl_diff_max:np.float64(0.027092337608337402) - rollout_corr/log_ppl_diff_min:np.float64(-0.015913575887680054) - rollout_corr/ppl_ratio:np.float64(1.0005910161416978) - rollout_corr/chi2_token:np.float64(0.0008642538450658321) - rollout_corr/chi2_seq:np.float64(0.19235544628463686) - actor/pg_loss:np.float64(0.027401684626966016) - 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(0.17280758917331696) - perf/mfu/actor:np.float64(0.045688296883408086) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.09777069091797) - actor/lr:np.float64(1e-05) - training/global_step:19 - training/epoch:0 - critic/score/mean:0.65625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.65625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00460882717743516 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00460882717743516 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:312.9609375 - response_length/max:1054.0 - response_length/min:18.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:312.9609375 - response_length_non_aborted/max:1054.0 - response_length_non_aborted/min:18.0 - 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:0.0001504644751548767 - timing_s/agent_loop/generate_sequences/min:np.float64(0.3043934088200331) - timing_s/agent_loop/generate_sequences/max:np.float64(8.564327601343393) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.7510185377977905) - timing_s/agent_loop/tool
|
|
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 target product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" is a nitrogen-containing compound with a fluorinated aromatic ring and a chlorinated aromatic ring. This suggests that the correct reactant must have the same core structure and functional groups as the target product.
|
|
|
|
- **Option A** contains a nitrogen in the correct position, the chlorinated aromatic ring, and the fluorinated aromatic ring, which matches the target product. It also has the correct substituents and functional groups.
|
|
- **Option B** introduces a different nitrogen position and a different ring structure, which does not align with the target product.
|
|
- **Option C** includes an oxygen atom and a hydroxyl group, which are not present in the target product.
|
|
- **Option D** includes a different carbon chain structure and lacks the correct nitrogen and fluorine placement.
|
|
|
|
Thus, **Option A** is the only correct reactant that matches the target product.
|
|
</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-Qwen3-8B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_20
|
|
step:20 - global_seqlen/min:16490 - global_seqlen/max:20298 - global_seqlen/minmax_diff:3808 - global_seqlen/balanced_min:18566 - global_seqlen/balanced_max:18569 - global_seqlen/mean:18567.875 - actor/entropy:0.5061439275741577 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2310560941696167 - training/rollout_probs_diff_mean:0.004149604123085737 - training/rollout_probs_diff_std:0.00835744570940733 - training/rollout_actor_probs_pearson_corr:0.9995498061180115 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.78125 - 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:0.78125 - rollout_corr/rollout_is_mean:0.9999780654907227 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.3944801092147827 - rollout_corr/rollout_is_min:0.6839407086372375 - rollout_corr/rollout_is_std:0.030308855697512627 - rollout_corr/rollout_is_eff_sample_size:0.9990822163153974 - rollout_corr/rollout_is_seq_mean:0.9999166131019592 - rollout_corr/rollout_is_seq_std:0.0023145442828536034 - rollout_corr/rollout_is_seq_max:1.004969596862793 - rollout_corr/rollout_is_seq_min:0.9827538728713989 - rollout_corr/rollout_is_seq_max_deviation:0.017246127128601074 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.68152741715312) - rollout_corr/training_log_ppl:np.float64(0.5100623590988107) - rollout_corr/kl:np.float64(0.0005645529809790162) - rollout_corr/k3_kl:np.float64(0.00048118357227622255) - rollout_corr/rollout_ppl:np.float64(1.680517458356917) - rollout_corr/rollout_log_ppl:np.float64(0.5094978031120263) - rollout_corr/log_ppl_diff:np.float64(0.0005645559867843986) - rollout_corr/log_ppl_abs_diff:np.float64(0.0017651560483500361) - rollout_corr/log_ppl_diff_max:np.float64(0.02067825198173523) - rollout_corr/log_ppl_diff_min:np.float64(-0.00447043776512146) - rollout_corr/ppl_ratio:np.float64(1.0005676480941474) - rollout_corr/chi2_token:np.float64(0.0007928451523184776) - rollout_corr/chi2_seq:np.float64(0.35891830269247293) - actor/pg_loss:np.float64(0.02594115215538295) - 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(0.15884989500045776) - perf/mfu/actor:np.float64(0.043443623512939705) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.10118865966797) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.6791666666666667) - val-aux/sciknoweval/reward/std@16:np.float64(0.109391289230873) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7227142857142858) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.09435221012626904) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6359285714285714) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.08906030749741908) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.6794333333333332) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.10919585971857018) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.7588285714285714) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.07623179703461139) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6017428571428571) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.06597450834426273) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.6824238095238097) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.08494460370860409) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.7894714285714285) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.05897949963421853) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.5752952380952382) - val-aux/sciknoweval/reward/worst@8/st
|
|
step:21 - global_seqlen/min:16461 - global_seqlen/max:21880 - global_seqlen/minmax_diff:5419 - global_seqlen/balanced_min:19476 - global_seqlen/balanced_max:19484 - global_seqlen/mean:19481.5 - actor/entropy:0.4757225811481476 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.09812116622924805 - training/rollout_probs_diff_mean:0.0038961053360253572 - training/rollout_probs_diff_std:0.008007421158254147 - training/rollout_actor_probs_pearson_corr:0.9995965957641602 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.796875 - 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:0.796875 - rollout_corr/rollout_is_mean:1.0000258684158325 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4084985256195068 - rollout_corr/rollout_is_min:0.6048171520233154 - rollout_corr/rollout_is_std:0.029357846826314926 - rollout_corr/rollout_is_eff_sample_size:0.9991388590394431 - rollout_corr/rollout_is_seq_mean:0.9998555183410645 - rollout_corr/rollout_is_seq_std:0.0022220418322831392 - rollout_corr/rollout_is_seq_max:1.0049490928649902 - rollout_corr/rollout_is_seq_min:0.9827251434326172 - rollout_corr/rollout_is_seq_max_deviation:0.017274856567382812 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6760855801403522) - rollout_corr/training_log_ppl:np.float64(0.50398448825581) - rollout_corr/kl:np.float64(0.0006250444803690414) - rollout_corr/k3_kl:np.float64(0.000480538414961984) - rollout_corr/rollout_ppl:np.float64(1.6750490693375468) - rollout_corr/rollout_log_ppl:np.float64(0.5033594403066672) - rollout_corr/log_ppl_diff:np.float64(0.0006250479491427541) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016198589000850916) - rollout_corr/log_ppl_diff_max:np.float64(0.020675331354141235) - rollout_corr/log_ppl_diff_min:np.float64(-0.004375159740447998) - rollout_corr/ppl_ratio:np.float64(1.0006280150264502) - rollout_corr/chi2_token:np.float64(0.0006643887609243393) - rollout_corr/chi2_seq:np.float64(0.3268553037196398) - actor/pg_loss:np.float64(0.026299531793938513) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.203125) - actor/grad_norm:np.float64(0.14760291576385498) - perf/mfu/actor:np.float64(0.04525548298328799) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(199.43410110473633) - actor/lr:np.float64(1e-05) - training/global_step:21 - 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:-0.004210854414850473 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.004210854414850473 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:325.609375 - response_length/max:860.0 - response_length/min:18.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:325.609375 - response_length_non_aborted/max:860.0 - response_length_non_aborted/min:18.0 - 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:6.809644401073456e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(0.27662935480475426) - timing_s/agent_loop/generate_sequences/max:np.float64(7.59946465305984) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.9350053522211965) - timing_s/agent_loop/tool_calls/min:np.float6
|
|
step:22 - global_seqlen/min:16437 - global_seqlen/max:18949 - global_seqlen/minmax_diff:2512 - global_seqlen/balanced_min:17334 - global_seqlen/balanced_max:17336 - global_seqlen/mean:17335.75 - actor/entropy:0.4336911737918854 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12206479907035828 - training/rollout_probs_diff_mean:0.0037276495713740587 - training/rollout_probs_diff_std:0.00804361142218113 - training/rollout_actor_probs_pearson_corr:0.9995757341384888 - self_distillation/success_group_fraction:0.65625 - self_distillation/success_sample_fraction:0.6484375 - 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:0.6484375 - rollout_corr/rollout_is_mean:1.0001039505004883 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6462105512619019 - rollout_corr/rollout_is_min:0.6167449355125427 - rollout_corr/rollout_is_std:0.028792036697268486 - rollout_corr/rollout_is_eff_sample_size:0.9991720622772111 - rollout_corr/rollout_is_seq_mean:1.0001753568649292 - rollout_corr/rollout_is_seq_std:0.0018125579226762056 - rollout_corr/rollout_is_seq_max:1.006730318069458 - rollout_corr/rollout_is_seq_min:0.9953950643539429 - rollout_corr/rollout_is_seq_max_deviation:0.006730318069458008 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6051223454996943) - rollout_corr/training_log_ppl:np.float64(0.4615049040876329) - rollout_corr/kl:np.float64(0.0002709297309984038) - rollout_corr/k3_kl:np.float64(0.00044632691296442317) - rollout_corr/rollout_ppl:np.float64(1.6047133235260844) - rollout_corr/rollout_log_ppl:np.float64(0.4612339739687741) - rollout_corr/log_ppl_diff:np.float64(0.00027093011885881424) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014038566732779145) - rollout_corr/log_ppl_diff_max:np.float64(0.005173385143280029) - rollout_corr/log_ppl_diff_min:np.float64(-0.005938291549682617) - rollout_corr/ppl_ratio:np.float64(1.0002725711092353) - rollout_corr/chi2_token:np.float64(0.001253241440281272) - rollout_corr/chi2_seq:np.float64(0.2761817956343293) - actor/pg_loss:np.float64(0.020855237748037325) - 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(0.12637825310230255) - perf/mfu/actor:np.float64(0.04079365860545161) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.3252067565918) - actor/lr:np.float64(1e-05) - training/global_step:22 - training/epoch:0 - critic/score/mean:0.5 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.5 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0029670908115804195 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0029670908115804195 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:286.1796875 - response_length/max:636.0 - response_length/min:92.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:286.1796875 - response_length_non_aborted/max:636.0 - response_length_non_aborted/min:92.0 - 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:0.00013215094804763794 - timing_s/agent_loop/generate_sequences/min:np.float64(1.2895570285618305) - timing_s/agent_loop/generate_sequences/max:np.float64(5.970777491107583) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.422414764885616) - timing_s/agent_loop/tool_calls/min:np.float64(0.0
|
|
step:23 - global_seqlen/min:13643 - global_seqlen/max:21133 - global_seqlen/minmax_diff:7490 - global_seqlen/balanced_min:18004 - global_seqlen/balanced_max:18005 - global_seqlen/mean:18004.375 - actor/entropy:0.4370790719985962 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1224580705165863 - training/rollout_probs_diff_mean:0.003933351021260023 - training/rollout_probs_diff_std:0.008326631039381027 - training/rollout_actor_probs_pearson_corr:0.9995477199554443 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.875 - self_distillation/correct_sample_fraction:0.75 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.875 - rollout_corr/rollout_is_mean:1.0000990629196167 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.3940829038619995 - rollout_corr/rollout_is_min:0.6908147931098938 - rollout_corr/rollout_is_std:0.029280593618750572 - rollout_corr/rollout_is_eff_sample_size:0.9991436192247598 - rollout_corr/rollout_is_seq_mean:1.0000967979431152 - rollout_corr/rollout_is_seq_std:0.0017464814009144902 - rollout_corr/rollout_is_seq_max:1.0064647197723389 - rollout_corr/rollout_is_seq_min:0.9940153956413269 - rollout_corr/rollout_is_seq_max_deviation:0.006464719772338867 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5920787728391588) - rollout_corr/training_log_ppl:np.float64(0.45516224118182436) - rollout_corr/kl:np.float64(0.0003444707492832322) - rollout_corr/k3_kl:np.float64(0.000441229893738182) - rollout_corr/rollout_ppl:np.float64(1.5915250945836306) - rollout_corr/rollout_log_ppl:np.float64(0.45481776879751123) - rollout_corr/log_ppl_diff:np.float64(0.00034447238431312144) - rollout_corr/log_ppl_abs_diff:np.float64(0.001376775820972398) - rollout_corr/log_ppl_diff_max:np.float64(0.0067074596881866455) - rollout_corr/log_ppl_diff_min:np.float64(-0.006047844886779785) - rollout_corr/ppl_ratio:np.float64(1.000346059910953) - rollout_corr/chi2_token:np.float64(0.0010785374324768782) - rollout_corr/chi2_seq:np.float64(0.2507378631271422) - actor/pg_loss:np.float64(0.029386846308625536) - 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(0.1569817066192627) - perf/mfu/actor:np.float64(0.04177945598935098) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.8329954147339) - actor/lr:np.float64(1e-05) - training/global_step:23 - training/epoch:0 - critic/score/mean:0.75 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.75 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0033675399608910084 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0033675399608910084 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:284.91796875 - response_length/max:649.0 - response_length/min:122.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:284.91796875 - response_length_non_aborted/max:649.0 - response_length_non_aborted/min:122.0 - 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:0.00012271665036678314 - timing_s/agent_loop/generate_sequences/min:np.float64(1.1439542472362518) - timing_s/agent_loop/generate_sequences/max:np.float64(6.067026806995273) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.2583775121092913) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - tim
|
|
step:24 - global_seqlen/min:15198 - global_seqlen/max:19559 - global_seqlen/minmax_diff:4361 - global_seqlen/balanced_min:17171 - global_seqlen/balanced_max:17176 - global_seqlen/mean:17174.125 - actor/entropy:0.44554394483566284 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1728690266609192 - training/rollout_probs_diff_mean:0.004030889365822077 - training/rollout_probs_diff_std:0.008343422785401344 - training/rollout_actor_probs_pearson_corr:0.999521017074585 - 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:0.9999503493309021 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.5239953427226283e-05 - rollout_corr/rollout_is_max:1.6097290515899658 - rollout_corr/rollout_is_min:0.3767780363559723 - rollout_corr/rollout_is_std:0.029928892850875854 - rollout_corr/rollout_is_eff_sample_size:0.9991049440319062 - rollout_corr/rollout_is_seq_mean:1.000017523765564 - rollout_corr/rollout_is_seq_std:0.0020160211715847254 - rollout_corr/rollout_is_seq_max:1.0062888860702515 - rollout_corr/rollout_is_seq_min:0.9952409267425537 - rollout_corr/rollout_is_seq_max_deviation:0.006288886070251465 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6035849140025675) - rollout_corr/training_log_ppl:np.float64(0.4631582065485418) - rollout_corr/kl:np.float64(0.000441291167081026) - rollout_corr/k3_kl:np.float64(0.0004588485890053562) - rollout_corr/rollout_ppl:np.float64(1.6028567119501531) - rollout_corr/rollout_log_ppl:np.float64(0.4627169130835682) - rollout_corr/log_ppl_diff:np.float64(0.0004412934649735689) - rollout_corr/log_ppl_abs_diff:np.float64(0.0016674348153173923) - rollout_corr/log_ppl_diff_max:np.float64(0.005479395389556885) - rollout_corr/log_ppl_diff_min:np.float64(-0.005660295486450195) - rollout_corr/ppl_ratio:np.float64(1.0004434485454112) - rollout_corr/chi2_token:np.float64(0.0009550715330988169) - rollout_corr/chi2_seq:np.float64(0.34876859025098383) - actor/pg_loss:np.float64(0.031064969865838066) - 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(0.17292560636997223) - perf/mfu/actor:np.float64(0.04008540172431062) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.79053497314453) - actor/lr:np.float64(1e-05) - training/global_step:24 - training/epoch:0 - critic/score/mean:0.7109375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7109375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0008648673538118601 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0008648673538118601 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:256.31640625 - response_length/max:530.0 - response_length/min:100.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:256.31640625 - response_length_non_aborted/max:530.0 - response_length_non_aborted/min:100.0 - 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:0.0001109093427658081 - timing_s/agent_loop/generate_sequences/min:np.float64(1.3898602649569511) - timing_s/agent_loop/generate_sequences/max:np.float64(5.099580256268382) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.1756780097348383) - timing_s/agen
|
|
step:25 - global_seqlen/min:15089 - global_seqlen/max:22830 - global_seqlen/minmax_diff:7741 - global_seqlen/balanced_min:18545 - global_seqlen/balanced_max:18553 - global_seqlen/mean:18550.0 - actor/entropy:0.4357784390449524 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.15484285354614258 - training/rollout_probs_diff_mean:0.003896940965205431 - training/rollout_probs_diff_std:0.008303479291498661 - training/rollout_actor_probs_pearson_corr:0.9995424151420593 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.8984375 - 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:0.8984375 - rollout_corr/rollout_is_mean:1.0002225637435913 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4750257730484009 - rollout_corr/rollout_is_min:0.65384840965271 - rollout_corr/rollout_is_std:0.029158201068639755 - rollout_corr/rollout_is_eff_sample_size:0.9991509976016397 - rollout_corr/rollout_is_seq_mean:1.0002158880233765 - rollout_corr/rollout_is_seq_std:0.0019787363708019257 - rollout_corr/rollout_is_seq_max:1.0072627067565918 - rollout_corr/rollout_is_seq_min:0.9931561946868896 - rollout_corr/rollout_is_seq_max_deviation:0.007262706756591797 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6172075737267733) - rollout_corr/training_log_ppl:np.float64(0.4692348457756452) - rollout_corr/kl:np.float64(0.00024129248307680484) - rollout_corr/k3_kl:np.float64(0.0004571954652305976) - rollout_corr/rollout_ppl:np.float64(1.616798825096339) - rollout_corr/rollout_log_ppl:np.float64(0.4689935490314383) - rollout_corr/log_ppl_diff:np.float64(0.00024129674420692027) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015522858884651214) - rollout_corr/log_ppl_diff_max:np.float64(0.0074301958084106445) - rollout_corr/log_ppl_diff_min:np.float64(-0.006698548793792725) - rollout_corr/ppl_ratio:np.float64(1.0002433084882796) - rollout_corr/chi2_token:np.float64(0.0013488694094121456) - rollout_corr/chi2_seq:np.float64(0.3894848038908094) - actor/pg_loss:np.float64(0.029641144054039614) - 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(0.1556435376405716) - perf/mfu/actor:np.float64(0.04320535946065441) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.80144739151) - actor/lr:np.float64(1e-05) - training/global_step:25 - 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:0.004848651587963104 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.004848651587963104 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:276.9375 - response_length/max:939.0 - response_length/min:110.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:276.9375 - response_length_non_aborted/max:939.0 - response_length_non_aborted/min:110.0 - 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:0.00013475492596626282 - timing_s/agent_loop/generate_sequences/min:np.float64(1.5073618683964014) - timing_s/agent_loop/generate_sequences/max:np.float64(7.373855691403151) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.293147319694981) - timing_s/agent_loop/tool_calls/min:np.float64(0.
|
|
step:26 - global_seqlen/min:16067 - global_seqlen/max:20476 - global_seqlen/minmax_diff:4409 - global_seqlen/balanced_min:17902 - global_seqlen/balanced_max:17905 - global_seqlen/mean:17903.75 - actor/entropy:0.4861890375614166 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1513535976409912 - training/rollout_probs_diff_mean:0.004183372016996145 - training/rollout_probs_diff_std:0.008406423032283783 - training/rollout_actor_probs_pearson_corr:0.9995557069778442 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8046875 - 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:0.8046875 - rollout_corr/rollout_is_mean:0.99986732006073 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.3731544017791748 - rollout_corr/rollout_is_min:0.6241110563278198 - rollout_corr/rollout_is_std:0.030226146802306175 - rollout_corr/rollout_is_eff_sample_size:0.9990869759610037 - rollout_corr/rollout_is_seq_mean:0.9998456239700317 - rollout_corr/rollout_is_seq_std:0.0020123575814068317 - rollout_corr/rollout_is_seq_max:1.005859136581421 - rollout_corr/rollout_is_seq_min:0.9947313070297241 - rollout_corr/rollout_is_seq_max_deviation:0.0058591365814208984 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6457447810098529) - rollout_corr/training_log_ppl:np.float64(0.4882779043982737) - rollout_corr/kl:np.float64(0.0006208247945327372) - rollout_corr/k3_kl:np.float64(0.00046650768092604267) - rollout_corr/rollout_ppl:np.float64(1.644661356229335) - rollout_corr/rollout_log_ppl:np.float64(0.4876570787746459) - rollout_corr/log_ppl_diff:np.float64(0.0006208256236277521) - rollout_corr/log_ppl_abs_diff:np.float64(0.001696924737188965) - rollout_corr/log_ppl_diff_max:np.float64(0.005778610706329346) - rollout_corr/log_ppl_diff_min:np.float64(-0.005435466766357422) - rollout_corr/ppl_ratio:np.float64(1.0006230880971998) - rollout_corr/chi2_token:np.float64(0.0006225099787116051) - rollout_corr/chi2_seq:np.float64(0.20502409618347883) - actor/pg_loss:np.float64(0.02809684054227546) - 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(0.1499176025390625) - perf/mfu/actor:np.float64(0.04176400110930352) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.79226684570312) - actor/lr:np.float64(1e-05) - training/global_step:26 - 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:0.004864513874053955 - critic/advantages/max:0.875 - critic/advantages/min:-0.625 - critic/returns/mean:0.004864513874053955 - critic/returns/max:0.875 - critic/returns/min:-0.625 - response_length/mean:263.0859375 - response_length/max:731.0 - response_length/min:132.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:263.0859375 - response_length_non_aborted/max:731.0 - response_length_non_aborted/min:132.0 - 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:0.00013700872659683228 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8560931850224733) - timing_s/agent_loop/generate_sequences/max:np.float64(6.419257763773203) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.315842459029227) - timing_s/agent_loop/tool_calls/min
|
|
step:27 - global_seqlen/min:15309 - global_seqlen/max:20923 - global_seqlen/minmax_diff:5614 - global_seqlen/balanced_min:18133 - global_seqlen/balanced_max:18378 - global_seqlen/mean:18194.5 - actor/entropy:0.4305836260318756 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23105758428573608 - training/rollout_probs_diff_mean:0.0036753243766725063 - training/rollout_probs_diff_std:0.00794100109487772 - training/rollout_actor_probs_pearson_corr:0.9995697736740112 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.92578125 - self_distillation/correct_sample_fraction:0.65234375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.92578125 - rollout_corr/rollout_is_mean:0.9999992251396179 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4467999935150146 - rollout_corr/rollout_is_min:0.6252691745758057 - rollout_corr/rollout_is_std:0.028309786692261696 - rollout_corr/rollout_is_eff_sample_size:0.9991991977445559 - rollout_corr/rollout_is_seq_mean:1.0000882148742676 - rollout_corr/rollout_is_seq_std:0.0018061090959236026 - rollout_corr/rollout_is_seq_max:1.0055360794067383 - rollout_corr/rollout_is_seq_min:0.9935215711593628 - rollout_corr/rollout_is_seq_max_deviation:0.006478428840637207 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.613325305748731) - rollout_corr/training_log_ppl:np.float64(0.46383187526953407) - rollout_corr/kl:np.float64(0.00034050049212908107) - rollout_corr/k3_kl:np.float64(0.00042870525533089676) - rollout_corr/rollout_ppl:np.float64(1.6127648334950209) - rollout_corr/rollout_log_ppl:np.float64(0.463491373549914) - rollout_corr/log_ppl_diff:np.float64(0.0003405017196200788) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014521153061650693) - rollout_corr/log_ppl_diff_max:np.float64(0.006880730390548706) - rollout_corr/log_ppl_diff_min:np.float64(-0.005160033702850342) - rollout_corr/ppl_ratio:np.float64(1.0003421721048653) - rollout_corr/chi2_token:np.float64(0.0010401380714029074) - rollout_corr/chi2_seq:np.float64(0.27827069256454706) - actor/pg_loss:np.float64(0.028016211595968343) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.07421875) - actor/grad_norm:np.float64(0.14197123050689697) - perf/mfu/actor:np.float64(0.04219079167790614) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.84881210327148) - actor/lr:np.float64(1e-05) - training/global_step:27 - training/epoch:0 - critic/score/mean:0.65234375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.65234375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0013475649757310748 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0013475649757310748 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:310.890625 - response_length/max:1149.0 - response_length/min:108.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:310.890625 - response_length_non_aborted/max:1149.0 - response_length_non_aborted/min:108.0 - 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:0.00017370283603668213 - timing_s/agent_loop/generate_sequences/min:np.float64(1.497892690822482) - timing_s/agent_loop/generate_sequences/max:np.float64(9.009137455374002) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.6840938990790164) - timing_s/agent_loop/to
|
|
step:28 - global_seqlen/min:16262 - global_seqlen/max:20364 - global_seqlen/minmax_diff:4102 - global_seqlen/balanced_min:18049 - global_seqlen/balanced_max:18054 - global_seqlen/mean:18051.5 - actor/entropy:0.4668128788471222 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12393772602081299 - training/rollout_probs_diff_mean:0.004177871625870466 - training/rollout_probs_diff_std:0.008510088548064232 - training/rollout_actor_probs_pearson_corr:0.9995258450508118 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.875 - self_distillation/correct_sample_fraction:0.703125 - 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:0.9999386072158813 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4428551197052002 - rollout_corr/rollout_is_min:0.6647944450378418 - rollout_corr/rollout_is_std:0.030269499868154526 - rollout_corr/rollout_is_eff_sample_size:0.9990843581503082 - rollout_corr/rollout_is_seq_mean:0.999984085559845 - rollout_corr/rollout_is_seq_std:0.0018264243844896555 - rollout_corr/rollout_is_seq_max:1.0060639381408691 - rollout_corr/rollout_is_seq_min:0.9941276907920837 - rollout_corr/rollout_is_seq_max_deviation:0.006063938140869141 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6296971552073956) - rollout_corr/training_log_ppl:np.float64(0.4774915195303038) - rollout_corr/kl:np.float64(0.0004894325135325062) - rollout_corr/k3_kl:np.float64(0.00047351719280186444) - rollout_corr/rollout_ppl:np.float64(1.6288958638906479) - rollout_corr/rollout_log_ppl:np.float64(0.4770020842552185) - rollout_corr/log_ppl_diff:np.float64(0.0004894352750852704) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014891911996528506) - rollout_corr/log_ppl_diff_max:np.float64(0.006566643714904785) - rollout_corr/log_ppl_diff_min:np.float64(-0.005632132291793823) - rollout_corr/ppl_ratio:np.float64(1.0004912153817713) - rollout_corr/chi2_token:np.float64(0.0009204922243952751) - rollout_corr/chi2_seq:np.float64(0.1660837340168655) - actor/pg_loss:np.float64(0.030589336332923267) - 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(0.14088349044322968) - perf/mfu/actor:np.float64(0.042166312827621004) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.78536558151245) - actor/lr:np.float64(1e-05) - training/global_step:28 - training/epoch:0 - critic/score/mean:0.703125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.703125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0017107236199080944 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.0017107236199080944 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:291.703125 - response_length/max:779.0 - response_length/min:127.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:291.703125 - response_length_non_aborted/max:779.0 - response_length_non_aborted/min:127.0 - 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:0.00017385929822921753 - timing_s/agent_loop/generate_sequences/min:np.float64(1.681123400107026) - timing_s/agent_loop/generate_sequences/max:np.float64(6.849359702318907) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.6768979362386744) - timing_s/agent_loop/tool_calls/min:np.float64(0.0)
|
|
step:29 - global_seqlen/min:13820 - global_seqlen/max:23114 - global_seqlen/minmax_diff:9294 - global_seqlen/balanced_min:18985 - global_seqlen/balanced_max:18995 - global_seqlen/mean:18991.75 - actor/entropy:0.4268430173397064 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.14973735809326172 - training/rollout_probs_diff_mean:0.003606261685490608 - training/rollout_probs_diff_std:0.007850543595850468 - training/rollout_actor_probs_pearson_corr:0.9996131658554077 - self_distillation/success_group_fraction:0.75 - self_distillation/success_sample_fraction:0.73828125 - self_distillation/correct_sample_fraction:0.578125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.73828125 - rollout_corr/rollout_is_mean:0.9999144673347473 - rollout_corr/rollout_is_ratio_fraction_high:1.233258535648929e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.256157159805298 - rollout_corr/rollout_is_min:0.7150127291679382 - rollout_corr/rollout_is_std:0.028309786692261696 - rollout_corr/rollout_is_eff_sample_size:0.9991989597077329 - rollout_corr/rollout_is_seq_mean:0.999914288520813 - rollout_corr/rollout_is_seq_std:0.0020249306689947844 - rollout_corr/rollout_is_seq_max:1.0086870193481445 - rollout_corr/rollout_is_seq_min:0.9941728711128235 - rollout_corr/rollout_is_seq_max_deviation:0.008687019348144531 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.598486723843962) - rollout_corr/training_log_ppl:np.float64(0.4570834499900229) - rollout_corr/kl:np.float64(0.0005140797892728699) - rollout_corr/k3_kl:np.float64(0.0004283656043639894) - rollout_corr/rollout_ppl:np.float64(1.5976283815689385) - rollout_corr/rollout_log_ppl:np.float64(0.4565693656040821) - rollout_corr/log_ppl_diff:np.float64(0.0005140843859408051) - rollout_corr/log_ppl_abs_diff:np.float64(0.001555114722577855) - rollout_corr/log_ppl_diff_max:np.float64(0.006442546844482422) - rollout_corr/log_ppl_diff_min:np.float64(-0.007996797561645508) - rollout_corr/ppl_ratio:np.float64(1.0005162707529962) - rollout_corr/chi2_token:np.float64(0.000691611785441637) - rollout_corr/chi2_seq:np.float64(0.2926765151787549) - actor/pg_loss:np.float64(0.02317226808372652) - 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(0.12512299418449402) - perf/mfu/actor:np.float64(0.0438484304068667) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.79710006713867) - actor/lr:np.float64(1e-05) - training/global_step:29 - training/epoch:0 - critic/score/mean:0.578125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.578125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0005904225399717689 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0005904225399717689 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:316.7421875 - response_length/max:1104.0 - response_length/min:123.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:316.7421875 - response_length_non_aborted/max:1104.0 - response_length_non_aborted/min:123.0 - 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:0.00011042691767215729 - timing_s/agent_loop/generate_sequences/min:np.float64(1.5206721350550652) - timing_s/agent_loop/generate_sequences/max:np.float64(8.448734015226364) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.715486570712528) - timing_s/agent_loop/
|
|
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 target product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" is a nitrogen-containing compound with a chlorinated aromatic ring, a trifluoromethyl group (C(F)(F)F), and a nitrogen-containing heterocycle. To identify the correct reactant, we must match the functional groups and structural features of the product.
|
|
|
|
- **Option A** contains the chlorinated aromatic ring (with a Cl), the trifluoromethyl group (C(F)(F)F), and a nitrogen-containing heterocycle, which matches the structure of the target product.
|
|
- **Option B** includes a different functional group (a hydroxyl group and a different aromatic substitution pattern), which does not match.
|
|
- **Option C** has an oxygen-containing group (hydroxyl) and an altered aromatic substitution pattern, which does not match.
|
|
- **Option D** includes a different fluorinated group and an incorrect ring structure, which also does not match.
|
|
|
|
Therefore, **Option A** is the only correct reactant that can be used in the synthesis of the given product.
|
|
</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-Qwen3-8B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_30
|
|
step:30 - global_seqlen/min:16791 - global_seqlen/max:23966 - global_seqlen/minmax_diff:7175 - global_seqlen/balanced_min:19289 - global_seqlen/balanced_max:19292 - global_seqlen/mean:19291.0 - actor/entropy:0.4370291829109192 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.14796662330627441 - training/rollout_probs_diff_mean:0.0038435272872447968 - training/rollout_probs_diff_std:0.00815175287425518 - training/rollout_actor_probs_pearson_corr:0.9995549321174622 - self_distillation/success_group_fraction:0.78125 - self_distillation/success_sample_fraction:0.74609375 - self_distillation/correct_sample_fraction:0.46484375 - 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:0.9998674988746643 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4941314458847046 - rollout_corr/rollout_is_min:0.6642436385154724 - rollout_corr/rollout_is_std:0.02865091897547245 - rollout_corr/rollout_is_eff_sample_size:0.9991795600879328 - rollout_corr/rollout_is_seq_mean:0.9998394846916199 - rollout_corr/rollout_is_seq_std:0.0017873558681458235 - rollout_corr/rollout_is_seq_max:1.0070340633392334 - rollout_corr/rollout_is_seq_min:0.9938014149665833 - rollout_corr/rollout_is_seq_max_deviation:0.0070340633392333984 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.591208673082292) - rollout_corr/training_log_ppl:np.float64(0.45267516531748697) - rollout_corr/kl:np.float64(0.0005911770146411754) - rollout_corr/k3_kl:np.float64(0.0004306302844554466) - rollout_corr/rollout_ppl:np.float64(1.5902782995253801) - rollout_corr/rollout_log_ppl:np.float64(0.45208398398244753) - rollout_corr/log_ppl_diff:np.float64(0.0005911813350394368) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014274888671934605) - rollout_corr/log_ppl_diff_max:np.float64(0.007044732570648193) - rollout_corr/log_ppl_diff_min:np.float64(-0.0062375664710998535) - rollout_corr/ppl_ratio:np.float64(1.000592951895669) - rollout_corr/chi2_token:np.float64(0.0005455284845083952) - rollout_corr/chi2_seq:np.float64(0.12012372142635286) - actor/pg_loss:np.float64(0.02229843009627075) - 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(0.13140323758125305) - perf/mfu/actor:np.float64(0.044951519198073386) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.60955429077148) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7080357142857143) - val-aux/sciknoweval/reward/std@16:np.float64(0.12276213250166364) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7581904761904761) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.11268263769056461) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6576047619047618) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.09505268189304389) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7083619047619047) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.12248134144974414) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.8044904761904762) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.09181517616699271) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6221142857142857) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.06372987511630503) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7094714285714286) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.10076923821304408) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.842304761904762) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.06542507169407996) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.5979571428571429) - val-aux/sciknoweval/r
|
|
step:31 - global_seqlen/min:16286 - global_seqlen/max:22783 - global_seqlen/minmax_diff:6497 - global_seqlen/balanced_min:19336 - global_seqlen/balanced_max:19346 - global_seqlen/mean:19343.75 - actor/entropy:0.4036998152732849 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18442967534065247 - training/rollout_probs_diff_mean:0.0035835476592183113 - training/rollout_probs_diff_std:0.00790021289139986 - training/rollout_actor_probs_pearson_corr:0.9995797276496887 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80859375 - 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:0.80859375 - rollout_corr/rollout_is_mean:0.9999337196350098 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.552642583847046 - rollout_corr/rollout_is_min:0.6482400298118591 - rollout_corr/rollout_is_std:0.028089968487620354 - rollout_corr/rollout_is_eff_sample_size:0.999211575815639 - rollout_corr/rollout_is_seq_mean:0.9999220371246338 - rollout_corr/rollout_is_seq_std:0.001655145431868732 - rollout_corr/rollout_is_seq_max:1.0047979354858398 - rollout_corr/rollout_is_seq_min:0.9949408769607544 - rollout_corr/rollout_is_seq_max_deviation:0.0050591230392456055 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5449855513870716) - rollout_corr/training_log_ppl:np.float64(0.4239463791018352) - rollout_corr/kl:np.float64(0.00049447579929085) - rollout_corr/k3_kl:np.float64(0.00041651196522707323) - rollout_corr/rollout_ppl:np.float64(1.5442148866131902) - rollout_corr/rollout_log_ppl:np.float64(0.4234518989978824) - rollout_corr/log_ppl_diff:np.float64(0.0004944801039528102) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013447333767544478) - rollout_corr/log_ppl_diff_max:np.float64(0.005715131759643555) - rollout_corr/log_ppl_diff_min:np.float64(-0.004388183355331421) - rollout_corr/ppl_ratio:np.float64(1.0004959942307323) - rollout_corr/chi2_token:np.float64(0.000681284349411726) - rollout_corr/chi2_seq:np.float64(0.18186982325278223) - actor/pg_loss:np.float64(0.023701957947196206) - 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(0.12994642555713654) - perf/mfu/actor:np.float64(0.044567037430020676) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.90562438964844) - actor/lr:np.float64(1e-05) - training/global_step:31 - training/epoch:0 - 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:-0.0003654794709291309 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0003654794709291309 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:324.6484375 - response_length/max:1078.0 - response_length/min:138.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:324.6484375 - response_length_non_aborted/max:1078.0 - response_length_non_aborted/min:138.0 - 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:8.709914982318878e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8712326791137457) - timing_s/agent_loop/generate_sequences/max:np.float64(8.549193287268281) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.9295072019085637) - timing_s/agent_loop/to
|
|
step:32 - global_seqlen/min:15042 - global_seqlen/max:20959 - global_seqlen/minmax_diff:5917 - global_seqlen/balanced_min:17996 - global_seqlen/balanced_max:18004 - global_seqlen/mean:18001.875 - actor/entropy:0.4281376600265503 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.122444748878479 - training/rollout_probs_diff_mean:0.0037491386756300926 - training/rollout_probs_diff_std:0.008076412603259087 - training/rollout_actor_probs_pearson_corr:0.9995706081390381 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - 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:0.83203125 - rollout_corr/rollout_is_mean:1.0000841617584229 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5177969932556152 - rollout_corr/rollout_is_min:0.6295897960662842 - rollout_corr/rollout_is_std:0.028709109872579575 - rollout_corr/rollout_is_eff_sample_size:0.999176584752751 - rollout_corr/rollout_is_seq_mean:1.0000839233398438 - rollout_corr/rollout_is_seq_std:0.0017924157436937094 - rollout_corr/rollout_is_seq_max:1.0059983730316162 - rollout_corr/rollout_is_seq_min:0.9946850538253784 - rollout_corr/rollout_is_seq_max_deviation:0.005998373031616211 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.566798876505345) - rollout_corr/training_log_ppl:np.float64(0.4383359969360754) - rollout_corr/kl:np.float64(0.0003456925159653901) - rollout_corr/k3_kl:np.float64(0.00042963139374307957) - rollout_corr/rollout_ppl:np.float64(1.5661763339303434) - rollout_corr/rollout_log_ppl:np.float64(0.4379902995424345) - rollout_corr/log_ppl_diff:np.float64(0.0003456973936408758) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014006176497787237) - rollout_corr/log_ppl_diff_max:np.float64(0.006148338317871094) - rollout_corr/log_ppl_diff_min:np.float64(-0.005340218544006348) - rollout_corr/ppl_ratio:np.float64(1.0003474047407508) - rollout_corr/chi2_token:np.float64(0.0010282525327056646) - rollout_corr/chi2_seq:np.float64(0.39566941489465535) - actor/pg_loss:np.float64(0.025634928795625456) - 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(0.12805183231830597) - perf/mfu/actor:np.float64(0.041760252927658635) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.7085189819336) - actor/lr:np.float64(1e-05) - training/global_step:32 - 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:-0.00042195056448690593 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00042195056448690593 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:299.71484375 - response_length/max:880.0 - response_length/min:154.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:299.71484375 - response_length_non_aborted/max:880.0 - response_length_non_aborted/min:154.0 - 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:9.644590318202972e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9897665549069643) - timing_s/agent_loop/generate_sequences/max:np.float64(7.183622565120459) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.580246256402461) - timing_s/agent_loo
|
|
step:33 - global_seqlen/min:16104 - global_seqlen/max:20808 - global_seqlen/minmax_diff:4704 - global_seqlen/balanced_min:18707 - global_seqlen/balanced_max:18799 - global_seqlen/mean:18723.375 - actor/entropy:0.44116878509521484 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.15575635433197021 - training/rollout_probs_diff_mean:0.0038432145956903696 - training/rollout_probs_diff_std:0.008133911527693272 - training/rollout_actor_probs_pearson_corr:0.9995786547660828 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.80078125 - 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:0.80078125 - rollout_corr/rollout_is_mean:0.9999343156814575 - rollout_corr/rollout_is_ratio_fraction_high:1.3407162441581022e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.2927515506744385 - rollout_corr/rollout_is_min:0.6440492272377014 - rollout_corr/rollout_is_std:0.028802385553717613 - rollout_corr/rollout_is_eff_sample_size:0.9991707531471955 - rollout_corr/rollout_is_seq_mean:0.9999381899833679 - rollout_corr/rollout_is_seq_std:0.0017237946158275008 - rollout_corr/rollout_is_seq_max:1.004521131515503 - rollout_corr/rollout_is_seq_min:0.9951814413070679 - rollout_corr/rollout_is_seq_max_deviation:0.004818558692932129 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5877922726795077) - rollout_corr/training_log_ppl:np.float64(0.451917833997868) - rollout_corr/kl:np.float64(0.0004878851554011021) - rollout_corr/k3_kl:np.float64(0.0004261053974516926) - rollout_corr/rollout_ppl:np.float64(1.587018565274775) - rollout_corr/rollout_log_ppl:np.float64(0.4514299462316558) - rollout_corr/log_ppl_diff:np.float64(0.00048788776621222496) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014335069572553039) - rollout_corr/log_ppl_diff_max:np.float64(0.005312144756317139) - rollout_corr/log_ppl_diff_min:np.float64(-0.003967940807342529) - rollout_corr/ppl_ratio:np.float64(1.0004894933663309) - rollout_corr/chi2_token:np.float64(0.000738984439522028) - rollout_corr/chi2_seq:np.float64(0.18156872992403805) - actor/pg_loss:np.float64(0.024620033291284926) - 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(0.13329696655273438) - perf/mfu/actor:np.float64(0.044006381532965086) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.2401008605957) - actor/lr:np.float64(1e-05) - training/global_step:33 - 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:-0.0016055076848715544 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0016055076848715544 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:291.35546875 - response_length/max:965.0 - response_length/min:141.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:291.35546875 - response_length_non_aborted/max:965.0 - response_length_non_aborted/min:141.0 - 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:0.00016062334179878235 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9058470353484154) - timing_s/agent_loop/generate_sequences/max:np.float64(7.637895109131932) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.7012352487290627) - ti
|
|
step:34 - global_seqlen/min:17172 - global_seqlen/max:23013 - global_seqlen/minmax_diff:5841 - global_seqlen/balanced_min:19124 - global_seqlen/balanced_max:19133 - global_seqlen/mean:19130.625 - actor/entropy:0.4375230371952057 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.30402058362960815 - training/rollout_probs_diff_mean:0.003805038519203663 - training/rollout_probs_diff_std:0.00835459679365158 - training/rollout_actor_probs_pearson_corr:0.99954754114151 - self_distillation/success_group_fraction:0.96875 - self_distillation/success_sample_fraction:0.953125 - 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:0.953125 - rollout_corr/rollout_is_mean:0.9998382329940796 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.5694720534374937e-05 - rollout_corr/rollout_is_max:1.5163826942443848 - rollout_corr/rollout_is_min:0.4629283547401428 - rollout_corr/rollout_is_std:0.028806524351239204 - rollout_corr/rollout_is_eff_sample_size:0.999170634135546 - rollout_corr/rollout_is_seq_mean:0.9998770952224731 - rollout_corr/rollout_is_seq_std:0.0017865466652438045 - rollout_corr/rollout_is_seq_max:1.0056116580963135 - rollout_corr/rollout_is_seq_min:0.9949389696121216 - rollout_corr/rollout_is_seq_max_deviation:0.0056116580963134766 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5851375460624695) - rollout_corr/training_log_ppl:np.float64(0.44951163048972376) - rollout_corr/kl:np.float64(0.0005517449747074465) - rollout_corr/k3_kl:np.float64(0.0004288690167868481) - rollout_corr/rollout_ppl:np.float64(1.5842331554740667) - rollout_corr/rollout_log_ppl:np.float64(0.4489598837099038) - rollout_corr/log_ppl_diff:np.float64(0.0005517467798199505) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014524034631904215) - rollout_corr/log_ppl_diff_max:np.float64(0.005543172359466553) - rollout_corr/log_ppl_diff_min:np.float64(-0.005078494548797607) - rollout_corr/ppl_ratio:np.float64(1.000553488265723) - rollout_corr/chi2_token:np.float64(0.0006128717213869095) - rollout_corr/chi2_seq:np.float64(0.11041362024843693) - actor/pg_loss:np.float64(0.02848288041059277) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.046875) - actor/grad_norm:np.float64(0.1472928375005722) - perf/mfu/actor:np.float64(0.04370032323590719) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.25754165649414) - actor/lr:np.float64(1e-05) - training/global_step:34 - training/epoch:0 - critic/score/mean:0.7109375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7109375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0011209321673959494 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0011209321673959494 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:304.05078125 - response_length/max:1003.0 - response_length/min:144.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:304.05078125 - response_length_non_aborted/max:1003.0 - response_length_non_aborted/min:144.0 - 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:0.00012355297803878784 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9959246180951595) - timing_s/agent_loop/generate_sequences/max:np.float64(7.991340881213546) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.725371319516853) - timing_s/a
|
|
step:35 - global_seqlen/min:17114 - global_seqlen/max:21206 - global_seqlen/minmax_diff:4092 - global_seqlen/balanced_min:19443 - global_seqlen/balanced_max:19653 - global_seqlen/mean:19472.375 - actor/entropy:0.37307053804397583 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1259000301361084 - training/rollout_probs_diff_mean:0.0032313757110387087 - training/rollout_probs_diff_std:0.007502178195863962 - training/rollout_actor_probs_pearson_corr:0.9996042251586914 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.85546875 - 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:0.85546875 - rollout_corr/rollout_is_mean:0.9999658465385437 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5713344812393188 - rollout_corr/rollout_is_min:0.6387621760368347 - rollout_corr/rollout_is_std:0.026190269738435745 - rollout_corr/rollout_is_eff_sample_size:0.9993143018339448 - rollout_corr/rollout_is_seq_mean:0.9999569654464722 - rollout_corr/rollout_is_seq_std:0.0015237914631143212 - rollout_corr/rollout_is_seq_max:1.0065796375274658 - rollout_corr/rollout_is_seq_min:0.9956242442131042 - rollout_corr/rollout_is_seq_max_deviation:0.00657963752746582 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4854215821251273) - rollout_corr/training_log_ppl:np.float64(0.38364325487054884) - rollout_corr/kl:np.float64(0.0004007244738568616) - rollout_corr/k3_kl:np.float64(0.0003576851627826727) - rollout_corr/rollout_ppl:np.float64(1.484797871671617) - rollout_corr/rollout_log_ppl:np.float64(0.3832425260625314) - rollout_corr/log_ppl_diff:np.float64(0.0004007288080174476) - rollout_corr/log_ppl_abs_diff:np.float64(0.0011870984162669629) - rollout_corr/log_ppl_diff_max:np.float64(0.004746556282043457) - rollout_corr/log_ppl_diff_min:np.float64(-0.0058882832527160645) - rollout_corr/ppl_ratio:np.float64(1.0004019455518574) - rollout_corr/chi2_token:np.float64(0.0006346092559397221) - rollout_corr/chi2_seq:np.float64(0.2272831026930362) - actor/pg_loss:np.float64(0.022161945465995814) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.14453125) - actor/grad_norm:np.float64(0.11812028288841248) - perf/mfu/actor:np.float64(0.045221251589272246) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.0724754333496) - actor/lr:np.float64(1e-05) - training/global_step:35 - 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:-0.00274647306650877 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00274647306650877 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:330.32421875 - response_length/max:1180.0 - response_length/min:150.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:330.32421875 - response_length_non_aborted/max:1180.0 - response_length_non_aborted/min:150.0 - 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:0.0001440141350030899 - timing_s/agent_loop/generate_sequences/min:np.float64(2.016096482053399) - timing_s/agent_loop/generate_sequences/max:np.float64(9.252621490508318) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.98015643555118) - timing_s/agent_loop/tool_
|
|
step:36 - global_seqlen/min:13807 - global_seqlen/max:22289 - global_seqlen/minmax_diff:8482 - global_seqlen/balanced_min:18504 - global_seqlen/balanced_max:18507 - global_seqlen/mean:18506.0 - actor/entropy:0.4335061013698578 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18137136101722717 - training/rollout_probs_diff_mean:0.0037304568104445934 - training/rollout_probs_diff_std:0.007991302758455276 - training/rollout_actor_probs_pearson_corr:0.9995746612548828 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.9296875 - 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:0.9296875 - rollout_corr/rollout_is_mean:1.0003021955490112 - rollout_corr/rollout_is_ratio_fraction_high:1.3007283996557817e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:4.5563740730285645 - rollout_corr/rollout_is_min:0.6244023442268372 - rollout_corr/rollout_is_std:0.029072217643260956 - rollout_corr/rollout_is_eff_sample_size:0.9991559959188603 - rollout_corr/rollout_is_seq_mean:1.0003197193145752 - rollout_corr/rollout_is_seq_std:0.001922975992783904 - rollout_corr/rollout_is_seq_max:1.005760669708252 - rollout_corr/rollout_is_seq_min:0.9958750605583191 - rollout_corr/rollout_is_seq_max_deviation:0.005760669708251953 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5684201624244452) - rollout_corr/training_log_ppl:np.float64(0.43929985890281387) - rollout_corr/kl:np.float64(0.00011283910309911604) - rollout_corr/k3_kl:np.float64(0.0004325444901098763) - rollout_corr/rollout_ppl:np.float64(1.5682631442323327) - rollout_corr/rollout_log_ppl:np.float64(0.4391870167746674) - rollout_corr/log_ppl_diff:np.float64(0.00011284212814643979) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014980111154727638) - rollout_corr/log_ppl_diff_max:np.float64(0.004500165581703186) - rollout_corr/log_ppl_diff_min:np.float64(-0.0051877498626708984) - rollout_corr/ppl_ratio:np.float64(1.000114627648145) - rollout_corr/chi2_token:np.float64(0.001557792304083705) - rollout_corr/chi2_seq:np.float64(0.7567500455770642) - actor/pg_loss:np.float64(0.02843063971886295) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.0703125) - actor/grad_norm:np.float64(0.12622880935668945) - perf/mfu/actor:np.float64(0.04258589085878693) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.0333662033081) - actor/lr:np.float64(1e-05) - training/global_step:36 - training/epoch:0 - critic/score/mean:0.734375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.734375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0007479188498109579 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0007479188498109579 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:300.3125 - response_length/max:752.0 - response_length/min:152.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:300.3125 - response_length_non_aborted/max:752.0 - response_length_non_aborted/min:152.0 - 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:0.00011322088539600372 - timing_s/agent_loop/generate_sequences/min:np.float64(2.054821155965328) - timing_s/agent_loop/generate_sequences/max:np.float64(6.598034670576453) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.7035879743198166) - timing_s/agent_loop/tool
|
|
step:37 - global_seqlen/min:15957 - global_seqlen/max:23278 - global_seqlen/minmax_diff:7321 - global_seqlen/balanced_min:19180 - global_seqlen/balanced_max:19186 - global_seqlen/mean:19181.625 - actor/entropy:0.43448370695114136 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.8773926496505737 - training/rollout_probs_diff_mean:0.0038428392726927996 - training/rollout_probs_diff_std:0.00864581111818552 - training/rollout_actor_probs_pearson_corr:0.9994995594024658 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - 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:0.8359375 - rollout_corr/rollout_is_mean:0.9999868869781494 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.362750572297955e-05 - rollout_corr/rollout_is_max:1.6952898502349854 - rollout_corr/rollout_is_min:0.10841354727745056 - rollout_corr/rollout_is_std:0.0290681179612875 - rollout_corr/rollout_is_eff_sample_size:0.9991557579026206 - rollout_corr/rollout_is_seq_mean:1.0000452995300293 - rollout_corr/rollout_is_seq_std:0.001684858463704586 - rollout_corr/rollout_is_seq_max:1.0051732063293457 - rollout_corr/rollout_is_seq_min:0.994458019733429 - rollout_corr/rollout_is_seq_max_deviation:0.005541980266571045 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5745648629963398) - rollout_corr/training_log_ppl:np.float64(0.44478763386723585) - rollout_corr/kl:np.float64(0.00039999501963450257) - rollout_corr/k3_kl:np.float64(0.0004452134256212048) - rollout_corr/rollout_ppl:np.float64(1.573883724398911) - rollout_corr/rollout_log_ppl:np.float64(0.44438763763173483) - rollout_corr/log_ppl_diff:np.float64(0.00039999623550102115) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013491098652593791) - rollout_corr/log_ppl_diff_max:np.float64(0.005976438522338867) - rollout_corr/log_ppl_diff_min:np.float64(-0.004648283123970032) - rollout_corr/ppl_ratio:np.float64(1.0004015662707388) - rollout_corr/chi2_token:np.float64(0.000969236483797431) - rollout_corr/chi2_seq:np.float64(0.1404132905881852) - actor/pg_loss:np.float64(0.023940196359035326) - 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(0.1259196400642395) - perf/mfu/actor:np.float64(0.04454120884749763) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.99682235717773) - actor/lr:np.float64(1e-05) - training/global_step:37 - 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:-0.0006864856113679707 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0006864856113679707 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:286.64453125 - response_length/max:891.0 - response_length/min:128.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:286.64453125 - response_length_non_aborted/max:891.0 - response_length_non_aborted/min:128.0 - 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:0.00013086386024951935 - timing_s/agent_loop/generate_sequences/min:np.float64(1.6448498629033566) - timing_s/agent_loop/generate_sequences/max:np.float64(7.376331951469183) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.5574712707748404) - tim
|
|
step:38 - global_seqlen/min:14393 - global_seqlen/max:20244 - global_seqlen/minmax_diff:5851 - global_seqlen/balanced_min:17331 - global_seqlen/balanced_max:17428 - global_seqlen/mean:17356.125 - actor/entropy:0.42084091901779175 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12431871891021729 - training/rollout_probs_diff_mean:0.0038049763534218073 - training/rollout_probs_diff_std:0.008086569607257843 - training/rollout_actor_probs_pearson_corr:0.999581515789032 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8046875 - self_distillation/correct_sample_fraction:0.6796875 - 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:0.9999146461486816 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.44087553024292 - rollout_corr/rollout_is_min:0.6290515065193176 - rollout_corr/rollout_is_std:0.0283203125 - rollout_corr/rollout_is_eff_sample_size:0.9991983646161715 - rollout_corr/rollout_is_seq_mean:0.9998958706855774 - rollout_corr/rollout_is_seq_std:0.001821843208745122 - rollout_corr/rollout_is_seq_max:1.0056995153427124 - rollout_corr/rollout_is_seq_min:0.9946661591529846 - rollout_corr/rollout_is_seq_max_deviation:0.005699515342712402 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.545995867345482) - rollout_corr/training_log_ppl:np.float64(0.42715942094218917) - rollout_corr/kl:np.float64(0.0005149057349402852) - rollout_corr/k3_kl:np.float64(0.00041077048251736414) - rollout_corr/rollout_ppl:np.float64(1.545235754456371) - rollout_corr/rollout_log_ppl:np.float64(0.42664451236487366) - rollout_corr/log_ppl_diff:np.float64(0.0005149085773155093) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014873218024149537) - rollout_corr/log_ppl_diff_max:np.float64(0.005653500556945801) - rollout_corr/log_ppl_diff_min:np.float64(-0.005117893218994141) - rollout_corr/ppl_ratio:np.float64(1.000516687752679) - rollout_corr/chi2_token:np.float64(0.0006164936348795891) - rollout_corr/chi2_seq:np.float64(0.2632097185123712) - actor/pg_loss:np.float64(0.02422216336344718) - 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(0.11867456883192062) - perf/mfu/actor:np.float64(0.04062237987611797) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.0080909729004) - actor/lr:np.float64(1e-05) - training/global_step:38 - training/epoch:0 - critic/score/mean:0.6796875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6796875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.002299562096595764 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.002299562096595764 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:279.22265625 - response_length/max:922.0 - response_length/min:83.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:279.22265625 - response_length_non_aborted/max:922.0 - response_length_non_aborted/min:83.0 - 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:0.000164153054356575 - timing_s/agent_loop/generate_sequences/min:np.float64(1.0932764671742916) - timing_s/agent_loop/generate_sequences/max:np.float64(7.419106082990766) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.4318311656970764) - timing_s/agent_loop/tool_calls/min:np.float6
|
|
step:39 - global_seqlen/min:14545 - global_seqlen/max:21771 - global_seqlen/minmax_diff:7226 - global_seqlen/balanced_min:17996 - global_seqlen/balanced_max:17999 - global_seqlen/mean:17997.5 - actor/entropy:0.45629140734672546 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12435358762741089 - training/rollout_probs_diff_mean:0.0038324615452438593 - training/rollout_probs_diff_std:0.007934743538498878 - training/rollout_actor_probs_pearson_corr:0.9995771050453186 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - 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:0.8359375 - rollout_corr/rollout_is_mean:0.9999353885650635 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5794267654418945 - rollout_corr/rollout_is_min:0.6099376082420349 - rollout_corr/rollout_is_std:0.02862594462931156 - rollout_corr/rollout_is_eff_sample_size:0.9991812262833748 - rollout_corr/rollout_is_seq_mean:0.9998792409896851 - rollout_corr/rollout_is_seq_std:0.0018032982479780912 - rollout_corr/rollout_is_seq_max:1.0046807527542114 - rollout_corr/rollout_is_seq_min:0.9951657056808472 - rollout_corr/rollout_is_seq_max_deviation:0.004834294319152832 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6169907390139997) - rollout_corr/training_log_ppl:np.float64(0.4675352103658952) - rollout_corr/kl:np.float64(0.0005434988521386686) - rollout_corr/k3_kl:np.float64(0.0004227347644416568) - rollout_corr/rollout_ppl:np.float64(1.6160588348284364) - rollout_corr/rollout_log_ppl:np.float64(0.46699170814827085) - rollout_corr/log_ppl_diff:np.float64(0.0005435022176243365) - rollout_corr/log_ppl_abs_diff:np.float64(0.001488720707129687) - rollout_corr/log_ppl_diff_max:np.float64(0.005336940288543701) - rollout_corr/log_ppl_diff_min:np.float64(-0.004028201103210449) - rollout_corr/ppl_ratio:np.float64(1.000545282382518) - rollout_corr/chi2_token:np.float64(0.00060637341812253) - rollout_corr/chi2_seq:np.float64(0.14641686133109033) - actor/pg_loss:np.float64(0.023575008252009866) - 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(0.12797197699546814) - perf/mfu/actor:np.float64(0.04214042652847487) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.87594985961914) - actor/lr:np.float64(1e-05) - training/global_step:39 - 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:-0.004740922711789608 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.004740922711789608 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:280.140625 - response_length/max:618.0 - response_length/min:145.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:280.140625 - response_length_non_aborted/max:618.0 - response_length_non_aborted/min:145.0 - 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:3.0295923352241516e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9028393104672432) - timing_s/agent_loop/generate_sequences/max:np.float64(5.705321965739131) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.4019602277257945) - timing_s/agent_loop/tool_calls/min:np.float6
|
|
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 target product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" is a complex organic molecule with a specific structure that includes a chlorinated aromatic ring, a fluorinated aromatic ring (C(F)(F)F), and a nitrogen-containing heterocyclic structure.
|
|
|
|
- **Option A** matches the structure of the target product closely. It contains the correct chlorinated and fluorinated aromatic rings, and the nitrogen-containing heterocyclic structure. It is chemically consistent with the product and would be a valid reactant in its synthesis.
|
|
|
|
- **Option B** has a different arrangement of functional groups and includes a different ring system, making it chemically inconsistent with the target product.
|
|
|
|
- **Option C** introduces a hydroxyl group and a different ring structure, which deviates from the structure of the target product.
|
|
|
|
- **Option D** has a different ring arrangement and introduces a different functional group (a fluorinated carbon chain), which does not match the target structure.
|
|
|
|
Therefore, **Option A** is the only correct reactant that matches the structure of the target product.
|
|
</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-Qwen3-8B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_40
|
|
step:40 - global_seqlen/min:15314 - global_seqlen/max:22016 - global_seqlen/minmax_diff:6702 - global_seqlen/balanced_min:19024 - global_seqlen/balanced_max:19029 - global_seqlen/mean:19027.5 - actor/entropy:0.5039446949958801 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23105758428573608 - training/rollout_probs_diff_mean:0.0042390669696033 - training/rollout_probs_diff_std:0.008486345410346985 - training/rollout_actor_probs_pearson_corr:0.9995676875114441 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.8671875 - 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:0.8671875 - rollout_corr/rollout_is_mean:1.0000135898590088 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5621331930160522 - rollout_corr/rollout_is_min:0.5215955972671509 - rollout_corr/rollout_is_std:0.030887408182024956 - rollout_corr/rollout_is_eff_sample_size:0.9990468773651655 - rollout_corr/rollout_is_seq_mean:1.0000314712524414 - rollout_corr/rollout_is_seq_std:0.0019727509934455156 - rollout_corr/rollout_is_seq_max:1.0056904554367065 - rollout_corr/rollout_is_seq_min:0.9948930144309998 - rollout_corr/rollout_is_seq_max_deviation:0.005690455436706543 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.686801542993635) - rollout_corr/training_log_ppl:np.float64(0.5095907527720556) - rollout_corr/kl:np.float64(0.0004519636506952196) - rollout_corr/k3_kl:np.float64(0.00048338308391748797) - rollout_corr/rollout_ppl:np.float64(1.6860664249397814) - rollout_corr/rollout_log_ppl:np.float64(0.5091387865832075) - rollout_corr/log_ppl_diff:np.float64(0.00045196618884801865) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015859833220019937) - rollout_corr/log_ppl_diff_max:np.float64(0.005783170461654663) - rollout_corr/log_ppl_diff_min:np.float64(-0.00496518611907959) - rollout_corr/ppl_ratio:np.float64(1.0004539778456092) - rollout_corr/chi2_token:np.float64(0.0010328476782888174) - rollout_corr/chi2_seq:np.float64(0.24973847717046738) - actor/pg_loss:np.float64(0.02871005253837211) - 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(0.15812700986862183) - perf/mfu/actor:np.float64(0.043890751732743254) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(195.8145294189453) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7288690476190476) - val-aux/sciknoweval/reward/std@16:np.float64(0.11009279114684199) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7763238095238096) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.09847831214732486) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6814428571428572) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.08916271473463086) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7282809523809524) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.11010990043238834) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.8175142857142857) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.07437484762599954) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6462999999999999) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.05818850382726964) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7326238095238096) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.09385421718008423) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.8474047619047619) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.04611925933760916) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6246904761904762) - val-aux/sciknoweval/reward/wors
|
|
step:41 - global_seqlen/min:14249 - global_seqlen/max:21495 - global_seqlen/minmax_diff:7246 - global_seqlen/balanced_min:17802 - global_seqlen/balanced_max:17811 - global_seqlen/mean:17808.25 - actor/entropy:0.43595263361930847 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23411676287651062 - training/rollout_probs_diff_mean:0.0038285646587610245 - training/rollout_probs_diff_std:0.008288653567433357 - training/rollout_actor_probs_pearson_corr:0.9995604157447815 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.859375 - 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:0.859375 - rollout_corr/rollout_is_mean:1.0000262260437012 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.335006527369842e-05 - rollout_corr/rollout_is_max:1.4618793725967407 - rollout_corr/rollout_is_min:0.4748973846435547 - rollout_corr/rollout_is_std:0.028746457770466805 - rollout_corr/rollout_is_eff_sample_size:0.9991743235098636 - rollout_corr/rollout_is_seq_mean:0.9999955892562866 - rollout_corr/rollout_is_seq_std:0.0018859205301851034 - rollout_corr/rollout_is_seq_max:1.006566047668457 - rollout_corr/rollout_is_seq_min:0.993057370185852 - rollout_corr/rollout_is_seq_max_deviation:0.006942629814147949 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5769322160631418) - rollout_corr/training_log_ppl:np.float64(0.4439886717009358) - rollout_corr/kl:np.float64(0.0004313593544196248) - rollout_corr/k3_kl:np.float64(0.0004269851143305914) - rollout_corr/rollout_ppl:np.float64(1.5762130464427173) - rollout_corr/rollout_log_ppl:np.float64(0.44355731329414994) - rollout_corr/log_ppl_diff:np.float64(0.00043135840678587556) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014722268679179251) - rollout_corr/log_ppl_diff_max:np.float64(0.00752025842666626) - rollout_corr/log_ppl_diff_min:np.float64(-0.005875349044799805) - rollout_corr/ppl_ratio:np.float64(1.000433244742453) - rollout_corr/chi2_token:np.float64(0.0008481652475893497) - rollout_corr/chi2_seq:np.float64(0.3106381921097636) - actor/pg_loss:np.float64(0.02670324167047511) - 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(0.13134446740150452) - perf/mfu/actor:np.float64(0.04164591175846025) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(197.63422393798828) - actor/lr:np.float64(1e-05) - training/global_step:41 - 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:-0.00030204522772692144 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00030204522772692144 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:292.6015625 - response_length/max:902.0 - response_length/min:130.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:292.6015625 - response_length_non_aborted/max:902.0 - response_length_non_aborted/min:130.0 - 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:0.00014559738337993622 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8552009724080563) - timing_s/agent_loop/generate_sequences/max:np.float64(7.65681566670537) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.606885417284502) - timing_s/agent
|
|
step:42 - global_seqlen/min:14354 - global_seqlen/max:20830 - global_seqlen/minmax_diff:6476 - global_seqlen/balanced_min:17430 - global_seqlen/balanced_max:17433 - global_seqlen/mean:17432.125 - actor/entropy:0.3931979238986969 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12414106726646423 - training/rollout_probs_diff_mean:0.003629918210208416 - training/rollout_probs_diff_std:0.007999058812856674 - training/rollout_actor_probs_pearson_corr:0.9995500445365906 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.8984375 - 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:0.8984375 - rollout_corr/rollout_is_mean:1.0001120567321777 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.7285059690475464 - rollout_corr/rollout_is_min:0.6970924735069275 - rollout_corr/rollout_is_std:0.02792396768927574 - rollout_corr/rollout_is_eff_sample_size:0.9992210976174688 - rollout_corr/rollout_is_seq_mean:1.0000967979431152 - rollout_corr/rollout_is_seq_std:0.001920396345667541 - rollout_corr/rollout_is_seq_max:1.0077484846115112 - rollout_corr/rollout_is_seq_min:0.9925958514213562 - rollout_corr/rollout_is_seq_max_deviation:0.0077484846115112305 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5090870992280543) - rollout_corr/training_log_ppl:np.float64(0.4039846356608905) - rollout_corr/kl:np.float64(0.0002998333663869479) - rollout_corr/k3_kl:np.float64(0.000396598432985229) - rollout_corr/rollout_ppl:np.float64(1.5086088585667312) - rollout_corr/rollout_log_ppl:np.float64(0.40368479923927225) - rollout_corr/log_ppl_diff:np.float64(0.0002998364216182381) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013943387020844966) - rollout_corr/log_ppl_diff_max:np.float64(0.007930517196655273) - rollout_corr/log_ppl_diff_min:np.float64(-0.007044434547424316) - rollout_corr/ppl_ratio:np.float64(1.000301711494103) - rollout_corr/chi2_token:np.float64(0.0009944927878677845) - rollout_corr/chi2_seq:np.float64(0.3706525939051062) - actor/pg_loss:np.float64(0.025583154369087424) - 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(0.11673144996166229) - perf/mfu/actor:np.float64(0.04039517675815754) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.7127342224121) - actor/lr:np.float64(1e-05) - training/global_step:42 - training/epoch:0 - 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:-0.0022606798447668552 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0022606798447668552 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:288.12890625 - response_length/max:632.0 - response_length/min:105.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:288.12890625 - response_length_non_aborted/max:632.0 - response_length_non_aborted/min:105.0 - 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:0.00018003582954406738 - timing_s/agent_loop/generate_sequences/min:np.float64(1.3817657735198736) - timing_s/agent_loop/generate_sequences/max:np.float64(5.968047175556421) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.532574918383034) - timing_s/agent_loop/tool_call
|
|
step:43 - global_seqlen/min:16966 - global_seqlen/max:21720 - global_seqlen/minmax_diff:4754 - global_seqlen/balanced_min:18807 - global_seqlen/balanced_max:18810 - global_seqlen/mean:18808.25 - actor/entropy:0.42845863103866577 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1632150411605835 - training/rollout_probs_diff_mean:0.003866584040224552 - training/rollout_probs_diff_std:0.008193894289433956 - training/rollout_actor_probs_pearson_corr:0.9995288848876953 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.8671875 - self_distillation/correct_sample_fraction:0.703125 - 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:1.0000653266906738 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6093928813934326 - rollout_corr/rollout_is_min:0.6941616535186768 - rollout_corr/rollout_is_std:0.028548799455165863 - rollout_corr/rollout_is_eff_sample_size:0.999185867857122 - rollout_corr/rollout_is_seq_mean:1.000070333480835 - rollout_corr/rollout_is_seq_std:0.001752030453644693 - rollout_corr/rollout_is_seq_max:1.0067416429519653 - rollout_corr/rollout_is_seq_min:0.9952163100242615 - rollout_corr/rollout_is_seq_max_deviation:0.006741642951965332 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5639699404127896) - rollout_corr/training_log_ppl:np.float64(0.4395692787365988) - rollout_corr/kl:np.float64(0.0003443371413149876) - rollout_corr/k3_kl:np.float64(0.00041464497468268746) - rollout_corr/rollout_ppl:np.float64(1.5634347605518997) - rollout_corr/rollout_log_ppl:np.float64(0.4392249371157959) - rollout_corr/log_ppl_diff:np.float64(0.00034434162080287933) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013916371390223503) - rollout_corr/log_ppl_diff_max:np.float64(0.005381196737289429) - rollout_corr/log_ppl_diff_min:np.float64(-0.0062261223793029785) - rollout_corr/ppl_ratio:np.float64(1.0003459374420345) - rollout_corr/chi2_token:np.float64(0.0009754123166203499) - rollout_corr/chi2_seq:np.float64(0.3160702483728528) - actor/pg_loss:np.float64(0.02642797108637751) - 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(0.12867224216461182) - perf/mfu/actor:np.float64(0.04323287964250154) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.61327743530273) - actor/lr:np.float64(1e-05) - training/global_step:43 - training/epoch:0 - critic/score/mean:0.703125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.703125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0018240667413920164 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0018240667413920164 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:284.8203125 - response_length/max:872.0 - response_length/min:134.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:284.8203125 - response_length_non_aborted/max:872.0 - response_length_non_aborted/min:134.0 - 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:7.39116221666336e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8455637786537409) - timing_s/agent_loop/generate_sequences/max:np.float64(7.242258792743087) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.5451021448534448) - timing_s/agent_loop/tool_calls/min:
|
|
step:44 - global_seqlen/min:16866 - global_seqlen/max:22547 - global_seqlen/minmax_diff:5681 - global_seqlen/balanced_min:19246 - global_seqlen/balanced_max:19250 - global_seqlen/mean:19248.375 - actor/entropy:0.42190542817115784 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2310553789138794 - training/rollout_probs_diff_mean:0.0038836172316223383 - training/rollout_probs_diff_std:0.00841892696917057 - training/rollout_actor_probs_pearson_corr:0.9994995594024658 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.85546875 - 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:0.85546875 - rollout_corr/rollout_is_mean:1.000082015991211 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.3472186765284277e-05 - rollout_corr/rollout_is_max:1.839381456375122 - rollout_corr/rollout_is_min:0.3176516592502594 - rollout_corr/rollout_is_std:0.02960249036550522 - rollout_corr/rollout_is_eff_sample_size:0.999124697756313 - rollout_corr/rollout_is_seq_mean:1.0000885725021362 - rollout_corr/rollout_is_seq_std:0.0018956754356622696 - rollout_corr/rollout_is_seq_max:1.0079100131988525 - rollout_corr/rollout_is_seq_min:0.9926266670227051 - rollout_corr/rollout_is_seq_max_deviation:0.007910013198852539 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5580365266650915) - rollout_corr/training_log_ppl:np.float64(0.43635796845774166) - rollout_corr/kl:np.float64(0.000368623170373894) - rollout_corr/k3_kl:np.float64(0.00045717331602190825) - rollout_corr/rollout_ppl:np.float64(1.5574525520205498) - rollout_corr/rollout_log_ppl:np.float64(0.43598934210604057) - rollout_corr/log_ppl_diff:np.float64(0.0003686263517010957) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014585884346161038) - rollout_corr/log_ppl_diff_max:np.float64(0.011438578367233276) - rollout_corr/log_ppl_diff_min:np.float64(-0.005643367767333984) - rollout_corr/ppl_ratio:np.float64(1.0003705746494234) - rollout_corr/chi2_token:np.float64(0.0010971922893077135) - rollout_corr/chi2_seq:np.float64(0.2806218769401312) - actor/pg_loss:np.float64(0.025179245734761935) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.14453125) - actor/grad_norm:np.float64(0.15673339366912842) - perf/mfu/actor:np.float64(0.04481737039083471) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.60191345214844) - actor/lr:np.float64(1e-05) - training/global_step:44 - 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:-0.0004243738658260554 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0004243738658260554 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:289.94921875 - response_length/max:872.0 - response_length/min:135.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:289.94921875 - response_length_non_aborted/max:872.0 - response_length_non_aborted/min:135.0 - 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:0.00010008923709392548 - timing_s/agent_loop/generate_sequences/min:np.float64(1.7268172409385443) - timing_s/agent_loop/generate_sequences/max:np.float64(7.175586240366101) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.572714293615718) - timing_
|
|
step:45 - global_seqlen/min:16682 - global_seqlen/max:25314 - global_seqlen/minmax_diff:8632 - global_seqlen/balanced_min:20213 - global_seqlen/balanced_max:20217 - global_seqlen/mean:20214.875 - actor/entropy:0.39611583948135376 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4400330185890198 - training/rollout_probs_diff_mean:0.003430995624512434 - training/rollout_probs_diff_std:0.007905502803623676 - training/rollout_actor_probs_pearson_corr:0.9995728731155396 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - 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:0.8359375 - rollout_corr/rollout_is_mean:1.0001569986343384 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.2321492249611765e-05 - rollout_corr/rollout_is_max:1.979709267616272 - rollout_corr/rollout_is_min:0.46178221702575684 - rollout_corr/rollout_is_std:0.027482904493808746 - rollout_corr/rollout_is_eff_sample_size:0.9992454980630003 - rollout_corr/rollout_is_seq_mean:1.0001763105392456 - rollout_corr/rollout_is_seq_std:0.0018226178362965584 - rollout_corr/rollout_is_seq_max:1.0077135562896729 - rollout_corr/rollout_is_seq_min:0.9930106401443481 - rollout_corr/rollout_is_seq_max_deviation:0.0077135562896728516 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5417101834900677) - rollout_corr/training_log_ppl:np.float64(0.4236733771394938) - rollout_corr/kl:np.float64(0.00022777361688497422) - rollout_corr/k3_kl:np.float64(0.0004041654358957203) - rollout_corr/rollout_ppl:np.float64(1.5413539111614227) - rollout_corr/rollout_log_ppl:np.float64(0.4234455991536379) - rollout_corr/log_ppl_diff:np.float64(0.000227777985855937) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013722628355026245) - rollout_corr/log_ppl_diff_max:np.float64(0.007453024387359619) - rollout_corr/log_ppl_diff_min:np.float64(-0.005736231803894043) - rollout_corr/ppl_ratio:np.float64(1.0002294208388776) - rollout_corr/chi2_token:np.float64(0.001172445947304368) - rollout_corr/chi2_seq:np.float64(0.3459507490042597) - actor/pg_loss:np.float64(0.0231021549734578) - 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(0.1241765171289444) - perf/mfu/actor:np.float64(0.04679945412828388) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.61876678466797) - actor/lr:np.float64(1e-05) - training/global_step:45 - 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:-0.005883512552827597 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.005883512552827597 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:317.02734375 - response_length/max:789.0 - response_length/min:85.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:317.02734375 - response_length_non_aborted/max:789.0 - response_length_non_aborted/min:85.0 - 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:3.9597973227500916e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.188296465203166) - timing_s/agent_loop/generate_sequences/max:np.float64(6.90229476056993) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.7401923308279947) - timing_s/ag
|
|
step:46 - global_seqlen/min:16227 - global_seqlen/max:21161 - global_seqlen/minmax_diff:4934 - global_seqlen/balanced_min:18528 - global_seqlen/balanced_max:18534 - global_seqlen/mean:18532.625 - actor/entropy:0.4359818994998932 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.17366677522659302 - training/rollout_probs_diff_mean:0.003783823922276497 - training/rollout_probs_diff_std:0.008063805289566517 - training/rollout_actor_probs_pearson_corr:0.9995725750923157 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.79296875 - 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:0.79296875 - rollout_corr/rollout_is_mean:0.9999687671661377 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5417418479919434 - rollout_corr/rollout_is_min:0.6651032567024231 - rollout_corr/rollout_is_std:0.02856549620628357 - rollout_corr/rollout_is_eff_sample_size:0.9991845586909296 - rollout_corr/rollout_is_seq_mean:1.0000532865524292 - rollout_corr/rollout_is_seq_std:0.0018359309760853648 - rollout_corr/rollout_is_seq_max:1.0064587593078613 - rollout_corr/rollout_is_seq_min:0.9931100606918335 - rollout_corr/rollout_is_seq_max_deviation:0.006889939308166504 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6022519608959556) - rollout_corr/training_log_ppl:np.float64(0.4606198475812562) - rollout_corr/kl:np.float64(0.0003739741666830554) - rollout_corr/k3_kl:np.float64(0.0004272379021017514) - rollout_corr/rollout_ppl:np.float64(1.6016698996536434) - rollout_corr/rollout_log_ppl:np.float64(0.4602458711597137) - rollout_corr/log_ppl_diff:np.float64(0.0003739764215424657) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014247739454731345) - rollout_corr/log_ppl_diff_max:np.float64(0.007501423358917236) - rollout_corr/log_ppl_diff_min:np.float64(-0.005899012088775635) - rollout_corr/ppl_ratio:np.float64(1.0003757197409868) - rollout_corr/chi2_token:np.float64(0.0009671284351497889) - rollout_corr/chi2_seq:np.float64(0.21831048862077296) - actor/pg_loss:np.float64(0.02200502128471271) - 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(0.13160297274589539) - perf/mfu/actor:np.float64(0.04267240014757405) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.5970687866211) - actor/lr:np.float64(1e-05) - training/global_step:46 - 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:-0.0003100381582044065 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0003100381582044065 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:303.95703125 - response_length/max:909.0 - response_length/min:103.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:303.95703125 - response_length_non_aborted/max:909.0 - response_length_non_aborted/min:103.0 - response_length_non_aborted/clip_ratio:0.0 - 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:9.754858911037445e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.4063380844891071) - timing_s/agent_loop/generate_sequences/max:np.float64(7.655340066179633) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.5860595810008817) - timing_s/agent_loop/tool_calls/
|
|
step:47 - global_seqlen/min:14963 - global_seqlen/max:22375 - global_seqlen/minmax_diff:7412 - global_seqlen/balanced_min:18263 - global_seqlen/balanced_max:18268 - global_seqlen/mean:18265.75 - actor/entropy:0.418122798204422 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.26676541566848755 - training/rollout_probs_diff_mean:0.003802663879469037 - training/rollout_probs_diff_std:0.00833720900118351 - training/rollout_actor_probs_pearson_corr:0.9995467066764832 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.86328125 - 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:0.86328125 - rollout_corr/rollout_is_mean:1.0000063180923462 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.9659818410873413 - rollout_corr/rollout_is_min:0.5187408924102783 - rollout_corr/rollout_is_std:0.029367996379733086 - rollout_corr/rollout_is_eff_sample_size:0.9991383830234061 - rollout_corr/rollout_is_seq_mean:1.0000537633895874 - rollout_corr/rollout_is_seq_std:0.0018420947017148137 - rollout_corr/rollout_is_seq_max:1.0052917003631592 - rollout_corr/rollout_is_seq_min:0.9952124953269958 - rollout_corr/rollout_is_seq_max_deviation:0.00529170036315918 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5636549689806998) - rollout_corr/training_log_ppl:np.float64(0.4363196868216619) - rollout_corr/kl:np.float64(0.0003868262038793091) - rollout_corr/k3_kl:np.float64(0.0004406081577030818) - rollout_corr/rollout_ppl:np.float64(1.5630644368939102) - rollout_corr/rollout_log_ppl:np.float64(0.43593285998213105) - rollout_corr/log_ppl_diff:np.float64(0.0003868268395308405) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014885150885675102) - rollout_corr/log_ppl_diff_max:np.float64(0.005997180938720703) - rollout_corr/log_ppl_diff_min:np.float64(-0.004631847143173218) - rollout_corr/ppl_ratio:np.float64(1.0003886055201292) - rollout_corr/chi2_token:np.float64(0.0009930671658366919) - rollout_corr/chi2_seq:np.float64(0.2101473300717771) - actor/pg_loss:np.float64(0.02535204555169912) - 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(0.1195836216211319) - perf/mfu/actor:np.float64(0.04261431840904702) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.52521896362305) - actor/lr:np.float64(1e-05) - training/global_step:47 - 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:-0.003205669578164816 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.003205669578164816 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:277.5234375 - response_length/max:1089.0 - response_length/min:124.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:277.5234375 - response_length_non_aborted/max:1089.0 - response_length_non_aborted/min:124.0 - 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:3.761984407901764e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.706325326114893) - timing_s/agent_loop/generate_sequences/max:np.float64(8.50525888055563) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.358023824002885) - timing_s/agent_loop/tool_calls
|
|
step:48 - global_seqlen/min:15479 - global_seqlen/max:20194 - global_seqlen/minmax_diff:4715 - global_seqlen/balanced_min:17529 - global_seqlen/balanced_max:17580 - global_seqlen/mean:17536.625 - actor/entropy:0.4028206765651703 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12434142827987671 - training/rollout_probs_diff_mean:0.003673039609566331 - training/rollout_probs_diff_std:0.007926157675683498 - training/rollout_actor_probs_pearson_corr:0.9995599985122681 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.92578125 - 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:0.92578125 - rollout_corr/rollout_is_mean:1.000057339668274 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5526751279830933 - rollout_corr/rollout_is_min:0.6994926929473877 - rollout_corr/rollout_is_std:0.027936771512031555 - rollout_corr/rollout_is_eff_sample_size:0.999220264452564 - rollout_corr/rollout_is_seq_mean:1.0000791549682617 - rollout_corr/rollout_is_seq_std:0.001777162542566657 - rollout_corr/rollout_is_seq_max:1.0057016611099243 - rollout_corr/rollout_is_seq_min:0.9951032996177673 - rollout_corr/rollout_is_seq_max_deviation:0.005701661109924316 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5320349587127566) - rollout_corr/training_log_ppl:np.float64(0.4171645875030663) - rollout_corr/kl:np.float64(0.000330831693624134) - rollout_corr/k3_kl:np.float64(0.00040998962001026484) - rollout_corr/rollout_ppl:np.float64(1.5315529028885067) - rollout_corr/rollout_log_ppl:np.float64(0.4168337558803614) - rollout_corr/log_ppl_diff:np.float64(0.00033083162270486355) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013908130349591374) - rollout_corr/log_ppl_diff_max:np.float64(0.005478084087371826) - rollout_corr/log_ppl_diff_min:np.float64(-0.005206495523452759) - rollout_corr/ppl_ratio:np.float64(1.000332427676767) - rollout_corr/chi2_token:np.float64(0.0009836687240749598) - rollout_corr/chi2_seq:np.float64(0.2130511358845979) - actor/pg_loss:np.float64(0.027172330894245533) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.07421875) - actor/grad_norm:np.float64(0.13036848604679108) - perf/mfu/actor:np.float64(0.040540908797480193) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.65386199951172) - actor/lr:np.float64(1e-05) - training/global_step:48 - 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:-0.005438016261905432 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.005438016261905432 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:282.30078125 - response_length/max:1097.0 - response_length/min:119.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:282.30078125 - response_length_non_aborted/max:1097.0 - response_length_non_aborted/min:119.0 - response_length_non_aborted/clip_ratio:0.0 - 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:0.0001454222947359085 - timing_s/agent_loop/generate_sequences/min:np.float64(1.2647006548941135) - timing_s/agent_loop/generate_sequences/max:np.float64(8.229641575366259) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.1935032372784917) - timing_s/agent_loop/tool_calls/min:
|
|
step:49 - global_seqlen/min:17470 - global_seqlen/max:19477 - global_seqlen/minmax_diff:2007 - global_seqlen/balanced_min:18440 - global_seqlen/balanced_max:18446 - global_seqlen/mean:18444.75 - actor/entropy:0.42139461636543274 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.13564960658550262 - training/rollout_probs_diff_mean:0.0037786359898746014 - training/rollout_probs_diff_std:0.008087078109383583 - training/rollout_actor_probs_pearson_corr:0.9995549321174622 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.9296875 - self_distillation/correct_sample_fraction:0.78515625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.9296875 - rollout_corr/rollout_is_mean:0.9999513626098633 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.3126460544299334e-05 - rollout_corr/rollout_is_max:1.4538602828979492 - rollout_corr/rollout_is_min:0.24729026854038239 - rollout_corr/rollout_is_std:0.028372880071401596 - rollout_corr/rollout_is_eff_sample_size:0.999195627204119 - rollout_corr/rollout_is_seq_mean:0.99993896484375 - rollout_corr/rollout_is_seq_std:0.0018597800517454743 - rollout_corr/rollout_is_seq_max:1.006990671157837 - rollout_corr/rollout_is_seq_min:0.994023323059082 - rollout_corr/rollout_is_seq_max_deviation:0.006990671157836914 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5495123639702797) - rollout_corr/training_log_ppl:np.float64(0.4296430809190497) - rollout_corr/kl:np.float64(0.0004818617892503596) - rollout_corr/k3_kl:np.float64(0.0004208038327533359) - rollout_corr/rollout_ppl:np.float64(1.5487493169493973) - rollout_corr/rollout_log_ppl:np.float64(0.4291612175293267) - rollout_corr/log_ppl_diff:np.float64(0.0004818633897230029) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014479222008958459) - rollout_corr/log_ppl_diff_max:np.float64(0.008307039737701416) - rollout_corr/log_ppl_diff_min:np.float64(-0.006430685520172119) - rollout_corr/ppl_ratio:np.float64(1.000483766430989) - rollout_corr/chi2_token:np.float64(0.0007128908764570951) - rollout_corr/chi2_seq:np.float64(0.35852123564109206) - actor/pg_loss:np.float64(0.02792465963284485) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.0703125) - actor/grad_norm:np.float64(0.13089485466480255) - perf/mfu/actor:np.float64(0.042404865190328433) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.6152114868164) - actor/lr:np.float64(1e-05) - training/global_step:49 - training/epoch:0 - critic/score/mean:0.78515625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.78515625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.005586949642747641 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.005586949642747641 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:297.5859375 - response_length/max:860.0 - response_length/min:108.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:297.5859375 - response_length_non_aborted/max:860.0 - response_length_non_aborted/min:108.0 - 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:6.860867142677307e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.4788524433970451) - timing_s/agent_loop/generate_sequences/max:np.float64(7.168281147256494) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.6172412733139936) - timing_s/agent
|
|
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 target product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" contains a chlorinated aromatic ring (with Cl), a trifluoromethyl group (C(F)(F)F), and a nitrogen-containing heterocyclic ring. The correct reactant must contain all these key structural features to be involved in the synthesis of this molecule.
|
|
|
|
- **Option A**: Contains the chlorinated aromatic ring, the trifluoromethyl group (C(F)(F)F), and the nitrogen-containing ring, matching the structure of the target product.
|
|
- **Option B**: Includes a different arrangement of functional groups, specifically a chlorinated ring and a different nitrogen-containing ring, which does not match the target structure.
|
|
- **Option C**: Has a hydroxyl group and a different arrangement of the fluorinated chain, which deviates from the target structure.
|
|
- **Option D**: Contains a different carbon chain and fluorinated group structure, making it structurally incompatible with the target product.
|
|
|
|
Thus, **Option A** is the only reactant that matches the required structural components of the target molecule.
|
|
</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-Qwen3-8B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_50
|
|
step:50 - global_seqlen/min:15764 - global_seqlen/max:23455 - global_seqlen/minmax_diff:7691 - global_seqlen/balanced_min:18459 - global_seqlen/balanced_max:18460 - global_seqlen/mean:18459.5 - actor/entropy:0.3588300347328186 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2412884533405304 - training/rollout_probs_diff_mean:0.003348836675286293 - training/rollout_probs_diff_std:0.00788651593029499 - training/rollout_actor_probs_pearson_corr:0.9995613694190979 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.890625 - 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:0.890625 - rollout_corr/rollout_is_mean:0.9998207688331604 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6725608110427856 - rollout_corr/rollout_is_min:0.6446537971496582 - rollout_corr/rollout_is_std:0.026825489476323128 - rollout_corr/rollout_is_eff_sample_size:0.9992804939528137 - rollout_corr/rollout_is_seq_mean:0.9997795224189758 - rollout_corr/rollout_is_seq_std:0.0017754227155819535 - rollout_corr/rollout_is_seq_max:1.0042121410369873 - rollout_corr/rollout_is_seq_min:0.9940100908279419 - rollout_corr/rollout_is_seq_max_deviation:0.0059899091720581055 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4925650302320719) - rollout_corr/training_log_ppl:np.float64(0.39076297875726596) - rollout_corr/kl:np.float64(0.0006134095791656691) - rollout_corr/k3_kl:np.float64(0.00039299291577776785) - rollout_corr/rollout_ppl:np.float64(1.491663983091712) - rollout_corr/rollout_log_ppl:np.float64(0.3901495693135075) - rollout_corr/log_ppl_diff:np.float64(0.0006134094437584281) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014219823642633855) - rollout_corr/log_ppl_diff_max:np.float64(0.0065149664878845215) - rollout_corr/log_ppl_diff_min:np.float64(-0.0037950873374938965) - rollout_corr/ppl_ratio:np.float64(1.0006151460111141) - rollout_corr/chi2_token:np.float64(0.000349690904840827) - rollout_corr/chi2_seq:np.float64(0.08765777293592691) - actor/pg_loss:np.float64(0.024352237644052366) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.109375) - actor/grad_norm:np.float64(0.12670068442821503) - perf/mfu/actor:np.float64(0.04226164855603727) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.64237308502197) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7196428571428571) - val-aux/sciknoweval/reward/std@16:np.float64(0.1265026225197639) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7699952380952382) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.11288458303810753) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6700619047619047) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.09997360562985441) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7195761904761905) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.1264928926210629) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.8150000000000001) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.09225361600736358) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6324428571428571) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.07170441999941289) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.721757142857143) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.0998404549480569) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.8530857142857142) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.06892258059718083) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6039714285714286) - val-aux/sciknoweval/reward/worst@8/
|
|
step:51 - global_seqlen/min:17567 - global_seqlen/max:21110 - global_seqlen/minmax_diff:3543 - global_seqlen/balanced_min:19483 - global_seqlen/balanced_max:19488 - global_seqlen/mean:19486.625 - actor/entropy:0.44690820574760437 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.25539886951446533 - training/rollout_probs_diff_mean:0.0039621260948479176 - training/rollout_probs_diff_std:0.008190932683646679 - training/rollout_actor_probs_pearson_corr:0.9995589852333069 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8046875 - 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:0.8046875 - rollout_corr/rollout_is_mean:0.9998981952667236 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.837836742401123 - rollout_corr/rollout_is_min:0.604153573513031 - rollout_corr/rollout_is_std:0.029266340658068657 - rollout_corr/rollout_is_eff_sample_size:0.999143976240487 - rollout_corr/rollout_is_seq_mean:0.9998791813850403 - rollout_corr/rollout_is_seq_std:0.0018273589666932821 - rollout_corr/rollout_is_seq_max:1.0075286626815796 - rollout_corr/rollout_is_seq_min:0.9949983358383179 - rollout_corr/rollout_is_seq_max_deviation:0.00752866268157959 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5894114742986858) - rollout_corr/training_log_ppl:np.float64(0.45557393025956117) - rollout_corr/kl:np.float64(0.0005590946644034034) - rollout_corr/k3_kl:np.float64(0.00043821071545835366) - rollout_corr/rollout_ppl:np.float64(1.5884660063311458) - rollout_corr/rollout_log_ppl:np.float64(0.4550148332200479) - rollout_corr/log_ppl_diff:np.float64(0.0005590970395132899) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014663320034742355) - rollout_corr/log_ppl_diff_max:np.float64(0.0055901408195495605) - rollout_corr/log_ppl_diff_min:np.float64(-0.006415367126464844) - rollout_corr/ppl_ratio:np.float64(1.0005609169602394) - rollout_corr/chi2_token:np.float64(0.0006425180472433567) - rollout_corr/chi2_seq:np.float64(0.2239590159151703) - actor/pg_loss:np.float64(0.024827675932101556) - 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(0.1343592256307602) - perf/mfu/actor:np.float64(0.04425238870358594) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.70580291748047) - actor/lr:np.float64(1e-05) - training/global_step:51 - 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:-0.001255188719369471 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.001255188719369471 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:296.42578125 - response_length/max:814.0 - response_length/min:131.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:296.42578125 - response_length_non_aborted/max:814.0 - response_length_non_aborted/min:131.0 - 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:0.00012967176735401154 - timing_s/agent_loop/generate_sequences/min:np.float64(1.269055314362049) - timing_s/agent_loop/generate_sequences/max:np.float64(6.624901965260506) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.364351192700269) - timing_s/agent_loop/tool_
|
|
step:52 - global_seqlen/min:15913 - global_seqlen/max:23614 - global_seqlen/minmax_diff:7701 - global_seqlen/balanced_min:18831 - global_seqlen/balanced_max:18833 - global_seqlen/mean:18832.5 - actor/entropy:0.398704469203949 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.11813041567802429 - training/rollout_probs_diff_mean:0.003580359974876046 - training/rollout_probs_diff_std:0.007930848747491837 - training/rollout_actor_probs_pearson_corr:0.9995623230934143 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83984375 - self_distillation/correct_sample_fraction:0.64453125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83984375 - rollout_corr/rollout_is_mean:0.9999911785125732 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.3796827793121338 - rollout_corr/rollout_is_min:0.6156680583953857 - rollout_corr/rollout_is_std:0.02782774716615677 - rollout_corr/rollout_is_eff_sample_size:0.9992260966360754 - rollout_corr/rollout_is_seq_mean:0.9999725222587585 - rollout_corr/rollout_is_seq_std:0.0017707835650071502 - rollout_corr/rollout_is_seq_max:1.0072441101074219 - rollout_corr/rollout_is_seq_min:0.9953335523605347 - rollout_corr/rollout_is_seq_max_deviation:0.007244110107421875 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.532066170591861) - rollout_corr/training_log_ppl:np.float64(0.4156618236738723) - rollout_corr/kl:np.float64(0.0004319135960955123) - rollout_corr/k3_kl:np.float64(0.00040442465362389157) - rollout_corr/rollout_ppl:np.float64(1.5314252390526235) - rollout_corr/rollout_log_ppl:np.float64(0.4152299124398269) - rollout_corr/log_ppl_diff:np.float64(0.0004319112340454012) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014288426900748163) - rollout_corr/log_ppl_diff_max:np.float64(0.005120009183883667) - rollout_corr/log_ppl_diff_min:np.float64(-0.006492435932159424) - rollout_corr/ppl_ratio:np.float64(1.0004335667472333) - rollout_corr/chi2_token:np.float64(0.0007549251895397902) - rollout_corr/chi2_seq:np.float64(0.3536420976743102) - actor/pg_loss:np.float64(0.025048665442227502) - 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(0.11144942790269852) - perf/mfu/actor:np.float64(0.043339508774113224) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(197.68711471557617) - actor/lr:np.float64(1e-05) - training/global_step:52 - training/epoch:0 - critic/score/mean:0.64453125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.64453125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0002924204745795578 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0002924204745795578 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:325.609375 - response_length/max:756.0 - response_length/min:129.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:325.609375 - response_length_non_aborted/max:756.0 - response_length_non_aborted/min:129.0 - 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:4.620291292667389e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.5304780043661594) - timing_s/agent_loop/generate_sequences/max:np.float64(6.849703526124358) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.7845842141396133) - timing_s/agent_loop/tool_ca
|
|
step:53 - global_seqlen/min:17692 - global_seqlen/max:22337 - global_seqlen/minmax_diff:4645 - global_seqlen/balanced_min:19429 - global_seqlen/balanced_max:19434 - global_seqlen/mean:19432.25 - actor/entropy:0.4355466663837433 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.11699336767196655 - training/rollout_probs_diff_mean:0.0037828637287020683 - training/rollout_probs_diff_std:0.007988695986568928 - training/rollout_actor_probs_pearson_corr:0.9995979070663452 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.8984375 - 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:0.8984375 - rollout_corr/rollout_is_mean:0.9999352693557739 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6097451448440552 - rollout_corr/rollout_is_min:0.6647977828979492 - rollout_corr/rollout_is_std:0.02837498113512993 - rollout_corr/rollout_is_eff_sample_size:0.9991953891689972 - rollout_corr/rollout_is_seq_mean:0.9999077916145325 - rollout_corr/rollout_is_seq_std:0.0018596879672259092 - rollout_corr/rollout_is_seq_max:1.0048993825912476 - rollout_corr/rollout_is_seq_min:0.9944272637367249 - rollout_corr/rollout_is_seq_max_deviation:0.0055727362632751465 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5918304533697665) - rollout_corr/training_log_ppl:np.float64(0.4555920782731846) - rollout_corr/kl:np.float64(0.0005068698899606261) - rollout_corr/k3_kl:np.float64(0.00041460993395503465) - rollout_corr/rollout_ppl:np.float64(1.5910366787575185) - rollout_corr/rollout_log_ppl:np.float64(0.4550852081156336) - rollout_corr/log_ppl_diff:np.float64(0.0005068701575510204) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014834986650384963) - rollout_corr/log_ppl_diff_max:np.float64(0.005933403968811035) - rollout_corr/log_ppl_diff_min:np.float64(-0.0042765140533447266) - rollout_corr/ppl_ratio:np.float64(1.0005087042227387) - rollout_corr/chi2_token:np.float64(0.0006478838622570038) - rollout_corr/chi2_seq:np.float64(0.2598018078133464) - actor/pg_loss:np.float64(0.025513055674309726) - 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(0.11645081639289856) - perf/mfu/actor:np.float64(0.044765792016770105) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.87960815429688) - actor/lr:np.float64(1e-05) - training/global_step:53 - training/epoch:0 - critic/score/mean:0.7109375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7109375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0034354086965322495 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0034354086965322495 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:311.6953125 - response_length/max:854.0 - response_length/min:102.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:311.6953125 - response_length_non_aborted/max:854.0 - response_length_non_aborted/min:102.0 - 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:9.389780461788177e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.365078415721655) - timing_s/agent_loop/generate_sequences/max:np.float64(7.309070995077491) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.717466606576636) - timing_s/agent_loop/tool_c
|
|
step:54 - global_seqlen/min:16386 - global_seqlen/max:21448 - global_seqlen/minmax_diff:5062 - global_seqlen/balanced_min:19173 - global_seqlen/balanced_max:19179 - global_seqlen/mean:19176.375 - actor/entropy:0.3905041813850403 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12245392799377441 - training/rollout_probs_diff_mean:0.003457350190728903 - training/rollout_probs_diff_std:0.007704423274844885 - training/rollout_actor_probs_pearson_corr:0.9995774626731873 - 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:0.9999135136604309 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.3528367280960083 - rollout_corr/rollout_is_min:0.5706366896629333 - rollout_corr/rollout_is_std:0.026921970769762993 - rollout_corr/rollout_is_eff_sample_size:0.999275494389906 - rollout_corr/rollout_is_seq_mean:0.9998381733894348 - rollout_corr/rollout_is_seq_std:0.0017586079193279147 - rollout_corr/rollout_is_seq_max:1.005534291267395 - rollout_corr/rollout_is_seq_min:0.9946604371070862 - rollout_corr/rollout_is_seq_max_deviation:0.0055342912673950195 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5337817729450762) - rollout_corr/training_log_ppl:np.float64(0.4180254357197555) - rollout_corr/kl:np.float64(0.0005513709736000294) - rollout_corr/k3_kl:np.float64(0.00038957832333608167) - rollout_corr/rollout_ppl:np.float64(1.5329339550808072) - rollout_corr/rollout_log_ppl:np.float64(0.4174740598682547) - rollout_corr/log_ppl_diff:np.float64(0.0005513758515007794) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013731399085372686) - rollout_corr/log_ppl_diff_max:np.float64(0.006084084510803223) - rollout_corr/log_ppl_diff_min:np.float64(-0.0051438212394714355) - rollout_corr/ppl_ratio:np.float64(1.000553070800379) - rollout_corr/chi2_token:np.float64(0.0004567198920994997) - rollout_corr/chi2_seq:np.float64(0.13874381058849394) - actor/pg_loss:np.float64(0.021754106717708055) - 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(0.11083228141069412) - perf/mfu/actor:np.float64(0.04440506089904515) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.63725662231445) - actor/lr:np.float64(1e-05) - training/global_step:54 - 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:-0.0006808100733906031 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0006808100733906031 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:328.48046875 - response_length/max:930.0 - response_length/min:136.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:328.48046875 - response_length_non_aborted/max:930.0 - response_length_non_aborted/min:136.0 - response_length_non_aborted/clip_ratio:0.0 - 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:9.906105697154999e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.845677923411131) - timing_s/agent_loop/generate_sequences/max:np.float64(7.780272418633103) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.857449359325983) - timing_s/agent_loop/tool_call
|
|
step:55 - global_seqlen/min:16750 - global_seqlen/max:20097 - global_seqlen/minmax_diff:3347 - global_seqlen/balanced_min:18977 - global_seqlen/balanced_max:18981 - global_seqlen/mean:18979.375 - actor/entropy:0.44882702827453613 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1382938027381897 - training/rollout_probs_diff_mean:0.0038533126935362816 - training/rollout_probs_diff_std:0.00807112269103527 - training/rollout_actor_probs_pearson_corr:0.9995792508125305 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.89453125 - self_distillation/correct_sample_fraction:0.71875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.89453125 - rollout_corr/rollout_is_mean:1.0000598430633545 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6400389671325684 - rollout_corr/rollout_is_min:0.6845523118972778 - rollout_corr/rollout_is_std:0.0288705974817276 - rollout_corr/rollout_is_eff_sample_size:0.9991675398426103 - rollout_corr/rollout_is_seq_mean:1.000090479850769 - rollout_corr/rollout_is_seq_std:0.0019493589643388987 - rollout_corr/rollout_is_seq_max:1.0075825452804565 - rollout_corr/rollout_is_seq_min:0.9932894706726074 - rollout_corr/rollout_is_seq_max_deviation:0.007582545280456543 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.6136538074351847) - rollout_corr/training_log_ppl:np.float64(0.46838534937705845) - rollout_corr/kl:np.float64(0.00034479378155438667) - rollout_corr/k3_kl:np.float64(0.0004352847810480398) - rollout_corr/rollout_ppl:np.float64(1.613085353281349) - rollout_corr/rollout_log_ppl:np.float64(0.46804055041866377) - rollout_corr/log_ppl_diff:np.float64(0.00034479895839467645) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015006595640443265) - rollout_corr/log_ppl_diff_max:np.float64(0.00708848237991333) - rollout_corr/log_ppl_diff_min:np.float64(-0.006899654865264893) - rollout_corr/ppl_ratio:np.float64(1.0003467292990535) - rollout_corr/chi2_token:np.float64(0.0010594082996249199) - rollout_corr/chi2_seq:np.float64(0.3098592855967581) - actor/pg_loss:np.float64(0.02823107153562887) - 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(0.1410229504108429) - perf/mfu/actor:np.float64(0.044401889033531566) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.61139297485352) - actor/lr:np.float64(1e-05) - training/global_step:55 - training/epoch:0 - critic/score/mean:0.71875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.71875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.003481856547296047 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.003481856547296047 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:288.60546875 - response_length/max:823.0 - response_length/min:132.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:288.60546875 - response_length_non_aborted/max:823.0 - response_length_non_aborted/min:132.0 - 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:3.1247735023498535e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.7431130316108465) - timing_s/agent_loop/generate_sequences/max:np.float64(6.865928471088409) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.4372487175205606) - timing_s/agent_loop/tool_calls/min
|
|
step:56 - global_seqlen/min:17097 - global_seqlen/max:22170 - global_seqlen/minmax_diff:5073 - global_seqlen/balanced_min:18738 - global_seqlen/balanced_max:18961 - global_seqlen/mean:18769.5 - actor/entropy:0.41118669509887695 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1757318377494812 - training/rollout_probs_diff_mean:0.003505242522805929 - training/rollout_probs_diff_std:0.007644045632332563 - training/rollout_actor_probs_pearson_corr:0.999602198600769 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.86328125 - 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:0.86328125 - rollout_corr/rollout_is_mean:0.9999428391456604 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.403804063796997 - rollout_corr/rollout_is_min:0.711456835269928 - rollout_corr/rollout_is_std:0.027056686580181122 - rollout_corr/rollout_is_eff_sample_size:0.999268352243962 - rollout_corr/rollout_is_seq_mean:0.9999257922172546 - rollout_corr/rollout_is_seq_std:0.0017106898594647646 - rollout_corr/rollout_is_seq_max:1.0056562423706055 - rollout_corr/rollout_is_seq_min:0.994813084602356 - rollout_corr/rollout_is_seq_max_deviation:0.005656242370605469 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5683106994256377) - rollout_corr/training_log_ppl:np.float64(0.438737221295014) - rollout_corr/kl:np.float64(0.0004667321902278587) - rollout_corr/k3_kl:np.float64(0.0003925500751904565) - rollout_corr/rollout_ppl:np.float64(1.5675687724724412) - rollout_corr/rollout_log_ppl:np.float64(0.4382704859599471) - rollout_corr/log_ppl_diff:np.float64(0.00046673533506691456) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013582040555775166) - rollout_corr/log_ppl_diff_max:np.float64(0.005755186080932617) - rollout_corr/log_ppl_diff_min:np.float64(-0.005272954702377319) - rollout_corr/ppl_ratio:np.float64(1.000468286452815) - rollout_corr/chi2_token:np.float64(0.000639752484858036) - rollout_corr/chi2_seq:np.float64(0.19365815469063818) - actor/pg_loss:np.float64(0.02293320205717464) - 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(0.11079184710979462) - perf/mfu/actor:np.float64(0.043579223740994176) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.71324157714844) - actor/lr:np.float64(1e-05) - training/global_step:56 - training/epoch:0 - critic/score/mean:0.640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.017775535583496094 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.017775535583496094 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:321.390625 - response_length/max:1416.0 - response_length/min:108.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:321.390625 - response_length_non_aborted/max:1416.0 - response_length_non_aborted/min:108.0 - 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:9.857676923274994e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.2711333073675632) - timing_s/agent_loop/generate_sequences/max:np.float64(10.542331211268902) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.765388753527077) - timing_s/agent_loop/tool_calls/min:np.
|
|
step:57 - global_seqlen/min:15752 - global_seqlen/max:20886 - global_seqlen/minmax_diff:5134 - global_seqlen/balanced_min:18638 - global_seqlen/balanced_max:18641 - global_seqlen/mean:18640.125 - actor/entropy:0.42558160424232483 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.14700806140899658 - training/rollout_probs_diff_mean:0.0036993061657994986 - training/rollout_probs_diff_std:0.007875271141529083 - training/rollout_actor_probs_pearson_corr:0.9995828866958618 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.92578125 - 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:0.92578125 - rollout_corr/rollout_is_mean:0.9999958872795105 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.3580538034439087 - rollout_corr/rollout_is_min:0.6847708821296692 - rollout_corr/rollout_is_std:0.027898341417312622 - rollout_corr/rollout_is_eff_sample_size:0.9992224068794121 - rollout_corr/rollout_is_seq_mean:1.0000523328781128 - rollout_corr/rollout_is_seq_std:0.0018940261797979474 - rollout_corr/rollout_is_seq_max:1.0050809383392334 - rollout_corr/rollout_is_seq_min:0.9951463341712952 - rollout_corr/rollout_is_seq_max_deviation:0.0050809383392333984 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5581915099173784) - rollout_corr/training_log_ppl:np.float64(0.4364832380088046) - rollout_corr/kl:np.float64(0.00035268154108791805) - rollout_corr/k3_kl:np.float64(0.00040503066281871725) - rollout_corr/rollout_ppl:np.float64(1.5576647697016597) - rollout_corr/rollout_log_ppl:np.float64(0.436130553658586) - rollout_corr/log_ppl_diff:np.float64(0.0003526843502186239) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015374367940239608) - rollout_corr/log_ppl_diff_max:np.float64(0.005321085453033447) - rollout_corr/log_ppl_diff_min:np.float64(-0.004596710205078125) - rollout_corr/ppl_ratio:np.float64(1.0003545337822288) - rollout_corr/chi2_token:np.float64(0.0009166451636701822) - rollout_corr/chi2_seq:np.float64(0.2423508686479181) - actor/pg_loss:np.float64(0.026474346062968834) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.07421875) - actor/grad_norm:np.float64(0.13544140756130219) - perf/mfu/actor:np.float64(0.04281035014313875) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.73678970336914) - actor/lr:np.float64(1e-05) - training/global_step:57 - training/epoch:0 - 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:-0.0006257842760533094 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0006257842760533094 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:295.72265625 - response_length/max:790.0 - response_length/min:132.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:295.72265625 - response_length_non_aborted/max:790.0 - response_length_non_aborted/min:132.0 - 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:0.00014242157340049744 - timing_s/agent_loop/generate_sequences/min:np.float64(1.766729136928916) - timing_s/agent_loop/generate_sequences/max:np.float64(6.959899395704269) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.6071650397134363) - timing_s/agent_
|
|
step:58 - global_seqlen/min:17453 - global_seqlen/max:21661 - global_seqlen/minmax_diff:4208 - global_seqlen/balanced_min:19541 - global_seqlen/balanced_max:19551 - global_seqlen/mean:19547.0 - actor/entropy:0.4006923735141754 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.17913368344306946 - training/rollout_probs_diff_mean:0.0034811769146472216 - training/rollout_probs_diff_std:0.007711675949394703 - training/rollout_actor_probs_pearson_corr:0.9995923638343811 - self_distillation/success_group_fraction:1.0 - self_distillation/success_sample_fraction:1.0 - 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:1.0 - rollout_corr/rollout_is_mean:1.000034213066101 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.558562159538269 - rollout_corr/rollout_is_min:0.6606763601303101 - rollout_corr/rollout_is_std:0.0273481085896492 - rollout_corr/rollout_is_eff_sample_size:0.9992527589134378 - rollout_corr/rollout_is_seq_mean:1.0000121593475342 - rollout_corr/rollout_is_seq_std:0.0018669103737920523 - rollout_corr/rollout_is_seq_max:1.0074886083602905 - rollout_corr/rollout_is_seq_min:0.9931515455245972 - rollout_corr/rollout_is_seq_max_deviation:0.007488608360290527 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5310271130874753) - rollout_corr/training_log_ppl:np.float64(0.41603791125817224) - rollout_corr/kl:np.float64(0.0003818474306171993) - rollout_corr/k3_kl:np.float64(0.00039403548791483445) - rollout_corr/rollout_ppl:np.float64(1.530425132252276) - rollout_corr/rollout_log_ppl:np.float64(0.4156560630071908) - rollout_corr/log_ppl_diff:np.float64(0.0003818482509814203) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014238196308724582) - rollout_corr/log_ppl_diff_max:np.float64(0.007478892803192139) - rollout_corr/log_ppl_diff_min:np.float64(-0.0067653656005859375) - rollout_corr/ppl_ratio:np.float64(1.0003836897667497) - rollout_corr/chi2_token:np.float64(0.0008168586064130068) - rollout_corr/chi2_seq:np.float64(0.433692816644907) - actor/pg_loss:np.float64(0.02866315551364096) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.0) - actor/grad_norm:np.float64(0.20574557781219482) - perf/mfu/actor:np.float64(0.0452280571470113) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.6932258605957) - actor/lr:np.float64(1e-05) - training/global_step:58 - training/epoch:0 - critic/score/mean:0.828125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.828125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.006002643145620823 - critic/advantages/max:0.75 - critic/advantages/min:-0.75 - critic/returns/mean:-0.006002643145620823 - critic/returns/max:0.75 - critic/returns/min:-0.75 - response_length/mean:334.0 - response_length/max:903.0 - response_length/min:140.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:334.0 - response_length_non_aborted/max:903.0 - response_length_non_aborted/min:140.0 - 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:4.879012703895569e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.7956734783947468) - timing_s/agent_loop/generate_sequences/max:np.float64(7.763147037476301) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.936746428575134) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timing_s/agent_loop/tool
|
|
step:59 - global_seqlen/min:17324 - global_seqlen/max:20844 - global_seqlen/minmax_diff:3520 - global_seqlen/balanced_min:18979 - global_seqlen/balanced_max:18983 - global_seqlen/mean:18980.625 - actor/entropy:0.3858022391796112 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.39921924471855164 - training/rollout_probs_diff_mean:0.003350769402459264 - training/rollout_probs_diff_std:0.00762861268594861 - training/rollout_actor_probs_pearson_corr:0.9995944499969482 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.89453125 - self_distillation/correct_sample_fraction:0.66015625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.89453125 - rollout_corr/rollout_is_mean:0.9998527765274048 - rollout_corr/rollout_is_ratio_fraction_high:1.203470765176462e-05 - rollout_corr/rollout_is_ratio_fraction_low:1.203470765176462e-05 - rollout_corr/rollout_is_max:2.059091806411743 - rollout_corr/rollout_is_min:0.24108079075813293 - rollout_corr/rollout_is_std:0.026432663202285767 - rollout_corr/rollout_is_eff_sample_size:0.9993016831318713 - rollout_corr/rollout_is_seq_mean:0.9998456239700317 - rollout_corr/rollout_is_seq_std:0.0016383329639211297 - rollout_corr/rollout_is_seq_max:1.003669261932373 - rollout_corr/rollout_is_seq_min:0.9938803315162659 - rollout_corr/rollout_is_seq_max_deviation:0.006119668483734131 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5132068768143654) - rollout_corr/training_log_ppl:np.float64(0.4052205993211828) - rollout_corr/kl:np.float64(0.0005355038742678175) - rollout_corr/k3_kl:np.float64(0.0003811805843838556) - rollout_corr/rollout_ppl:np.float64(1.5123617579229176) - rollout_corr/rollout_log_ppl:np.float64(0.4046850940794684) - rollout_corr/log_ppl_diff:np.float64(0.0005355052417144179) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013248705072328448) - rollout_corr/log_ppl_diff_max:np.float64(0.00656050443649292) - rollout_corr/log_ppl_diff_min:np.float64(-0.0032739639282226562) - rollout_corr/ppl_ratio:np.float64(1.0005370606668293) - rollout_corr/chi2_token:np.float64(0.00044868001714348793) - rollout_corr/chi2_seq:np.float64(0.11397636123001575) - actor/pg_loss:np.float64(0.021889999512495706) - 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(0.11556123942136765) - perf/mfu/actor:np.float64(0.04430367476268178) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(196.64712524414062) - actor/lr:np.float64(1e-05) - training/global_step:59 - training/epoch:0 - critic/score/mean:0.66015625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.66015625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0056954254396259785 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0056954254396259785 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:324.58203125 - response_length/max:1043.0 - response_length/min:151.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:324.58203125 - response_length_non_aborted/max:1043.0 - response_length_non_aborted/min:151.0 - 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:0.00012916699051856995 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9462187886238098) - timing_s/agent_loop/generate_sequences/max:np.float64(8.34768102876842) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.7
|
|
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>
|
|
To determine the correct reactant for the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1", we need to identify which of the options matches the structural features of the target molecule.
|
|
|
|
The target molecule has:
|
|
- A chlorinated aromatic ring (Cl on a benzene ring)
|
|
- A fluorinated aromatic ring (C(F)(F)F, indicating a trifluoromethyl group)
|
|
- A nitrile group (Nc1ccn...)
|
|
- A nitrogen-containing heterocycle
|
|
|
|
Now, let's analyze the options:
|
|
|
|
- **Option A**: "O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1"
|
|
This option contains all the key structural elements of the target molecule: the chlorinated aromatic ring, the fluorinated aromatic ring, and the nitrogen-containing heterocycle. It also has a carbonyl group, which could be part of a synthetic pathway. This is a strong candidate.
|
|
|
|
- **Option B**: "O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1"
|
|
This option has a different arrangement of functional groups and includes a different chlorinated ring structure. It does not match the target molecule in terms of the fluorinated aromatic ring or the nitrogen-containing heterocycle.
|
|
|
|
- **Option C**: "O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1"
|
|
This option introduces a hydroxyl group and a different spatial arrangement of the fluorinated and chlorinated rings. It is structurally different from the target molecule.
|
|
|
|
- **Option D**: "O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F"
|
|
This option has a different ring system and includes a different fluorinated group, which is not part of the target molecule's structure.
|
|
|
|
Thus, **Option A** is the only one that matches the structural features of the target molecule.
|
|
</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-Qwen3-8B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_60
|
|
step:60 - global_seqlen/min:16531 - global_seqlen/max:19958 - global_seqlen/minmax_diff:3427 - global_seqlen/balanced_min:18241 - global_seqlen/balanced_max:18257 - global_seqlen/mean:18252.125 - actor/entropy:0.40044477581977844 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.14492791891098022 - training/rollout_probs_diff_mean:0.0035471946466714144 - training/rollout_probs_diff_std:0.007729396689683199 - training/rollout_actor_probs_pearson_corr:0.9995673298835754 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.89453125 - self_distillation/correct_sample_fraction:0.6796875 - 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:1.0000903606414795 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.562499761581421 - rollout_corr/rollout_is_min:0.6212186813354492 - rollout_corr/rollout_is_std:0.027206074446439743 - rollout_corr/rollout_is_eff_sample_size:0.9992604960013418 - rollout_corr/rollout_is_seq_mean:1.0000561475753784 - rollout_corr/rollout_is_seq_std:0.0016895101871341467 - rollout_corr/rollout_is_seq_max:1.0051277875900269 - rollout_corr/rollout_is_seq_min:0.9949024319648743 - rollout_corr/rollout_is_seq_max_deviation:0.0051277875900268555 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5430215983651578) - rollout_corr/training_log_ppl:np.float64(0.424149729864439) - rollout_corr/kl:np.float64(0.00034226323059449726) - rollout_corr/k3_kl:np.float64(0.00039832691112451357) - rollout_corr/rollout_ppl:np.float64(1.5425134282559156) - rollout_corr/rollout_log_ppl:np.float64(0.4238074625027366) - rollout_corr/log_ppl_diff:np.float64(0.00034226736170239747) - rollout_corr/log_ppl_abs_diff:np.float64(0.001299959869356826) - rollout_corr/log_ppl_diff_max:np.float64(0.0055245161056518555) - rollout_corr/log_ppl_diff_min:np.float64(-0.0045236945152282715) - rollout_corr/ppl_ratio:np.float64(1.0003437479026616) - rollout_corr/chi2_token:np.float64(0.000914017902687192) - rollout_corr/chi2_seq:np.float64(0.23901105695404112) - actor/pg_loss:np.float64(0.024737755411479156) - 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(0.11609493941068649) - perf/mfu/actor:np.float64(0.04205042076670848) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(197.53561401367188) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7136904761904762) - val-aux/sciknoweval/reward/std@16:np.float64(0.12518602507764898) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7633285714285715) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.11071671555892403) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6628190476190476) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.10071134253883397) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7131714285714286) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.12549671696459871) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.8080809523809523) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.0884673739845118) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.625604761904762) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.07346178837283983) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7166095238095238) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.09911450660194439) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.8445714285714286) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.0623856331842929) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.5967761904761905) - val-aux/sciknoweval
|
|
step:61 - global_seqlen/min:15060 - global_seqlen/max:22161 - global_seqlen/minmax_diff:7101 - global_seqlen/balanced_min:19077 - global_seqlen/balanced_max:19084 - global_seqlen/mean:19082.375 - actor/entropy:0.36471593379974365 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12173795700073242 - training/rollout_probs_diff_mean:0.0032218515407294035 - training/rollout_probs_diff_std:0.007416716311126947 - training/rollout_actor_probs_pearson_corr:0.9996089339256287 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.87109375 - 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:0.87109375 - rollout_corr/rollout_is_mean:1.0000851154327393 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6381045579910278 - rollout_corr/rollout_is_min:0.6432838439941406 - rollout_corr/rollout_is_std:0.025892721489071846 - rollout_corr/rollout_is_eff_sample_size:0.9993300161348888 - rollout_corr/rollout_is_seq_mean:1.0000503063201904 - rollout_corr/rollout_is_seq_std:0.0016334683168679476 - rollout_corr/rollout_is_seq_max:1.005224347114563 - rollout_corr/rollout_is_seq_min:0.9958754181861877 - rollout_corr/rollout_is_seq_max_deviation:0.005224347114562988 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4750278238207102) - rollout_corr/training_log_ppl:np.float64(0.38085875037359074) - rollout_corr/kl:np.float64(0.00029834461101074794) - rollout_corr/k3_kl:np.float64(0.00034861160895616194) - rollout_corr/rollout_ppl:np.float64(1.4745741924270988) - rollout_corr/rollout_log_ppl:np.float64(0.380560403020354) - rollout_corr/log_ppl_diff:np.float64(0.00029834735323674977) - rollout_corr/log_ppl_abs_diff:np.float64(0.001295119378482923) - rollout_corr/log_ppl_diff_max:np.float64(0.004462242126464844) - rollout_corr/log_ppl_diff_min:np.float64(-0.004393279552459717) - rollout_corr/ppl_ratio:np.float64(1.0002997086849064) - rollout_corr/chi2_token:np.float64(0.0008040203247219324) - rollout_corr/chi2_seq:np.float64(0.45568988169543445) - actor/pg_loss:np.float64(0.022238394272790174) - 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(0.11320225894451141) - perf/mfu/actor:np.float64(0.044117055188793955) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(199.08259201049805) - actor/lr:np.float64(1e-05) - training/global_step:61 - 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:-0.002721169963479042 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.002721169963479042 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:334.29296875 - response_length/max:943.0 - response_length/min:129.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:334.29296875 - response_length_non_aborted/max:943.0 - response_length_non_aborted/min:129.0 - 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:0.00013157352805137634 - timing_s/agent_loop/generate_sequences/min:np.float64(1.7653572764247656) - timing_s/agent_loop/generate_sequences/max:np.float64(7.825763072818518) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.9960993432541727) - timing_s/agent_l
|
|
step:62 - global_seqlen/min:17321 - global_seqlen/max:21554 - global_seqlen/minmax_diff:4233 - global_seqlen/balanced_min:19101 - global_seqlen/balanced_max:19102 - global_seqlen/mean:19101.75 - actor/entropy:0.44737741351127625 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.11699187755584717 - training/rollout_probs_diff_mean:0.0038474206812679768 - training/rollout_probs_diff_std:0.007895644754171371 - training/rollout_actor_probs_pearson_corr:0.9995954632759094 - self_distillation/success_group_fraction:0.8125 - self_distillation/success_sample_fraction:0.8046875 - 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:0.8046875 - rollout_corr/rollout_is_mean:0.9999257922172546 - rollout_corr/rollout_is_ratio_fraction_high:1.3122325981385075e-05 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:2.2864253520965576 - rollout_corr/rollout_is_min:0.6261085867881775 - rollout_corr/rollout_is_std:0.028734013438224792 - rollout_corr/rollout_is_eff_sample_size:0.9991747995601472 - rollout_corr/rollout_is_seq_mean:0.9999727010726929 - rollout_corr/rollout_is_seq_std:0.0018650178099051118 - rollout_corr/rollout_is_seq_max:1.0050915479660034 - rollout_corr/rollout_is_seq_min:0.9952458739280701 - rollout_corr/rollout_is_seq_max_deviation:0.005091547966003418 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5924441083334386) - rollout_corr/training_log_ppl:np.float64(0.4572776145651005) - rollout_corr/kl:np.float64(0.0004526559134667707) - rollout_corr/k3_kl:np.float64(0.00042531462557349187) - rollout_corr/rollout_ppl:np.float64(1.5917160850949585) - rollout_corr/rollout_log_ppl:np.float64(0.45682495782966726) - rollout_corr/log_ppl_diff:np.float64(0.00045265673543326557) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014936592488083988) - rollout_corr/log_ppl_diff_max:np.float64(0.005309104919433594) - rollout_corr/log_ppl_diff_min:np.float64(-0.0046530961990356445) - rollout_corr/ppl_ratio:np.float64(1.0004544605035335) - rollout_corr/chi2_token:np.float64(0.0008097072131931782) - rollout_corr/chi2_seq:np.float64(0.20479423366487026) - actor/pg_loss:np.float64(0.022945368251384934) - 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(0.11080524325370789) - perf/mfu/actor:np.float64(0.04438774888984196) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.52357482910156) - actor/lr:np.float64(1e-05) - training/global_step:62 - 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:0.0008841167436912656 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0008841167436912656 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:297.6796875 - response_length/max:688.0 - response_length/min:129.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:297.6796875 - response_length_non_aborted/max:688.0 - response_length_non_aborted/min:129.0 - 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:0.00012746453285217285 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8271600063890219) - timing_s/agent_loop/generate_sequences/max:np.float64(6.742027984932065) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.753349417464051) - timing_s/a
|
|
step:63 - global_seqlen/min:17963 - global_seqlen/max:21792 - global_seqlen/minmax_diff:3829 - global_seqlen/balanced_min:19538 - global_seqlen/balanced_max:19562 - global_seqlen/mean:19546.125 - actor/entropy:0.41168642044067383 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12245827913284302 - training/rollout_probs_diff_mean:0.0033766920678317547 - training/rollout_probs_diff_std:0.0074678584933280945 - training/rollout_actor_probs_pearson_corr:0.9996307492256165 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.8515625 - 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:0.8515625 - rollout_corr/rollout_is_mean:1.000119686126709 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.559311866760254 - rollout_corr/rollout_is_min:0.6956509947776794 - rollout_corr/rollout_is_std:0.026697421446442604 - rollout_corr/rollout_is_eff_sample_size:0.9992879933909776 - rollout_corr/rollout_is_seq_mean:1.0000903606414795 - rollout_corr/rollout_is_seq_std:0.001833563786931336 - rollout_corr/rollout_is_seq_max:1.0069832801818848 - rollout_corr/rollout_is_seq_min:0.9946748614311218 - rollout_corr/rollout_is_seq_max_deviation:0.006983280181884766 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5700851790606976) - rollout_corr/training_log_ppl:np.float64(0.4412491190014407) - rollout_corr/kl:np.float64(0.0002873682499267005) - rollout_corr/k3_kl:np.float64(0.00037768021957163) - rollout_corr/rollout_ppl:np.float64(1.5696247527375817) - rollout_corr/rollout_log_ppl:np.float64(0.44096174818696454) - rollout_corr/log_ppl_diff:np.float64(0.0002873708144761622) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014114673831500113) - rollout_corr/log_ppl_diff_max:np.float64(0.005928069353103638) - rollout_corr/log_ppl_diff_min:np.float64(-0.006410956382751465) - rollout_corr/ppl_ratio:np.float64(1.0002890892792493) - rollout_corr/chi2_token:np.float64(0.0009420521091669798) - rollout_corr/chi2_seq:np.float64(0.36746392282657325) - actor/pg_loss:np.float64(0.021043284359620884) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1484375) - actor/grad_norm:np.float64(0.10576817393302917) - perf/mfu/actor:np.float64(0.045123485656912266) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.08647537231445) - actor/lr:np.float64(1e-05) - training/global_step:63 - training/epoch:1 - 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:-0.0029664128087460995 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0029664128087460995 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:324.59765625 - response_length/max:992.0 - response_length/min:126.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:324.59765625 - response_length_non_aborted/max:992.0 - response_length_non_aborted/min:126.0 - 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:0.00013443082571029663 - timing_s/agent_loop/generate_sequences/min:np.float64(1.6516579203307629) - timing_s/agent_loop/generate_sequences/max:np.float64(8.248821556568146) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.7947063500396325) - timing_s/agent_loop/t
|
|
step:64 - global_seqlen/min:15434 - global_seqlen/max:22895 - global_seqlen/minmax_diff:7461 - global_seqlen/balanced_min:19898 - global_seqlen/balanced_max:19949 - global_seqlen/mean:19912.5 - actor/entropy:0.4157674312591553 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12214788794517517 - training/rollout_probs_diff_mean:0.0034745135344564915 - training/rollout_probs_diff_std:0.0075850896537303925 - training/rollout_actor_probs_pearson_corr:0.9996185302734375 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.93359375 - self_distillation/correct_sample_fraction:0.8359375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.93359375 - rollout_corr/rollout_is_mean:0.999928891658783 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.2834035158157349 - rollout_corr/rollout_is_min:0.6945799589157104 - rollout_corr/rollout_is_std:0.026846589520573616 - rollout_corr/rollout_is_eff_sample_size:0.999279660688855 - rollout_corr/rollout_is_seq_mean:0.9999498724937439 - rollout_corr/rollout_is_seq_std:0.00161889533046633 - rollout_corr/rollout_is_seq_max:1.0053268671035767 - rollout_corr/rollout_is_seq_min:0.9947873950004578 - rollout_corr/rollout_is_seq_max_deviation:0.00532686710357666 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5546787269413471) - rollout_corr/training_log_ppl:np.float64(0.43200751044787467) - rollout_corr/kl:np.float64(0.00042410349355037624) - rollout_corr/k3_kl:np.float64(0.00037402251840035206) - rollout_corr/rollout_ppl:np.float64(1.5540012088604271) - rollout_corr/rollout_log_ppl:np.float64(0.4315834029985126) - rollout_corr/log_ppl_diff:np.float64(0.00042410744936205447) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012981545587535948) - rollout_corr/log_ppl_diff_max:np.float64(0.005594015121459961) - rollout_corr/log_ppl_diff_min:np.float64(-0.004669129848480225) - rollout_corr/ppl_ratio:np.float64(1.0004255042877048) - rollout_corr/chi2_token:np.float64(0.0006495912093669176) - rollout_corr/chi2_seq:np.float64(0.1763608583714813) - actor/pg_loss:np.float64(0.023846463836889598) - 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(0.1116981953382492) - perf/mfu/actor:np.float64(0.04551338606839088) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(197.99003076553345) - actor/lr:np.float64(1e-05) - training/global_step:64 - training/epoch:1 - critic/score/mean:0.8359375 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.8359375 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.00046380815911106765 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.00046380815911106765 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:329.515625 - response_length/max:1253.0 - response_length/min:115.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:329.515625 - response_length_non_aborted/max:1253.0 - response_length_non_aborted/min:115.0 - 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:0.00011455267667770386 - timing_s/agent_loop/generate_sequences/min:np.float64(1.5306183770298958) - timing_s/agent_loop/generate_sequences/max:np.float64(9.492812514305115) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.947460310511815) - timing_s/agent_loop/tool_ca
|
|
step:65 - global_seqlen/min:16356 - global_seqlen/max:22894 - global_seqlen/minmax_diff:6538 - global_seqlen/balanced_min:19797 - global_seqlen/balanced_max:19800 - global_seqlen/mean:19798.375 - actor/entropy:0.4274553060531616 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1525484323501587 - training/rollout_probs_diff_mean:0.003610642161220312 - training/rollout_probs_diff_std:0.00765934307128191 - training/rollout_actor_probs_pearson_corr:0.9996089339256287 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.93359375 - self_distillation/correct_sample_fraction:0.8125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.93359375 - rollout_corr/rollout_is_mean:1.000006914138794 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5713839530944824 - rollout_corr/rollout_is_min:0.720781147480011 - rollout_corr/rollout_is_std:0.02796875685453415 - rollout_corr/rollout_is_eff_sample_size:0.999218360080856 - rollout_corr/rollout_is_seq_mean:0.9999938607215881 - rollout_corr/rollout_is_seq_std:0.001600720570422709 - rollout_corr/rollout_is_seq_max:1.0049405097961426 - rollout_corr/rollout_is_seq_min:0.9948761463165283 - rollout_corr/rollout_is_seq_max_deviation:0.00512385368347168 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.556871825363487) - rollout_corr/training_log_ppl:np.float64(0.43447944574290887) - rollout_corr/kl:np.float64(0.00040245426350549174) - rollout_corr/k3_kl:np.float64(0.0003962767012808399) - rollout_corr/rollout_ppl:np.float64(1.5562205673195422) - rollout_corr/rollout_log_ppl:np.float64(0.4340769902919419) - rollout_corr/log_ppl_diff:np.float64(0.00040245545096695423) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012919731670990586) - rollout_corr/log_ppl_diff_max:np.float64(0.005877017974853516) - rollout_corr/log_ppl_diff_min:np.float64(-0.004272162914276123) - rollout_corr/ppl_ratio:np.float64(1.0004038133192807) - rollout_corr/chi2_token:np.float64(0.0007853044662624598) - rollout_corr/chi2_seq:np.float64(0.2540235547348857) - actor/pg_loss:np.float64(0.02478727472407627) - 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(0.11220979690551758) - perf/mfu/actor:np.float64(0.04588716696135464) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(197.89067840576172) - actor/lr:np.float64(1e-05) - training/global_step:65 - training/epoch:1 - critic/score/mean:0.8125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.8125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.010113711468875408 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.010113711468875408 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:335.10546875 - response_length/max:924.0 - response_length/min:124.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:335.10546875 - response_length_non_aborted/max:924.0 - response_length_non_aborted/min:124.0 - 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:4.026666283607483e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.7985296081751585) - timing_s/agent_loop/generate_sequences/max:np.float64(8.12501939944923) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.260380339197582) - timing_s/agent_loop/tool_calls/min:np.float6
|
|
step:66 - global_seqlen/min:16581 - global_seqlen/max:21813 - global_seqlen/minmax_diff:5232 - global_seqlen/balanced_min:20175 - global_seqlen/balanced_max:20191 - global_seqlen/mean:20188.0 - actor/entropy:0.375995010137558 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.22288969159126282 - training/rollout_probs_diff_mean:0.0031780749559402466 - training/rollout_probs_diff_std:0.00733711663633585 - training/rollout_actor_probs_pearson_corr:0.9996129274368286 - 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:0.9998968839645386 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6083227396011353 - rollout_corr/rollout_is_min:0.6693816781044006 - rollout_corr/rollout_is_std:0.025982344523072243 - rollout_corr/rollout_is_eff_sample_size:0.999325254173322 - rollout_corr/rollout_is_seq_mean:0.999854326248169 - rollout_corr/rollout_is_seq_std:0.0015588144306093454 - rollout_corr/rollout_is_seq_max:1.004082441329956 - rollout_corr/rollout_is_seq_min:0.9936635494232178 - rollout_corr/rollout_is_seq_max_deviation:0.0063364505767822266 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.497855738736689) - rollout_corr/training_log_ppl:np.float64(0.39445046201581135) - rollout_corr/kl:np.float64(0.0005056746705349724) - rollout_corr/k3_kl:np.float64(0.0003600496421825028) - rollout_corr/rollout_ppl:np.float64(1.4970720945857465) - rollout_corr/rollout_log_ppl:np.float64(0.39394478738540784) - rollout_corr/log_ppl_diff:np.float64(0.0005056746304035187) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012444012099877) - rollout_corr/log_ppl_diff_max:np.float64(0.006895571947097778) - rollout_corr/log_ppl_diff_min:np.float64(-0.003816962242126465) - rollout_corr/ppl_ratio:np.float64(1.0005070415791124) - rollout_corr/chi2_token:np.float64(0.0004320081789046526) - rollout_corr/chi2_seq:np.float64(0.0932999555952847) - actor/pg_loss:np.float64(0.019179151926437044) - 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(0.08925329148769379) - perf/mfu/actor:np.float64(0.04599040867509213) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(197.88832092285156) - actor/lr:np.float64(1e-05) - training/global_step:66 - 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:0.0011897935764864087 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.0011897935764864087 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:340.625 - response_length/max:1211.0 - response_length/min:126.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:340.625 - response_length_non_aborted/max:1211.0 - response_length_non_aborted/min:126.0 - 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:0.00015368498861789703 - timing_s/agent_loop/generate_sequences/min:np.float64(1.693924829363823) - timing_s/agent_loop/generate_sequences/max:np.float64(9.594088276848197) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.065205566876102) - timing_s/agent_loop/tool_calls/min:np.float64(0.0) - timi
|
|
step:67 - global_seqlen/min:15070 - global_seqlen/max:21175 - global_seqlen/minmax_diff:6105 - global_seqlen/balanced_min:18909 - global_seqlen/balanced_max:18914 - global_seqlen/mean:18912.75 - actor/entropy:0.40055450797080994 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12214356660842896 - training/rollout_probs_diff_mean:0.0035869525745511055 - training/rollout_probs_diff_std:0.007802943233400583 - training/rollout_actor_probs_pearson_corr:0.9995772242546082 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.8984375 - self_distillation/correct_sample_fraction:0.71875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.8984375 - rollout_corr/rollout_is_mean:1.0000096559524536 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.461733341217041 - rollout_corr/rollout_is_min:0.7160922884941101 - rollout_corr/rollout_is_std:0.026988307014107704 - rollout_corr/rollout_is_eff_sample_size:0.999272042339952 - rollout_corr/rollout_is_seq_mean:1.0000311136245728 - rollout_corr/rollout_is_seq_std:0.0016181101091206074 - rollout_corr/rollout_is_seq_max:1.0043017864227295 - rollout_corr/rollout_is_seq_min:0.9960224032402039 - rollout_corr/rollout_is_seq_max_deviation:0.004301786422729492 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.526858836878091) - rollout_corr/training_log_ppl:np.float64(0.4155436938162893) - rollout_corr/kl:np.float64(0.00034789064843465667) - rollout_corr/k3_kl:np.float64(0.00037900395125234354) - rollout_corr/rollout_ppl:np.float64(1.5263166134245694) - rollout_corr/rollout_log_ppl:np.float64(0.415195803710958) - rollout_corr/log_ppl_diff:np.float64(0.0003478901053313166) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013251047057565302) - rollout_corr/log_ppl_diff_max:np.float64(0.004528522491455078) - rollout_corr/log_ppl_diff_min:np.float64(-0.003804445266723633) - rollout_corr/ppl_ratio:np.float64(1.0003492587711662) - rollout_corr/chi2_token:np.float64(0.0008259830065071583) - rollout_corr/chi2_seq:np.float64(0.16636192821897566) - actor/pg_loss:np.float64(0.023465468737413175) - 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(0.11083564162254333) - perf/mfu/actor:np.float64(0.04420330819029732) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(197.85264205932617) - actor/lr:np.float64(1e-05) - training/global_step:67 - training/epoch:1 - critic/score/mean:0.71875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.71875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.004832549951970577 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:-0.004832549951970577 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:299.1796875 - response_length/max:906.0 - response_length/min:134.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:299.1796875 - response_length_non_aborted/max:906.0 - response_length_non_aborted/min:134.0 - 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:0.0001032855361700058 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9132667556405067) - timing_s/agent_loop/generate_sequences/max:np.float64(7.524463154375553) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.654308578770724) - timing_s/agent_loop/tool_calls/min:np
|
|
step:68 - global_seqlen/min:15979 - global_seqlen/max:24690 - global_seqlen/minmax_diff:8711 - global_seqlen/balanced_min:19986 - global_seqlen/balanced_max:20015 - global_seqlen/mean:20008.375 - actor/entropy:0.39408883452415466 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2594304084777832 - training/rollout_probs_diff_mean:0.0034657458309084177 - training/rollout_probs_diff_std:0.007732602767646313 - training/rollout_actor_probs_pearson_corr:0.9995718598365784 - self_distillation/success_group_fraction:0.71875 - self_distillation/success_sample_fraction:0.71484375 - 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:0.71484375 - rollout_corr/rollout_is_mean:1.0000478029251099 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.2599060028151143e-05 - rollout_corr/rollout_is_max:1.617263913154602 - rollout_corr/rollout_is_min:0.36691319942474365 - rollout_corr/rollout_is_std:0.0269794724881649 - rollout_corr/rollout_is_eff_sample_size:0.999272637519277 - rollout_corr/rollout_is_seq_mean:1.0000691413879395 - rollout_corr/rollout_is_seq_std:0.0017838911153376102 - rollout_corr/rollout_is_seq_max:1.0051558017730713 - rollout_corr/rollout_is_seq_min:0.995134174823761 - rollout_corr/rollout_is_seq_max_deviation:0.005155801773071289 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5447495942935348) - rollout_corr/training_log_ppl:np.float64(0.42585801042150706) - rollout_corr/kl:np.float64(0.0003275737801375911) - rollout_corr/k3_kl:np.float64(0.00039677792463521655) - rollout_corr/rollout_ppl:np.float64(1.5442284056916833) - rollout_corr/rollout_log_ppl:np.float64(0.4255304363905452) - rollout_corr/log_ppl_diff:np.float64(0.00032757403096184134) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013972398010082543) - rollout_corr/log_ppl_diff_max:np.float64(0.006733328104019165) - rollout_corr/log_ppl_diff_min:np.float64(-0.004459559917449951) - rollout_corr/ppl_ratio:np.float64(1.0003292602486908) - rollout_corr/chi2_token:np.float64(0.0009306683205068111) - rollout_corr/chi2_seq:np.float64(0.19451901805587113) - actor/pg_loss:np.float64(0.020688947131930036) - 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(0.11469472199678421) - perf/mfu/actor:np.float64(0.04556282014597235) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(197.81998825073242) - actor/lr:np.float64(1e-05) - training/global_step:68 - training/epoch:1 - 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:-0.005115218460559845 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.005115218460559845 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:310.04296875 - response_length/max:1353.0 - response_length/min:31.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:310.04296875 - response_length_non_aborted/max:1353.0 - response_length_non_aborted/min:31.0 - 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:0.00010886043310165405 - timing_s/agent_loop/generate_sequences/min:np.float64(0.45913008227944374) - timing_s/agent_loop/generate_sequences/max:np.float64(9.884037492796779) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.627023144472332) - timin
|
|
step:69 - global_seqlen/min:16792 - global_seqlen/max:25100 - global_seqlen/minmax_diff:8308 - global_seqlen/balanced_min:20801 - global_seqlen/balanced_max:20816 - global_seqlen/mean:20812.25 - actor/entropy:0.4084627628326416 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2103867530822754 - training/rollout_probs_diff_mean:0.0034431288950145245 - training/rollout_probs_diff_std:0.007618641946464777 - training/rollout_actor_probs_pearson_corr:0.9996013641357422 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83984375 - 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:0.83984375 - rollout_corr/rollout_is_mean:1.000182867050171 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6337867975234985 - rollout_corr/rollout_is_min:0.6301739811897278 - rollout_corr/rollout_is_std:0.02755005471408367 - rollout_corr/rollout_is_eff_sample_size:0.999241927191657 - rollout_corr/rollout_is_seq_mean:1.0002250671386719 - rollout_corr/rollout_is_seq_std:0.0016926646931096911 - rollout_corr/rollout_is_seq_max:1.0049790143966675 - rollout_corr/rollout_is_seq_min:0.994762122631073 - rollout_corr/rollout_is_seq_max_deviation:0.005237877368927002 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5649619437754154) - rollout_corr/training_log_ppl:np.float64(0.43678645958425477) - rollout_corr/kl:np.float64(0.00018486184533728078) - rollout_corr/k3_kl:np.float64(0.0004098876303260113) - rollout_corr/rollout_ppl:np.float64(1.5646906518377364) - rollout_corr/rollout_log_ppl:np.float64(0.43660159487626515) - rollout_corr/log_ppl_diff:np.float64(0.00018486470798961818) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013399154704529792) - rollout_corr/log_ppl_diff_max:np.float64(0.005812525749206543) - rollout_corr/log_ppl_diff_min:np.float64(-0.004365712404251099) - rollout_corr/ppl_ratio:np.float64(1.000186308985576) - rollout_corr/chi2_token:np.float64(0.0012745179701596498) - rollout_corr/chi2_seq:np.float64(0.3666792728472501) - actor/pg_loss:np.float64(0.023392049077301635) - 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(0.1004951074719429) - perf/mfu/actor:np.float64(0.04728507791771519) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(197.77862930297852) - actor/lr:np.float64(1e-05) - training/global_step:69 - 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:-0.0022423325572162867 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0022423325572162867 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:340.5703125 - response_length/max:1217.0 - response_length/min:150.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:340.5703125 - response_length_non_aborted/max:1217.0 - response_length_non_aborted/min:150.0 - 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:2.7043744921684265e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9933093041181564) - timing_s/agent_loop/generate_sequences/max:np.float64(9.56888977251947) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.033180785758304) - timing_s/agent_loop/too
|
|
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 target molecule "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1" contains a chlorinated aromatic ring (with Cl), a fluorinated aromatic ring (with C(F)(F)F), and a nitrogen-containing heterocyclic ring. These are key structural features that must be matched by the correct reactant.
|
|
|
|
- **Option A** contains the same chlorinated and fluorinated aromatic rings as the target, along with the nitrogen-containing ring. It is chemically consistent with the target structure.
|
|
- **Option B** introduces a different ring structure with a chlorinated aromatic ring but lacks the fluorinated group and has a different nitrogen arrangement.
|
|
- **Option C** includes an additional hydroxyl group and a different substituent pattern, which deviates from the target.
|
|
- **Option D** has a different arrangement of the fluorinated group and lacks the chlorinated aromatic ring, making it inconsistent.
|
|
|
|
Therefore, **Option A** is the only reactant that matches the target molecule's structure.
|
|
</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-Qwen3-8B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_70
|
|
step:70 - global_seqlen/min:14101 - global_seqlen/max:21559 - global_seqlen/minmax_diff:7458 - global_seqlen/balanced_min:19138 - global_seqlen/balanced_max:19159 - global_seqlen/mean:19151.0 - actor/entropy:0.4222305417060852 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.17251300811767578 - training/rollout_probs_diff_mean:0.003655370557680726 - training/rollout_probs_diff_std:0.007862560451030731 - training/rollout_actor_probs_pearson_corr:0.9995858073234558 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.92578125 - 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:0.92578125 - rollout_corr/rollout_is_mean:0.9999498128890991 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.3009991562284995e-05 - rollout_corr/rollout_is_max:1.4511783123016357 - rollout_corr/rollout_is_min:0.46163684129714966 - rollout_corr/rollout_is_std:0.02803049050271511 - rollout_corr/rollout_is_eff_sample_size:0.9992149084256371 - rollout_corr/rollout_is_seq_mean:0.9999523758888245 - rollout_corr/rollout_is_seq_std:0.0019060494378209114 - rollout_corr/rollout_is_seq_max:1.0059359073638916 - rollout_corr/rollout_is_seq_min:0.9945008158683777 - rollout_corr/rollout_is_seq_max_deviation:0.0059359073638916016 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5844248533248901) - rollout_corr/training_log_ppl:np.float64(0.44899149145931005) - rollout_corr/kl:np.float64(0.0004694185424423125) - rollout_corr/k3_kl:np.float64(0.0004217251729414784) - rollout_corr/rollout_ppl:np.float64(1.583574797026813) - rollout_corr/rollout_log_ppl:np.float64(0.44852206809446216) - rollout_corr/log_ppl_diff:np.float64(0.0004694233648478985) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015368495369330049) - rollout_corr/log_ppl_diff_max:np.float64(0.0063057541847229) - rollout_corr/log_ppl_diff_min:np.float64(-0.00528794527053833) - rollout_corr/ppl_ratio:np.float64(1.0004714189562947) - rollout_corr/chi2_token:np.float64(0.0007506799884140491) - rollout_corr/chi2_seq:np.float64(0.24837386328727007) - actor/pg_loss:np.float64(0.026233967942971503) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.07421875) - actor/grad_norm:np.float64(0.13381417095661163) - perf/mfu/actor:np.float64(0.044537481009946076) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.17802047729492) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7291666666666666) - val-aux/sciknoweval/reward/std@16:np.float64(0.11358182519277789) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7742809523809523) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.10358911875592322) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6844095238095238) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.08736307902530736) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.729404761904762) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.11377167124274151) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.8166333333333334) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.08696209653766836) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6535952380952381) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.061771105060867526) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7313142857142858) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.0890504575047043) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.8524428571428571) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.06531944184317622) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6309142857142858) - val-a
|
|
step:71 - global_seqlen/min:16521 - global_seqlen/max:21698 - global_seqlen/minmax_diff:5177 - global_seqlen/balanced_min:19058 - global_seqlen/balanced_max:19067 - global_seqlen/mean:19063.875 - actor/entropy:0.4110512435436249 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.4120006859302521 - training/rollout_probs_diff_mean:0.003641315968707204 - training/rollout_probs_diff_std:0.007941591553390026 - training/rollout_actor_probs_pearson_corr:0.9995434284210205 - self_distillation/success_group_fraction:1.0 - self_distillation/success_sample_fraction:0.984375 - 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:0.984375 - rollout_corr/rollout_is_mean:0.9998087286949158 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.3315046999196056e-05 - rollout_corr/rollout_is_max:1.5040754079818726 - rollout_corr/rollout_is_min:0.22439981997013092 - rollout_corr/rollout_is_std:0.027481820434331894 - rollout_corr/rollout_is_eff_sample_size:0.9992450219453466 - rollout_corr/rollout_is_seq_mean:0.9998637437820435 - rollout_corr/rollout_is_seq_std:0.0018327267607674003 - rollout_corr/rollout_is_seq_max:1.0059372186660767 - rollout_corr/rollout_is_seq_min:0.9950578808784485 - rollout_corr/rollout_is_seq_max_deviation:0.00593721866607666 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.559848547913134) - rollout_corr/training_log_ppl:np.float64(0.4348594650509767) - rollout_corr/kl:np.float64(0.0005368578483810893) - rollout_corr/k3_kl:np.float64(0.00040066643899194787) - rollout_corr/rollout_ppl:np.float64(1.5590190719813108) - rollout_corr/rollout_log_ppl:np.float64(0.43432260648114607) - rollout_corr/log_ppl_diff:np.float64(0.000536858569830656) - rollout_corr/log_ppl_abs_diff:np.float64(0.0015107898507267237) - rollout_corr/log_ppl_diff_max:np.float64(0.005342572927474976) - rollout_corr/log_ppl_diff_min:np.float64(-0.005357205867767334) - rollout_corr/ppl_ratio:np.float64(1.0005387011915445) - rollout_corr/chi2_token:np.float64(0.0005255555734038353) - rollout_corr/chi2_seq:np.float64(0.1581101689953357) - actor/pg_loss:np.float64(0.027555041258892743) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.015625) - actor/grad_norm:np.float64(0.12825407087802887) - perf/mfu/actor:np.float64(0.04447708076246408) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.6194839477539) - actor/lr:np.float64(1e-05) - training/global_step:71 - 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:-0.0035717615392059088 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0035717615392059088 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:293.37109375 - response_length/max:956.0 - response_length/min:121.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:293.37109375 - response_length_non_aborted/max:956.0 - response_length_non_aborted/min:121.0 - 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:7.373839616775513e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8360966704785824) - timing_s/agent_loop/generate_sequences/max:np.float64(8.122816558927298) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.701703845974407) - timing_s/agent
|
|
step:72 - global_seqlen/min:14257 - global_seqlen/max:20355 - global_seqlen/minmax_diff:6098 - global_seqlen/balanced_min:18102 - global_seqlen/balanced_max:18111 - global_seqlen/mean:18108.5 - actor/entropy:0.44086089730262756 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12245860695838928 - training/rollout_probs_diff_mean:0.0038282007444649935 - training/rollout_probs_diff_std:0.00789185892790556 - training/rollout_actor_probs_pearson_corr:0.9995994567871094 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.9375 - self_distillation/correct_sample_fraction:0.796875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.9375 - rollout_corr/rollout_is_mean:0.9999794363975525 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.424440860748291 - rollout_corr/rollout_is_min:0.7489334940910339 - rollout_corr/rollout_is_std:0.02808784507215023 - rollout_corr/rollout_is_eff_sample_size:0.999211575815639 - rollout_corr/rollout_is_seq_mean:1.000002145767212 - rollout_corr/rollout_is_seq_std:0.0017483425326645374 - rollout_corr/rollout_is_seq_max:1.0042827129364014 - rollout_corr/rollout_is_seq_min:0.9935374855995178 - rollout_corr/rollout_is_seq_max_deviation:0.006462514400482178 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5690998318605125) - rollout_corr/training_log_ppl:np.float64(0.4413370717084035) - rollout_corr/kl:np.float64(0.00039703592849704705) - rollout_corr/k3_kl:np.float64(0.00039925843628907387) - rollout_corr/rollout_ppl:np.float64(1.568475884385407) - rollout_corr/rollout_log_ppl:np.float64(0.44094003591453657) - rollout_corr/log_ppl_diff:np.float64(0.0003970357938669622) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014144566957838833) - rollout_corr/log_ppl_diff_max:np.float64(0.007111489772796631) - rollout_corr/log_ppl_diff_min:np.float64(-0.003797292709350586) - rollout_corr/ppl_ratio:np.float64(1.000398627249524) - rollout_corr/chi2_token:np.float64(0.0008073083590716124) - rollout_corr/chi2_seq:np.float64(0.18585897679440677) - actor/pg_loss:np.float64(0.025938525581295835) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.0625) - actor/grad_norm:np.float64(0.11628760397434235) - perf/mfu/actor:np.float64(0.04208368067334796) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.2704849243164) - actor/lr:np.float64(1e-05) - training/global_step:72 - training/epoch:1 - critic/score/mean:0.796875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.796875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.009716739878058434 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.009716739878058434 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:279.046875 - response_length/max:913.0 - response_length/min:106.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:279.046875 - response_length_non_aborted/max:913.0 - response_length_non_aborted/min:106.0 - 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:0.0001526903361082077 - timing_s/agent_loop/generate_sequences/min:np.float64(1.470910020172596) - timing_s/agent_loop/generate_sequences/max:np.float64(7.320230815559626) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.4056005862075835) - timing_s/agent_loop/tool_calls/min:np.float64(0.0)
|
|
step:73 - global_seqlen/min:17378 - global_seqlen/max:21039 - global_seqlen/minmax_diff:3661 - global_seqlen/balanced_min:18800 - global_seqlen/balanced_max:18882 - global_seqlen/mean:18815.0 - actor/entropy:0.4189092516899109 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.21938490867614746 - training/rollout_probs_diff_mean:0.0035435601603239775 - training/rollout_probs_diff_std:0.00768677331507206 - training/rollout_actor_probs_pearson_corr:0.999596357345581 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.84375 - 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:0.84375 - rollout_corr/rollout_is_mean:0.9999712705612183 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4065992832183838 - rollout_corr/rollout_is_min:0.687300443649292 - rollout_corr/rollout_is_std:0.027495915070176125 - rollout_corr/rollout_is_eff_sample_size:0.9992444267989176 - rollout_corr/rollout_is_seq_mean:0.999947726726532 - rollout_corr/rollout_is_seq_std:0.0017560485284775496 - rollout_corr/rollout_is_seq_max:1.005643367767334 - rollout_corr/rollout_is_seq_min:0.9933860301971436 - rollout_corr/rollout_is_seq_max_deviation:0.006613969802856445 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.570341675542295) - rollout_corr/training_log_ppl:np.float64(0.44261712132720277) - rollout_corr/kl:np.float64(0.00045443968903136067) - rollout_corr/k3_kl:np.float64(0.0004021287906255111) - rollout_corr/rollout_ppl:np.float64(1.5696153114549816) - rollout_corr/rollout_log_ppl:np.float64(0.44216267787851393) - rollout_corr/log_ppl_diff:np.float64(0.0004544434486888349) - rollout_corr/log_ppl_abs_diff:np.float64(0.001368885685224086) - rollout_corr/log_ppl_diff_max:np.float64(0.007200330495834351) - rollout_corr/log_ppl_diff_min:np.float64(-0.005126357078552246) - rollout_corr/ppl_ratio:np.float64(1.0004560791421682) - rollout_corr/chi2_token:np.float64(0.0007027185056358576) - rollout_corr/chi2_seq:np.float64(0.24342316132970154) - actor/pg_loss:np.float64(0.02281117175698455) - 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(0.11419490724802017) - perf/mfu/actor:np.float64(0.04382265623096121) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.18217086791992) - actor/lr:np.float64(1e-05) - training/global_step:73 - 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:-0.006608258001506329 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.006608258001506329 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:307.75 - response_length/max:1118.0 - response_length/min:150.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:307.75 - response_length_non_aborted/max:1118.0 - response_length_non_aborted/min:150.0 - 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:0.0001467689871788025 - timing_s/agent_loop/generate_sequences/min:np.float64(1.3163737431168556) - timing_s/agent_loop/generate_sequences/max:np.float64(8.468793088570237) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.3453373987067607) - timing_s/agent_loop/tool_calls/min:np.float64(0.0
|
|
step:74 - global_seqlen/min:14580 - global_seqlen/max:21623 - global_seqlen/minmax_diff:7043 - global_seqlen/balanced_min:18235 - global_seqlen/balanced_max:18249 - global_seqlen/mean:18243.625 - actor/entropy:0.3892906606197357 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12245851755142212 - training/rollout_probs_diff_mean:0.0035136842634528875 - training/rollout_probs_diff_std:0.007783967070281506 - training/rollout_actor_probs_pearson_corr:0.999575138092041 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.92578125 - 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:0.92578125 - rollout_corr/rollout_is_mean:1.0001652240753174 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.3339412212371826 - rollout_corr/rollout_is_min:0.5362560749053955 - rollout_corr/rollout_is_std:0.027369895949959755 - rollout_corr/rollout_is_eff_sample_size:0.9992518066647457 - rollout_corr/rollout_is_seq_mean:1.0001499652862549 - rollout_corr/rollout_is_seq_std:0.0018182116327807307 - rollout_corr/rollout_is_seq_max:1.0078661441802979 - rollout_corr/rollout_is_seq_min:0.9955509901046753 - rollout_corr/rollout_is_seq_max_deviation:0.007866144180297852 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5045169708319008) - rollout_corr/training_log_ppl:np.float64(0.3999948497803416) - rollout_corr/kl:np.float64(0.00024378678905012663) - rollout_corr/k3_kl:np.float64(0.0003936363919763153) - rollout_corr/rollout_ppl:np.float64(1.5041436273604631) - rollout_corr/rollout_log_ppl:np.float64(0.3997510618646629) - rollout_corr/log_ppl_diff:np.float64(0.00024378791567869484) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014158300182316452) - rollout_corr/log_ppl_diff_max:np.float64(0.004975378513336182) - rollout_corr/log_ppl_diff_min:np.float64(-0.007210075855255127) - rollout_corr/ppl_ratio:np.float64(1.0002454952336848) - rollout_corr/chi2_token:np.float64(0.0010851591359823942) - rollout_corr/chi2_seq:np.float64(0.42862841091118753) - actor/pg_loss:np.float64(0.02352263317152392) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.07421875) - actor/grad_norm:np.float64(0.11920825392007828) - perf/mfu/actor:np.float64(0.04163547336430955) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.14832305908203) - actor/lr:np.float64(1e-05) - training/global_step:74 - 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:0.005139421205967665 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.005139421205967665 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:296.70703125 - response_length/max:1019.0 - response_length/min:125.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:296.70703125 - response_length_non_aborted/max:1019.0 - response_length_non_aborted/min:125.0 - 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:0.00013957731425762177 - timing_s/agent_loop/generate_sequences/min:np.float64(1.428647417575121) - timing_s/agent_loop/generate_sequences/max:np.float64(8.084354534745216) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.4840043943404453) - timing_s/agent_loop/tool_
|
|
step:75 - global_seqlen/min:14638 - global_seqlen/max:21414 - global_seqlen/minmax_diff:6776 - global_seqlen/balanced_min:19389 - global_seqlen/balanced_max:19579 - global_seqlen/mean:19418.0 - actor/entropy:0.4358784854412079 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.10959482192993164 - training/rollout_probs_diff_mean:0.0037747458554804325 - training/rollout_probs_diff_std:0.00785855296999216 - training/rollout_actor_probs_pearson_corr:0.9995870590209961 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.87109375 - self_distillation/correct_sample_fraction:0.78125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.87109375 - rollout_corr/rollout_is_mean:1.0000176429748535 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.331486964772921e-05 - rollout_corr/rollout_is_max:1.3575057983398438 - rollout_corr/rollout_is_min:0.4878958761692047 - rollout_corr/rollout_is_std:0.028326625004410744 - rollout_corr/rollout_is_eff_sample_size:0.9991982455979442 - rollout_corr/rollout_is_seq_mean:1.0000746250152588 - rollout_corr/rollout_is_seq_std:0.0018576147267594934 - rollout_corr/rollout_is_seq_max:1.0058623552322388 - rollout_corr/rollout_is_seq_min:0.9945737719535828 - rollout_corr/rollout_is_seq_max_deviation:0.0058623552322387695 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.579587769228965) - rollout_corr/training_log_ppl:np.float64(0.4477337296702899) - rollout_corr/kl:np.float64(0.00034227544096054885) - rollout_corr/k3_kl:np.float64(0.0004169754695340089) - rollout_corr/rollout_ppl:np.float64(1.5790764666162431) - rollout_corr/rollout_log_ppl:np.float64(0.4473914531408809) - rollout_corr/log_ppl_diff:np.float64(0.00034227652940899134) - rollout_corr/log_ppl_abs_diff:np.float64(0.001440780353732407) - rollout_corr/log_ppl_diff_max:np.float64(0.005937933921813965) - rollout_corr/log_ppl_diff_min:np.float64(-0.00519484281539917) - rollout_corr/ppl_ratio:np.float64(1.0003440557047725) - rollout_corr/chi2_token:np.float64(0.0009851281065493822) - rollout_corr/chi2_seq:np.float64(0.45370376645587385) - actor/pg_loss:np.float64(0.02632107673707651) - 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(0.12157409638166428) - perf/mfu/actor:np.float64(0.044928533812832794) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.1626319885254) - actor/lr:np.float64(1e-05) - training/global_step:75 - training/epoch:1 - critic/score/mean:0.78125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.78125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0076843444257974625 - critic/advantages/max:0.875 - critic/advantages/min:-0.375 - critic/returns/mean:-0.0076843444257974625 - critic/returns/max:0.875 - critic/returns/min:-0.375 - response_length/mean:293.375 - response_length/max:1046.0 - response_length/min:133.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:293.375 - response_length_non_aborted/max:1046.0 - response_length_non_aborted/min:133.0 - 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:9.872205555438995e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.6137052197009325) - timing_s/agent_loop/generate_sequences/max:np.float64(8.329487580806017) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.335676260445325) - timing_s/agent_loop/too
|
|
step:76 - global_seqlen/min:17512 - global_seqlen/max:24761 - global_seqlen/minmax_diff:7249 - global_seqlen/balanced_min:20479 - global_seqlen/balanced_max:20515 - global_seqlen/mean:20486.125 - actor/entropy:0.3951570391654968 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1383596658706665 - training/rollout_probs_diff_mean:0.003291900735348463 - training/rollout_probs_diff_std:0.007475008722394705 - training/rollout_actor_probs_pearson_corr:0.9996172189712524 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.8359375 - 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:0.8359375 - rollout_corr/rollout_is_mean:0.9999387860298157 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6347383260726929 - rollout_corr/rollout_is_min:0.665208637714386 - rollout_corr/rollout_is_std:0.02640107460319996 - rollout_corr/rollout_is_eff_sample_size:0.9993032306913955 - rollout_corr/rollout_is_seq_mean:0.9999539852142334 - rollout_corr/rollout_is_seq_std:0.0016997948987409472 - rollout_corr/rollout_is_seq_max:1.0048187971115112 - rollout_corr/rollout_is_seq_min:0.9915561676025391 - rollout_corr/rollout_is_seq_max_deviation:0.008443832397460938 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5492758695036173) - rollout_corr/training_log_ppl:np.float64(0.42393943030037917) - rollout_corr/kl:np.float64(0.00042287084150505194) - rollout_corr/k3_kl:np.float64(0.00037680903727732584) - rollout_corr/rollout_ppl:np.float64(1.5485817645676434) - rollout_corr/rollout_log_ppl:np.float64(0.4235165556310676) - rollout_corr/log_ppl_diff:np.float64(0.00042287466931156814) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013275915698613971) - rollout_corr/log_ppl_diff_max:np.float64(0.009181201457977295) - rollout_corr/log_ppl_diff_min:np.float64(-0.004236400127410889) - rollout_corr/ppl_ratio:np.float64(1.0004244146402925) - rollout_corr/chi2_token:np.float64(0.0006638092454522848) - rollout_corr/chi2_seq:np.float64(0.25199120957404375) - actor/pg_loss:np.float64(0.02108556974781095) - 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(0.0969318151473999) - perf/mfu/actor:np.float64(0.04700956309888034) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.13067626953125) - actor/lr:np.float64(1e-05) - training/global_step:76 - training/epoch:1 - critic/score/mean:0.65625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.65625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0057725366204977036 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0057725366204977036 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:354.50390625 - response_length/max:1063.0 - response_length/min:123.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:354.50390625 - response_length_non_aborted/max:1063.0 - response_length_non_aborted/min:123.0 - 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:9.890086948871613e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.6719591114670038) - timing_s/agent_loop/generate_sequences/max:np.float64(8.466542003676295) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.114450175591628) - timing_s/agent_loop/tool_calls
|
|
step:77 - global_seqlen/min:16934 - global_seqlen/max:22240 - global_seqlen/minmax_diff:5306 - global_seqlen/balanced_min:19600 - global_seqlen/balanced_max:19613 - global_seqlen/mean:19610.125 - actor/entropy:0.4589277505874634 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.11700129508972168 - training/rollout_probs_diff_mean:0.003826945088803768 - training/rollout_probs_diff_std:0.007760029286146164 - training/rollout_actor_probs_pearson_corr:0.9996224045753479 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.875 - 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:0.875 - rollout_corr/rollout_is_mean:0.9998858571052551 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.3804817199707031 - rollout_corr/rollout_is_min:0.6699560284614563 - rollout_corr/rollout_is_std:0.02820431813597679 - rollout_corr/rollout_is_eff_sample_size:0.9992050296821476 - rollout_corr/rollout_is_seq_mean:0.9999222755432129 - rollout_corr/rollout_is_seq_std:0.0017754852306097746 - rollout_corr/rollout_is_seq_max:1.0049725770950317 - rollout_corr/rollout_is_seq_min:0.9942664504051208 - rollout_corr/rollout_is_seq_max_deviation:0.00573354959487915 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5996759617701173) - rollout_corr/training_log_ppl:np.float64(0.4612838347093202) - rollout_corr/kl:np.float64(0.0004860428932991212) - rollout_corr/k3_kl:np.float64(0.0004083824371150513) - rollout_corr/rollout_ppl:np.float64(1.598894499707967) - rollout_corr/rollout_log_ppl:np.float64(0.46079778991406783) - rollout_corr/log_ppl_diff:np.float64(0.00048604479525238276) - rollout_corr/log_ppl_abs_diff:np.float64(0.001427395618520677) - rollout_corr/log_ppl_diff_max:np.float64(0.006351053714752197) - rollout_corr/log_ppl_diff_min:np.float64(-0.00450509786605835) - rollout_corr/ppl_ratio:np.float64(1.0004877541214228) - rollout_corr/chi2_token:np.float64(0.0006633717566728592) - rollout_corr/chi2_seq:np.float64(0.15285813342779875) - actor/pg_loss:np.float64(0.024593278827524045) - 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(0.11761290580034256) - perf/mfu/actor:np.float64(0.04536050619670104) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.0846710205078) - actor/lr:np.float64(1e-05) - training/global_step:77 - training/epoch:1 - critic/score/mean:0.7421875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.7421875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.00541740981861949 - critic/advantages/max:0.75 - critic/advantages/min:-0.625 - critic/returns/mean:-0.00541740981861949 - critic/returns/max:0.75 - critic/returns/min:-0.625 - response_length/mean:295.72265625 - response_length/max:1024.0 - response_length/min:116.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:295.72265625 - response_length_non_aborted/max:1024.0 - response_length_non_aborted/min:116.0 - 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:0.00010614469647407532 - timing_s/agent_loop/generate_sequences/min:np.float64(1.491341369226575) - timing_s/agent_loop/generate_sequences/max:np.float64(8.11718424782157) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.497864163473423) - timing_s/agent_loop/tool_calls/min:np.float64(0
|
|
step:78 - global_seqlen/min:16202 - global_seqlen/max:23560 - global_seqlen/minmax_diff:7358 - global_seqlen/balanced_min:19437 - global_seqlen/balanced_max:19447 - global_seqlen/mean:19444.75 - actor/entropy:0.41157546639442444 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.21512961387634277 - training/rollout_probs_diff_mean:0.003451885189861059 - training/rollout_probs_diff_std:0.007595372851938009 - training/rollout_actor_probs_pearson_corr:0.9995942711830139 - self_distillation/success_group_fraction:0.96875 - self_distillation/success_sample_fraction:0.9609375 - self_distillation/correct_sample_fraction:0.76953125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.9609375 - rollout_corr/rollout_is_mean:0.9999512434005737 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.382772445678711 - rollout_corr/rollout_is_min:0.5737053751945496 - rollout_corr/rollout_is_std:0.02684214897453785 - rollout_corr/rollout_is_eff_sample_size:0.9992798987641299 - rollout_corr/rollout_is_seq_mean:0.999904990196228 - rollout_corr/rollout_is_seq_std:0.0016190879978239536 - rollout_corr/rollout_is_seq_max:1.0048513412475586 - rollout_corr/rollout_is_seq_min:0.9938386678695679 - rollout_corr/rollout_is_seq_max_deviation:0.006161332130432129 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5593909919261932) - rollout_corr/training_log_ppl:np.float64(0.43316946757840924) - rollout_corr/kl:np.float64(0.0004828531190828045) - rollout_corr/k3_kl:np.float64(0.00038787729332057097) - rollout_corr/rollout_ppl:np.float64(1.5585966338403523) - rollout_corr/rollout_log_ppl:np.float64(0.4326866158808116) - rollout_corr/log_ppl_diff:np.float64(0.0004828516975976527) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012700915685854852) - rollout_corr/log_ppl_diff_max:np.float64(0.006852507591247559) - rollout_corr/log_ppl_diff_min:np.float64(-0.004224538803100586) - rollout_corr/ppl_ratio:np.float64(1.000484290998429) - rollout_corr/chi2_token:np.float64(0.0005866813007742167) - rollout_corr/chi2_seq:np.float64(0.13113136892206967) - actor/pg_loss:np.float64(0.026065922198540648) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.0390625) - actor/grad_norm:np.float64(0.11347637325525284) - perf/mfu/actor:np.float64(0.044139462726232155) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.10301208496094) - actor/lr:np.float64(1e-05) - training/global_step:78 - training/epoch:1 - critic/score/mean:0.76953125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.76953125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.017423205077648163 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.017423205077648163 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:328.0859375 - response_length/max:1065.0 - response_length/min:115.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:328.0859375 - response_length_non_aborted/max:1065.0 - response_length_non_aborted/min:115.0 - 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:9.630806744098663e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.5468091703951359) - timing_s/agent_loop/generate_sequences/max:np.float64(8.492994800209999) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.833776883177052) - timing_s/agent_loop/tool_
|
|
step:79 - global_seqlen/min:17469 - global_seqlen/max:21609 - global_seqlen/minmax_diff:4140 - global_seqlen/balanced_min:19289 - global_seqlen/balanced_max:19293 - global_seqlen/mean:19290.75 - actor/entropy:0.393234521150589 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.22149217128753662 - training/rollout_probs_diff_mean:0.0033473705407232046 - training/rollout_probs_diff_std:0.007645653560757637 - training/rollout_actor_probs_pearson_corr:0.9996024966239929 - 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:1.0001161098480225 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.1524189176270738e-05 - rollout_corr/rollout_is_max:1.6161819696426392 - rollout_corr/rollout_is_min:0.25997433066368103 - rollout_corr/rollout_is_std:0.0269617922604084 - rollout_corr/rollout_is_eff_sample_size:0.9992739469162876 - rollout_corr/rollout_is_seq_mean:1.0001323223114014 - rollout_corr/rollout_is_seq_std:0.0017513000639155507 - rollout_corr/rollout_is_seq_max:1.0070801973342896 - rollout_corr/rollout_is_seq_min:0.9951545596122742 - rollout_corr/rollout_is_seq_max_deviation:0.007080197334289551 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5293668895028532) - rollout_corr/training_log_ppl:np.float64(0.41441104281693697) - rollout_corr/kl:np.float64(0.00025625684655716796) - rollout_corr/k3_kl:np.float64(0.0003886133479227283) - rollout_corr/rollout_ppl:np.float64(1.528971301857382) - rollout_corr/rollout_log_ppl:np.float64(0.4141547832987271) - rollout_corr/log_ppl_diff:np.float64(0.00025625951820984483) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013637715601362288) - rollout_corr/log_ppl_diff_max:np.float64(0.005415737628936768) - rollout_corr/log_ppl_diff_min:np.float64(-0.0064498186111450195) - rollout_corr/ppl_ratio:np.float64(1.0002578359562904) - rollout_corr/chi2_token:np.float64(0.0010421324986964464) - rollout_corr/chi2_seq:np.float64(0.6562901881989092) - actor/pg_loss:np.float64(0.0217051109757449) - 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(0.09964113682508469) - perf/mfu/actor:np.float64(0.04482786116005441) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.08424758911133) - actor/lr:np.float64(1e-05) - training/global_step:79 - 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:0.014336091466248035 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.014336091466248035 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:338.9609375 - response_length/max:937.0 - response_length/min:127.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:338.9609375 - response_length_non_aborted/max:937.0 - response_length_non_aborted/min:127.0 - 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:0.00016025826334953308 - timing_s/agent_loop/generate_sequences/min:np.float64(1.028574887663126) - timing_s/agent_loop/generate_sequences/max:np.float64(7.930324262008071) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.5301734151435085) - 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': 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>
|
|
To determine the correct reactant for the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1", we need to compare the structure of each option with the target product.
|
|
|
|
The product contains:
|
|
- A nitrogen-containing heterocyclic ring (likely a pyrrolidine or similar).
|
|
- A chlorinated aromatic ring (with a Cl atom).
|
|
- A fluorinated aromatic ring (with a C(F)(F)F group).
|
|
- A carbon chain connecting the nitrogen ring to the fluorinated aromatic ring.
|
|
|
|
Let's evaluate the options:
|
|
|
|
- **Option A**: "O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1"
|
|
This structure contains a chlorinated aromatic ring (with Cl), a fluorinated aromatic ring (C(F)(F)F), and a nitrogen-containing heterocycle with the correct connectivity. It matches the product structure closely.
|
|
|
|
- **Option B**: "O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1"
|
|
This has a different arrangement of the chlorine atoms and does not contain the correct fluorinated aromatic ring. It does not match the target structure.
|
|
|
|
- **Option C**: "O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1"
|
|
This includes an oxygen-containing group (hydroxyl) that is not present in the product, and the fluorinated ring is not directly attached to the nitrogen ring as required.
|
|
|
|
- **Option D**: "O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F"
|
|
This lacks the correct chlorinated aromatic ring and has a different carbon chain structure. It does not match the target.
|
|
|
|
Therefore, **Option A** is the only correct reactant that matches the structure and connectivity of the product.
|
|
</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-Qwen3-8B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_80
|
|
step:80 - global_seqlen/min:17404 - global_seqlen/max:22819 - global_seqlen/minmax_diff:5415 - global_seqlen/balanced_min:20630 - global_seqlen/balanced_max:20636 - global_seqlen/mean:20633.625 - actor/entropy:0.4224976897239685 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1190071702003479 - training/rollout_probs_diff_mean:0.00357939675450325 - training/rollout_probs_diff_std:0.00770284328609705 - training/rollout_actor_probs_pearson_corr:0.9995966553688049 - self_distillation/success_group_fraction:0.96875 - self_distillation/success_sample_fraction:0.9609375 - self_distillation/correct_sample_fraction:0.76953125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.9609375 - rollout_corr/rollout_is_mean:1.0000313520431519 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6125468015670776 - rollout_corr/rollout_is_min:0.5972538590431213 - rollout_corr/rollout_is_std:0.027621358633041382 - rollout_corr/rollout_is_eff_sample_size:0.9992376421797334 - rollout_corr/rollout_is_seq_mean:1.0000790357589722 - rollout_corr/rollout_is_seq_std:0.0017634446267038584 - rollout_corr/rollout_is_seq_max:1.0044612884521484 - rollout_corr/rollout_is_seq_min:0.9956191182136536 - rollout_corr/rollout_is_seq_max_deviation:0.0044612884521484375 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5759053975343704) - rollout_corr/training_log_ppl:np.float64(0.44484703088528477) - rollout_corr/kl:np.float64(0.0003184272226661733) - rollout_corr/k3_kl:np.float64(0.00039754411551484736) - rollout_corr/rollout_ppl:np.float64(1.575384461786598) - rollout_corr/rollout_log_ppl:np.float64(0.4445286004047375) - rollout_corr/log_ppl_diff:np.float64(0.0003184304805472493) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014140368439257145) - rollout_corr/log_ppl_diff_max:np.float64(0.004790067672729492) - rollout_corr/log_ppl_diff_min:np.float64(-0.004132270812988281) - rollout_corr/ppl_ratio:np.float64(1.000320031773299) - rollout_corr/chi2_token:np.float64(0.0009581439662724733) - rollout_corr/chi2_seq:np.float64(0.37358179083094) - actor/pg_loss:np.float64(0.026770716220198665) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.0390625) - actor/grad_norm:np.float64(0.11496338993310928) - perf/mfu/actor:np.float64(0.04721151056967112) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.05580139160156) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7375) - val-aux/sciknoweval/reward/std@16:np.float64(0.14370285906368757) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.7983809523809524) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.13396467740621712) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6769904761904763) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.10916077551731444) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7376428571428572) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.14327878534013916) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.8532809523809525) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.10677235385341363) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6347999999999999) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.06823870032970898) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7432809523809524) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.12265749370103732) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.8961666666666667) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.0736492300341428) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6101142857142857) - val-aux/sciknoweval/reward/worst@8/std:n
|
|
step:81 - global_seqlen/min:16951 - global_seqlen/max:24477 - global_seqlen/minmax_diff:7526 - global_seqlen/balanced_min:20852 - global_seqlen/balanced_max:20988 - global_seqlen/mean:20873.5 - actor/entropy:0.3995857536792755 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.2773001194000244 - training/rollout_probs_diff_mean:0.0035114840138703585 - training/rollout_probs_diff_std:0.007972010411322117 - training/rollout_actor_probs_pearson_corr:0.9995556473731995 - self_distillation/success_group_fraction:0.96875 - self_distillation/success_sample_fraction:0.953125 - 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:0.953125 - rollout_corr/rollout_is_mean:1.0000276565551758 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5546014308929443 - rollout_corr/rollout_is_min:0.5128262042999268 - rollout_corr/rollout_is_std:0.02770969271659851 - rollout_corr/rollout_is_eff_sample_size:0.9992328810984762 - rollout_corr/rollout_is_seq_mean:1.000017523765564 - rollout_corr/rollout_is_seq_std:0.0017161129508167505 - rollout_corr/rollout_is_seq_max:1.0067123174667358 - rollout_corr/rollout_is_seq_min:0.9945694804191589 - rollout_corr/rollout_is_seq_max_deviation:0.00671231746673584 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5391268385574222) - rollout_corr/training_log_ppl:np.float64(0.4214484076946974) - rollout_corr/kl:np.float64(0.0003985792236473884) - rollout_corr/k3_kl:np.float64(0.00041610692892390944) - rollout_corr/rollout_ppl:np.float64(1.5384613214991987) - rollout_corr/rollout_log_ppl:np.float64(0.4210498257307336) - rollout_corr/log_ppl_diff:np.float64(0.00039858196396380663) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013196455547586083) - rollout_corr/log_ppl_diff_max:np.float64(0.005821883678436279) - rollout_corr/log_ppl_diff_min:np.float64(-0.0059986114501953125) - rollout_corr/ppl_ratio:np.float64(1.0004001329652965) - rollout_corr/chi2_token:np.float64(0.0008705053478479385) - rollout_corr/chi2_seq:np.float64(0.313229059567675) - actor/pg_loss:np.float64(0.025826958941252087) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.046875) - actor/grad_norm:np.float64(0.12749090790748596) - perf/mfu/actor:np.float64(0.04774606383762947) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(199.4612693786621) - actor/lr:np.float64(1e-05) - training/global_step:81 - 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:0.0015914811519905925 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.0015914811519905925 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:344.546875 - response_length/max:1271.0 - response_length/min:130.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:344.546875 - response_length_non_aborted/max:1271.0 - response_length_non_aborted/min:130.0 - 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:0.00015782006084918976 - timing_s/agent_loop/generate_sequences/min:np.float64(1.6993379052728415) - timing_s/agent_loop/generate_sequences/max:np.float64(10.211031097918749) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.096621022385079) - timing_s/agent_loop/tool_calls/min:np.floa
|
|
step:82 - global_seqlen/min:18493 - global_seqlen/max:23185 - global_seqlen/minmax_diff:4692 - global_seqlen/balanced_min:20621 - global_seqlen/balanced_max:20670 - global_seqlen/mean:20637.125 - actor/entropy:0.37220197916030884 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12834173440933228 - training/rollout_probs_diff_mean:0.003158497391268611 - training/rollout_probs_diff_std:0.007322004064917564 - training/rollout_actor_probs_pearson_corr:0.9996234178543091 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.9296875 - 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:0.9296875 - rollout_corr/rollout_is_mean:1.0000524520874023 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5408705472946167 - rollout_corr/rollout_is_min:0.7248677015304565 - rollout_corr/rollout_is_std:0.02568238601088524 - rollout_corr/rollout_is_eff_sample_size:0.9993408497665217 - rollout_corr/rollout_is_seq_mean:1.000057339668274 - rollout_corr/rollout_is_seq_std:0.0015503127360716462 - rollout_corr/rollout_is_seq_max:1.0064513683319092 - rollout_corr/rollout_is_seq_min:0.9959645867347717 - rollout_corr/rollout_is_seq_max_deviation:0.00645136833190918 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4953496786765754) - rollout_corr/training_log_ppl:np.float64(0.39294266584329307) - rollout_corr/kl:np.float64(0.000294816790257002) - rollout_corr/k3_kl:np.float64(0.00035218617600207835) - rollout_corr/rollout_ppl:np.float64(1.4948710771277547) - rollout_corr/rollout_log_ppl:np.float64(0.3926478484645486) - rollout_corr/log_ppl_diff:np.float64(0.000294817378744483) - rollout_corr/log_ppl_abs_diff:np.float64(0.001223124680109322) - rollout_corr/log_ppl_diff_max:np.float64(0.004598289728164673) - rollout_corr/log_ppl_diff_min:np.float64(-0.005899399518966675) - rollout_corr/ppl_ratio:np.float64(1.0002960797864944) - rollout_corr/chi2_token:np.float64(0.0008225198835134506) - rollout_corr/chi2_seq:np.float64(0.3081965073943138) - actor/pg_loss:np.float64(0.023082646808688878) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.0703125) - actor/grad_norm:np.float64(0.10587161779403687) - perf/mfu/actor:np.float64(0.04735461194266269) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(199.1742706298828) - actor/lr:np.float64(1e-05) - training/global_step:82 - training/epoch:1 - critic/score/mean:0.6875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.0051129041239619255 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0051129041239619255 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:378.84765625 - response_length/max:1270.0 - response_length/min:144.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:378.84765625 - response_length_non_aborted/max:1270.0 - response_length_non_aborted/min:144.0 - 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:0.00016219168901443481 - timing_s/agent_loop/generate_sequences/min:np.float64(1.92292376793921) - timing_s/agent_loop/generate_sequences/max:np.float64(10.040248956531286) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.438673419092083) - timing_s/agent_loop/tool_calls/min:np.fl
|
|
step:83 - global_seqlen/min:16710 - global_seqlen/max:21349 - global_seqlen/minmax_diff:4639 - global_seqlen/balanced_min:19040 - global_seqlen/balanced_max:19049 - global_seqlen/mean:19044.875 - actor/entropy:0.40517759323120117 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.13839542865753174 - training/rollout_probs_diff_mean:0.003609153674915433 - training/rollout_probs_diff_std:0.007879640907049179 - training/rollout_actor_probs_pearson_corr:0.999563992023468 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83984375 - 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:0.83984375 - rollout_corr/rollout_is_mean:1.0000033378601074 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.758644700050354 - rollout_corr/rollout_is_min:0.6626258492469788 - rollout_corr/rollout_is_std:0.027495915070176125 - rollout_corr/rollout_is_eff_sample_size:0.9992444267989176 - rollout_corr/rollout_is_seq_mean:1.0000196695327759 - rollout_corr/rollout_is_seq_std:0.0017711295513436198 - rollout_corr/rollout_is_seq_max:1.004800796508789 - rollout_corr/rollout_is_seq_min:0.9950864315032959 - rollout_corr/rollout_is_seq_max_deviation:0.0049135684967041016 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5294637000188231) - rollout_corr/training_log_ppl:np.float64(0.41653524636058137) - rollout_corr/kl:np.float64(0.00037080937508199296) - rollout_corr/k3_kl:np.float64(0.0003903741740032274) - rollout_corr/rollout_ppl:np.float64(1.528886781539768) - rollout_corr/rollout_log_ppl:np.float64(0.41616443532984704) - rollout_corr/log_ppl_diff:np.float64(0.0003708110307343304) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014502297271974385) - rollout_corr/log_ppl_diff_max:np.float64(0.005322396755218506) - rollout_corr/log_ppl_diff_min:np.float64(-0.004239082336425781) - rollout_corr/ppl_ratio:np.float64(1.000372430542484) - rollout_corr/chi2_token:np.float64(0.0008249275851994753) - rollout_corr/chi2_seq:np.float64(0.24598133168183267) - actor/pg_loss:np.float64(0.022324581877910532) - 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(0.1188177689909935) - perf/mfu/actor:np.float64(0.044192956314555706) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.59613800048828) - actor/lr:np.float64(1e-05) - training/global_step:83 - training/epoch:1 - 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:-0.007243964355438948 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.007243964355438948 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:296.24609375 - response_length/max:871.0 - response_length/min:109.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:296.24609375 - response_length_non_aborted/max:871.0 - response_length_non_aborted/min:109.0 - 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:0.00010315701365470886 - timing_s/agent_loop/generate_sequences/min:np.float64(1.489199798554182) - timing_s/agent_loop/generate_sequences/max:np.float64(7.379217518493533) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.5365343318117084) - timing_s/agent_loop
|
|
step:84 - global_seqlen/min:14701 - global_seqlen/max:20350 - global_seqlen/minmax_diff:5649 - global_seqlen/balanced_min:18214 - global_seqlen/balanced_max:18220 - global_seqlen/mean:18217.625 - actor/entropy:0.3717782497406006 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.16888469457626343 - training/rollout_probs_diff_mean:0.0033551182132214308 - training/rollout_probs_diff_std:0.007667208556085825 - training/rollout_actor_probs_pearson_corr:0.9995649456977844 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.92578125 - 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:0.92578125 - rollout_corr/rollout_is_mean:1.0000234842300415 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:2.478038368280977e-05 - rollout_corr/rollout_is_max:1.4499679803848267 - rollout_corr/rollout_is_min:0.3131377696990967 - rollout_corr/rollout_is_std:0.026493476703763008 - rollout_corr/rollout_is_eff_sample_size:0.9992988261114965 - rollout_corr/rollout_is_seq_mean:1.0001063346862793 - rollout_corr/rollout_is_seq_std:0.001685771974734962 - rollout_corr/rollout_is_seq_max:1.0058950185775757 - rollout_corr/rollout_is_seq_min:0.9954847693443298 - rollout_corr/rollout_is_seq_max_deviation:0.005895018577575684 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4890748248435557) - rollout_corr/training_log_ppl:np.float64(0.39051099558128044) - rollout_corr/kl:np.float64(0.0002647332206819897) - rollout_corr/k3_kl:np.float64(0.0003711414171903016) - rollout_corr/rollout_ppl:np.float64(1.4887068322859704) - rollout_corr/rollout_log_ppl:np.float64(0.39024625689489767) - rollout_corr/log_ppl_diff:np.float64(0.00026473868638277054) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013162189861759543) - rollout_corr/log_ppl_diff_max:np.float64(0.0052817463874816895) - rollout_corr/log_ppl_diff_min:np.float64(-0.005239009857177734) - rollout_corr/ppl_ratio:np.float64(1.0002661920152605) - rollout_corr/chi2_token:np.float64(0.000953644048422575) - rollout_corr/chi2_seq:np.float64(0.28764038090594113) - actor/pg_loss:np.float64(0.023024803875159705) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.07421875) - actor/grad_norm:np.float64(0.11200632899999619) - perf/mfu/actor:np.float64(0.04198279864447934) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.36781311035156) - actor/lr:np.float64(1e-05) - training/global_step:84 - training/epoch:1 - 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:-0.003766618436202407 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.003766618436202407 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:315.26953125 - response_length/max:876.0 - response_length/min:135.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:315.26953125 - response_length_non_aborted/max:876.0 - response_length_non_aborted/min:135.0 - 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:5.776435136795044e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.860195454210043) - timing_s/agent_loop/generate_sequences/max:np.float64(7.453522594645619) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.8417371912146336) -
|
|
step:85 - global_seqlen/min:16425 - global_seqlen/max:22906 - global_seqlen/minmax_diff:6481 - global_seqlen/balanced_min:19482 - global_seqlen/balanced_max:19489 - global_seqlen/mean:19487.25 - actor/entropy:0.4401249885559082 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12588274478912354 - training/rollout_probs_diff_mean:0.003797614248469472 - training/rollout_probs_diff_std:0.007930994965136051 - training/rollout_actor_probs_pearson_corr:0.9995816946029663 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.8984375 - 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:0.8984375 - rollout_corr/rollout_is_mean:0.9997216463088989 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.3540738821029663 - rollout_corr/rollout_is_min:0.7105928659439087 - rollout_corr/rollout_is_std:0.027824534103274345 - rollout_corr/rollout_is_eff_sample_size:0.9992258585864361 - rollout_corr/rollout_is_seq_mean:0.9997164011001587 - rollout_corr/rollout_is_seq_std:0.001581866294145584 - rollout_corr/rollout_is_seq_max:1.0051721334457397 - rollout_corr/rollout_is_seq_min:0.9948992133140564 - rollout_corr/rollout_is_seq_max_deviation:0.005172133445739746 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.584278158377856) - rollout_corr/training_log_ppl:np.float64(0.4499193246010691) - rollout_corr/kl:np.float64(0.0006890382838964371) - rollout_corr/k3_kl:np.float64(0.0004054484571156536) - rollout_corr/rollout_ppl:np.float64(1.5831760363653302) - rollout_corr/rollout_log_ppl:np.float64(0.44923028588527814) - rollout_corr/log_ppl_diff:np.float64(0.0006890387157909572) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013385045458562672) - rollout_corr/log_ppl_diff_max:np.float64(0.0054593682289123535) - rollout_corr/log_ppl_diff_min:np.float64(-0.004691362380981445) - rollout_corr/ppl_ratio:np.float64(1.0006905470509082) - rollout_corr/chi2_token:np.float64(0.00024570291861891747) - rollout_corr/chi2_seq:np.float64(-0.027990202652290463) - actor/pg_loss:np.float64(0.026418285742693115) - 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(0.12258151173591614) - perf/mfu/actor:np.float64(0.04494003714395457) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.34210062026978) - actor/lr:np.float64(1e-05) - training/global_step:85 - training/epoch:1 - critic/score/mean:0.76171875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.76171875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.002087913453578949 - critic/advantages/max:0.875 - critic/advantages/min:-0.625 - critic/returns/mean:0.002087913453578949 - critic/returns/max:0.875 - critic/returns/min:-0.625 - response_length/mean:309.1640625 - response_length/max:814.0 - response_length/min:113.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:309.1640625 - response_length_non_aborted/max:814.0 - response_length_non_aborted/min:113.0 - 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:0.00014099851250648499 - timing_s/agent_loop/generate_sequences/min:np.float64(1.5464082807302475) - timing_s/agent_loop/generate_sequences/max:np.float64(7.190803783014417) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.681909154896857) - timing_s/agent_loop/tool_
|
|
step:86 - global_seqlen/min:17737 - global_seqlen/max:20999 - global_seqlen/minmax_diff:3262 - global_seqlen/balanced_min:19085 - global_seqlen/balanced_max:19092 - global_seqlen/mean:19089.5 - actor/entropy:0.37723323702812195 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.23082247376441956 - training/rollout_probs_diff_mean:0.003395188134163618 - training/rollout_probs_diff_std:0.007731357589364052 - training/rollout_actor_probs_pearson_corr:0.9996012449264526 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.921875 - self_distillation/correct_sample_fraction:0.66015625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.921875 - rollout_corr/rollout_is_mean:0.9999049305915833 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.5170440673828125 - rollout_corr/rollout_is_min:0.5380076766014099 - rollout_corr/rollout_is_std:0.026578830555081367 - rollout_corr/rollout_is_eff_sample_size:0.9992938263651499 - rollout_corr/rollout_is_seq_mean:0.9998374581336975 - rollout_corr/rollout_is_seq_std:0.0016897612949833274 - rollout_corr/rollout_is_seq_max:1.004929542541504 - rollout_corr/rollout_is_seq_min:0.9923271536827087 - rollout_corr/rollout_is_seq_max_deviation:0.00767284631729126 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.498649399727583) - rollout_corr/training_log_ppl:np.float64(0.3945946304593235) - rollout_corr/kl:np.float64(0.0005340524880228514) - rollout_corr/k3_kl:np.float64(0.00037150386150131) - rollout_corr/rollout_ppl:np.float64(1.497807027772069) - rollout_corr/rollout_log_ppl:np.float64(0.39406057534506544) - rollout_corr/log_ppl_diff:np.float64(0.0005340551142580807) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013611234608106315) - rollout_corr/log_ppl_diff_max:np.float64(0.008284002542495728) - rollout_corr/log_ppl_diff_min:np.float64(-0.004299581050872803) - rollout_corr/ppl_ratio:np.float64(1.0005356636829674) - rollout_corr/chi2_token:np.float64(0.000420228810980916) - rollout_corr/chi2_seq:np.float64(0.220992537913844) - actor/pg_loss:np.float64(0.02341267381416401) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.078125) - actor/grad_norm:np.float64(0.1145884171128273) - perf/mfu/actor:np.float64(0.044133150812278484) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.2140350341797) - actor/lr:np.float64(1e-05) - training/global_step:86 - training/epoch:1 - critic/score/mean:0.66015625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.66015625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.005195329897105694 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.005195329897105694 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:319.171875 - response_length/max:827.0 - response_length/min:128.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:319.171875 - response_length_non_aborted/max:827.0 - response_length_non_aborted/min:128.0 - 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:0.0001481175422668457 - timing_s/agent_loop/generate_sequences/min:np.float64(1.4202340003103018) - timing_s/agent_loop/generate_sequences/max:np.float64(7.130688466131687) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.720032443823584) - timing_s/agent_loop/tool_calls/min:np.float6
|
|
step:87 - global_seqlen/min:16907 - global_seqlen/max:21769 - global_seqlen/minmax_diff:4862 - global_seqlen/balanced_min:18868 - global_seqlen/balanced_max:18873 - global_seqlen/mean:18870.875 - actor/entropy:0.3849543035030365 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12734666466712952 - training/rollout_probs_diff_mean:0.0034306973684579134 - training/rollout_probs_diff_std:0.007617479655891657 - training/rollout_actor_probs_pearson_corr:0.999609112739563 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.9375 - self_distillation/correct_sample_fraction:0.78125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.9375 - rollout_corr/rollout_is_mean:1.0000180006027222 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:1.3275452147354372e-05 - rollout_corr/rollout_is_max:1.4114339351654053 - rollout_corr/rollout_is_min:0.4281553328037262 - rollout_corr/rollout_is_std:0.026802148669958115 - rollout_corr/rollout_is_eff_sample_size:0.9992822795231191 - rollout_corr/rollout_is_seq_mean:0.999997615814209 - rollout_corr/rollout_is_seq_std:0.001799858990125358 - rollout_corr/rollout_is_seq_max:1.0063596963882446 - rollout_corr/rollout_is_seq_min:0.99202561378479 - rollout_corr/rollout_is_seq_max_deviation:0.007974386215209961 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5035898280330002) - rollout_corr/training_log_ppl:np.float64(0.3992501661123242) - rollout_corr/kl:np.float64(0.00037876126474856164) - rollout_corr/k3_kl:np.float64(0.0003764367431813298) - rollout_corr/rollout_ppl:np.float64(1.5030373092740774) - rollout_corr/rollout_log_ppl:np.float64(0.3988714033039287) - rollout_corr/log_ppl_diff:np.float64(0.00037876280839554965) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013777429412584752) - rollout_corr/log_ppl_diff_max:np.float64(0.00855901837348938) - rollout_corr/log_ppl_diff_min:np.float64(-0.005897998809814453) - rollout_corr/ppl_ratio:np.float64(1.0003804499283433) - rollout_corr/chi2_token:np.float64(0.000750402919948101) - rollout_corr/chi2_seq:np.float64(0.2456935232039541) - actor/pg_loss:np.float64(0.023766542466546525) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.0625) - actor/grad_norm:np.float64(0.1240430623292923) - perf/mfu/actor:np.float64(0.04362549643295263) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.30377197265625) - actor/lr:np.float64(1e-05) - training/global_step:87 - training/epoch:1 - critic/score/mean:0.78125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.78125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.008876299485564232 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:0.008876299485564232 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:294.24609375 - response_length/max:839.0 - response_length/min:111.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:294.24609375 - response_length_non_aborted/max:839.0 - response_length_non_aborted/min:111.0 - 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:0.000174839049577713 - timing_s/agent_loop/generate_sequences/min:np.float64(1.040698753669858) - timing_s/agent_loop/generate_sequences/max:np.float64(7.230389727279544) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.3120938681531698) - timing_s/agent_loop/tool_calls/min
|
|
step:88 - global_seqlen/min:16263 - global_seqlen/max:19926 - global_seqlen/minmax_diff:3663 - global_seqlen/balanced_min:18094 - global_seqlen/balanced_max:18099 - global_seqlen/mean:18098.125 - actor/entropy:0.4026566445827484 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.13445866107940674 - training/rollout_probs_diff_mean:0.0035605633165687323 - training/rollout_probs_diff_std:0.0077731190249323845 - training/rollout_actor_probs_pearson_corr:0.9995751976966858 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.90234375 - self_distillation/correct_sample_fraction:0.75 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.90234375 - rollout_corr/rollout_is_mean:0.9999142289161682 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.3945938348770142 - rollout_corr/rollout_is_min:0.6247751712799072 - rollout_corr/rollout_is_std:0.02713148295879364 - rollout_corr/rollout_is_eff_sample_size:0.9992641860393089 - rollout_corr/rollout_is_seq_mean:0.999900758266449 - rollout_corr/rollout_is_seq_std:0.0017919992096722126 - rollout_corr/rollout_is_seq_max:1.0064406394958496 - rollout_corr/rollout_is_seq_min:0.9920283555984497 - rollout_corr/rollout_is_seq_max_deviation:0.007971644401550293 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5180707965046167) - rollout_corr/training_log_ppl:np.float64(0.4069743760046549) - rollout_corr/kl:np.float64(0.0004803496275656016) - rollout_corr/k3_kl:np.float64(0.00038113498456482375) - rollout_corr/rollout_ppl:np.float64(1.5172541765496135) - rollout_corr/rollout_log_ppl:np.float64(0.40649402409326285) - rollout_corr/log_ppl_diff:np.float64(0.0004803519113920629) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013665432925336063) - rollout_corr/log_ppl_diff_max:np.float64(0.008656024932861328) - rollout_corr/log_ppl_diff_min:np.float64(-0.005883395671844482) - rollout_corr/ppl_ratio:np.float64(1.0004821170587093) - rollout_corr/chi2_token:np.float64(0.0005676755681633949) - rollout_corr/chi2_seq:np.float64(0.2910463814623654) - actor/pg_loss:np.float64(0.024544738887925632) - 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(0.12066686898469925) - perf/mfu/actor:np.float64(0.0421833356929109) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.16246032714844) - actor/lr:np.float64(1e-05) - training/global_step:88 - training/epoch:1 - critic/score/mean:0.75 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.75 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0052373334765434265 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0052373334765434265 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:296.75390625 - response_length/max:777.0 - response_length/min:132.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:296.75390625 - response_length_non_aborted/max:777.0 - response_length_non_aborted/min:132.0 - 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:0.00012115761637687683 - timing_s/agent_loop/generate_sequences/min:np.float64(1.790180156007409) - timing_s/agent_loop/generate_sequences/max:np.float64(6.706206066533923) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.5888049019122263) - timing_s/agent_loop/tool_calls/min:n
|
|
step:89 - global_seqlen/min:17367 - global_seqlen/max:24336 - global_seqlen/minmax_diff:6969 - global_seqlen/balanced_min:20219 - global_seqlen/balanced_max:20232 - global_seqlen/mean:20229.0 - actor/entropy:0.3856755793094635 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.19635042548179626 - training/rollout_probs_diff_mean:0.0033116349950432777 - training/rollout_probs_diff_std:0.0075814551673829556 - training/rollout_actor_probs_pearson_corr:0.9996069073677063 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.92578125 - 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:0.92578125 - rollout_corr/rollout_is_mean:1.0000357627868652 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6833916902542114 - rollout_corr/rollout_is_min:0.5454350113868713 - rollout_corr/rollout_is_std:0.026791026815772057 - rollout_corr/rollout_is_eff_sample_size:0.9992827556762782 - rollout_corr/rollout_is_seq_mean:1.0000516176223755 - rollout_corr/rollout_is_seq_std:0.0017000410007312894 - rollout_corr/rollout_is_seq_max:1.0048518180847168 - rollout_corr/rollout_is_seq_min:0.9942007064819336 - rollout_corr/rollout_is_seq_max_deviation:0.005799293518066406 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5469604074023664) - rollout_corr/training_log_ppl:np.float64(0.42298222141107544) - rollout_corr/kl:np.float64(0.0003418542377886524) - rollout_corr/k3_kl:np.float64(0.0003934762273445358) - rollout_corr/rollout_ppl:np.float64(1.5464338595047593) - rollout_corr/rollout_log_ppl:np.float64(0.42264036287087947) - rollout_corr/log_ppl_diff:np.float64(0.00034185854019597173) - rollout_corr/log_ppl_abs_diff:np.float64(0.001321688701864332) - rollout_corr/log_ppl_diff_max:np.float64(0.006345689296722412) - rollout_corr/log_ppl_diff_min:np.float64(-0.004188179969787598) - rollout_corr/ppl_ratio:np.float64(1.0003433658275753) - rollout_corr/chi2_token:np.float64(0.0008899655658751726) - rollout_corr/chi2_seq:np.float64(0.21638701064512134) - actor/pg_loss:np.float64(0.024673637155501638) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.07421875) - actor/grad_norm:np.float64(0.11104342341423035) - perf/mfu/actor:np.float64(0.04657847873187733) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.2346534729004) - actor/lr:np.float64(1e-05) - training/global_step:89 - 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:0.0015100633027032018 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0015100633027032018 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:340.8125 - response_length/max:1225.0 - response_length/min:132.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:340.8125 - response_length_non_aborted/max:1225.0 - response_length_non_aborted/min:132.0 - 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:7.351301610469818e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.8044025283306837) - timing_s/agent_loop/generate_sequences/max:np.float64(9.335278917104006) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.016844092337124) - timing_s/agent_loop/tool_calls/m
|
|
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>
|
|
To determine the correct reactant for the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1", we must compare the structural features of the product with the options provided.
|
|
|
|
The product has the following key features:
|
|
- A nitrogen-containing ring (likely a pyrrole or similar structure).
|
|
- A chlorinated aromatic ring (Cc2ccc(Cl)cc2).
|
|
- A fluorinated aromatic ring (C(F)(F)F) attached to the chlorinated ring.
|
|
|
|
Now, analyzing the options:
|
|
|
|
- **Option A**: "O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1"
|
|
This option contains the correct chlorinated aromatic ring (Cc2ccc(Cl)cc2), the fluorinated aromatic ring (C(F)(F)F), and the nitrogen-containing ring with the correct substitution pattern. It also includes the necessary functional groups for the synthesis.
|
|
|
|
- **Option B**: "O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1"
|
|
This option has a chlorinated ring but the fluorinated ring is not present, and the nitrogen substitution pattern does not match the target product.
|
|
|
|
- **Option C**: "O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1"
|
|
This option introduces an oxygen atom and a different arrangement of the fluorinated and chlorinated rings, which does not match the product's structure.
|
|
|
|
- **Option D**: "O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F"
|
|
This option has a fluorinated group but lacks the chlorinated aromatic ring and the correct nitrogen substitution pattern.
|
|
|
|
Thus, **Option A** is the only reactant that matches the structure of the product.
|
|
</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-Qwen3-8B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_90
|
|
step:90 - global_seqlen/min:17993 - global_seqlen/max:24534 - global_seqlen/minmax_diff:6541 - global_seqlen/balanced_min:20530 - global_seqlen/balanced_max:20543 - global_seqlen/mean:20540.125 - actor/entropy:0.35813578963279724 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.24059292674064636 - training/rollout_probs_diff_mean:0.0031041023321449757 - training/rollout_probs_diff_std:0.00729778315871954 - training/rollout_actor_probs_pearson_corr:0.999591052532196 - self_distillation/success_group_fraction:0.96875 - self_distillation/success_sample_fraction:0.95703125 - 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:0.95703125 - rollout_corr/rollout_is_mean:0.999988853931427 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.751498818397522 - rollout_corr/rollout_is_min:0.5970545411109924 - rollout_corr/rollout_is_std:0.025474995374679565 - rollout_corr/rollout_is_eff_sample_size:0.9993513264688703 - rollout_corr/rollout_is_seq_mean:0.9999680519104004 - rollout_corr/rollout_is_seq_std:0.0015169174876064062 - rollout_corr/rollout_is_seq_max:1.0037466287612915 - rollout_corr/rollout_is_seq_min:0.9956980347633362 - rollout_corr/rollout_is_seq_max_deviation:0.004301965236663818 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4888556944206357) - rollout_corr/training_log_ppl:np.float64(0.38689433032413945) - rollout_corr/kl:np.float64(0.0003815739005741392) - rollout_corr/k3_kl:np.float64(0.0003495998294340552) - rollout_corr/rollout_ppl:np.float64(1.4882625960744917) - rollout_corr/rollout_log_ppl:np.float64(0.3865127532335464) - rollout_corr/log_ppl_diff:np.float64(0.0003815770905930549) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012148020032327622) - rollout_corr/log_ppl_diff_max:np.float64(0.004585593938827515) - rollout_corr/log_ppl_diff_min:np.float64(-0.0033484697341918945) - rollout_corr/ppl_ratio:np.float64(1.0003828115295619) - rollout_corr/chi2_token:np.float64(0.0006405133754014969) - rollout_corr/chi2_seq:np.float64(0.20064060017466545) - actor/pg_loss:np.float64(0.022427467538364) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.04296875) - actor/grad_norm:np.float64(0.10638584196567535) - perf/mfu/actor:np.float64(0.04655599481898082) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.15143203735352) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7383928571428572) - val-aux/sciknoweval/reward/std@16:np.float64(0.14365431605650666) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.8011952380952381) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.1307038228818137) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.6770761904761905) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.11351933668678355) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7393857142857143) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.14329942361312858) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.8569714285714286) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.09830401861938036) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6338476190476191) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.07286850459718364) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7421190476190478) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.12378209678135982) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.8949142857142857) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.059950229295658866) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6086047619047619) - val-aux/sciknoweval/re
|
|
step:91 - global_seqlen/min:16867 - global_seqlen/max:27207 - global_seqlen/minmax_diff:10340 - global_seqlen/balanced_min:20730 - global_seqlen/balanced_max:20743 - global_seqlen/mean:20741.0 - actor/entropy:0.39037877321243286 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1890607476234436 - training/rollout_probs_diff_mean:0.0032652076333761215 - training/rollout_probs_diff_std:0.007450577337294817 - training/rollout_actor_probs_pearson_corr:0.9996206760406494 - self_distillation/success_group_fraction:0.96875 - self_distillation/success_sample_fraction:0.96484375 - 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:0.96484375 - rollout_corr/rollout_is_mean:0.9998917579650879 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6048963069915771 - rollout_corr/rollout_is_min:0.7054476141929626 - rollout_corr/rollout_is_std:0.026394300162792206 - rollout_corr/rollout_is_eff_sample_size:0.9993037068645211 - rollout_corr/rollout_is_seq_mean:0.9998647570610046 - rollout_corr/rollout_is_seq_std:0.0017023973632603884 - rollout_corr/rollout_is_seq_max:1.005475401878357 - rollout_corr/rollout_is_seq_min:0.9921031594276428 - rollout_corr/rollout_is_seq_max_deviation:0.007896840572357178 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5244424263946712) - rollout_corr/training_log_ppl:np.float64(0.4122463323874399) - rollout_corr/kl:np.float64(0.0005031662056946828) - rollout_corr/k3_kl:np.float64(0.0003679188381795484) - rollout_corr/rollout_ppl:np.float64(1.523619743064046) - rollout_corr/rollout_log_ppl:np.float64(0.41174316353863105) - rollout_corr/log_ppl_diff:np.float64(0.0005031688488088548) - rollout_corr/log_ppl_abs_diff:np.float64(0.001289652253035456) - rollout_corr/log_ppl_diff_max:np.float64(0.00843280553817749) - rollout_corr/log_ppl_diff_min:np.float64(-0.004891514778137207) - rollout_corr/ppl_ratio:np.float64(1.0005047880113125) - rollout_corr/chi2_token:np.float64(0.00046828039921820164) - rollout_corr/chi2_seq:np.float64(0.16324121574871242) - actor/pg_loss:np.float64(0.02536357068674988) - 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(0.1203346699476242) - perf/mfu/actor:np.float64(0.04710681279234217) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(199.1850996017456) - actor/lr:np.float64(1e-05) - training/global_step:91 - 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:0.007172529119998217 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.007172529119998217 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:362.03125 - response_length/max:1340.0 - response_length/min:153.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:362.03125 - response_length_non_aborted/max:1340.0 - response_length_non_aborted/min:153.0 - response_length_non_aborted/clip_ratio:0.0 - 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:3.1614676117897034e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.4471836313605309) - timing_s/agent_loop/generate_sequences/max:np.float64(10.356217896565795) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.027023465205275) - timing_s/agent_loop/tool_calls/mi
|
|
step:92 - global_seqlen/min:17966 - global_seqlen/max:22682 - global_seqlen/minmax_diff:4716 - global_seqlen/balanced_min:20182 - global_seqlen/balanced_max:20188 - global_seqlen/mean:20186.375 - actor/entropy:0.4032350778579712 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.19272543489933014 - training/rollout_probs_diff_mean:0.0034193717874586582 - training/rollout_probs_diff_std:0.007561853621155024 - training/rollout_actor_probs_pearson_corr:0.999582052230835 - self_distillation/success_group_fraction:0.84375 - self_distillation/success_sample_fraction:0.83203125 - self_distillation/correct_sample_fraction:0.62890625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.83203125 - rollout_corr/rollout_is_mean:1.0000437498092651 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.6260637044906616 - rollout_corr/rollout_is_min:0.5597866177558899 - rollout_corr/rollout_is_std:0.02645745500922203 - rollout_corr/rollout_is_eff_sample_size:0.9993004927047298 - rollout_corr/rollout_is_seq_mean:1.0000483989715576 - rollout_corr/rollout_is_seq_std:0.001514876028522849 - rollout_corr/rollout_is_seq_max:1.0047374963760376 - rollout_corr/rollout_is_seq_min:0.9956532716751099 - rollout_corr/rollout_is_seq_max_deviation:0.004737496376037598 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5263971290551126) - rollout_corr/training_log_ppl:np.float64(0.4141978725383524) - rollout_corr/kl:np.float64(0.000313541638590209) - rollout_corr/k3_kl:np.float64(0.0003620258462433412) - rollout_corr/rollout_ppl:np.float64(1.5259038410149515) - rollout_corr/rollout_log_ppl:np.float64(0.41388432699022815) - rollout_corr/log_ppl_diff:np.float64(0.00031354554812423885) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012031655933242291) - rollout_corr/log_ppl_diff_max:np.float64(0.00475078821182251) - rollout_corr/log_ppl_diff_min:np.float64(-0.004315018653869629) - rollout_corr/ppl_ratio:np.float64(1.0003147441893816) - rollout_corr/chi2_token:np.float64(0.0008282156195491552) - rollout_corr/chi2_seq:np.float64(0.2682835680898279) - actor/pg_loss:np.float64(0.021490446630195947) - 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(0.10505128651857376) - perf/mfu/actor:np.float64(0.046589224730296405) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(199.13045501708984) - actor/lr:np.float64(1e-05) - training/global_step:92 - training/epoch:1 - critic/score/mean:0.62890625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.62890625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.009068050421774387 - critic/advantages/max:0.875 - critic/advantages/min:-0.625 - critic/returns/mean:0.009068050421774387 - critic/returns/max:0.875 - critic/returns/min:-0.625 - response_length/mean:343.32421875 - response_length/max:769.0 - response_length/min:127.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:343.32421875 - response_length_non_aborted/max:769.0 - response_length_non_aborted/min:127.0 - 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:0.00011093169450759888 - timing_s/agent_loop/generate_sequences/min:np.float64(1.6635501701384783) - timing_s/agent_loop/generate_sequences/max:np.float64(6.915476659312844) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.076195679830562) - timing_s/agent_loop/tool_ca
|
|
step:93 - global_seqlen/min:14912 - global_seqlen/max:23076 - global_seqlen/minmax_diff:8164 - global_seqlen/balanced_min:18394 - global_seqlen/balanced_max:18396 - global_seqlen/mean:18395.5 - actor/entropy:0.41236981749534607 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.13426834344863892 - training/rollout_probs_diff_mean:0.003574734553694725 - training/rollout_probs_diff_std:0.007658370304852724 - training/rollout_actor_probs_pearson_corr:0.9995920658111572 - self_distillation/success_group_fraction:0.96875 - self_distillation/success_sample_fraction:0.96484375 - self_distillation/correct_sample_fraction:0.89453125 - 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:0.9999799132347107 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.2887338399887085 - rollout_corr/rollout_is_min:0.6945141553878784 - rollout_corr/rollout_is_std:0.02712269313633442 - rollout_corr/rollout_is_eff_sample_size:0.9992649002433537 - rollout_corr/rollout_is_seq_mean:1.0000003576278687 - rollout_corr/rollout_is_seq_std:0.001636735163629055 - rollout_corr/rollout_is_seq_max:1.0064903497695923 - rollout_corr/rollout_is_seq_min:0.9961798191070557 - rollout_corr/rollout_is_seq_max_deviation:0.006490349769592285 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5282936920411885) - rollout_corr/training_log_ppl:np.float64(0.4146989045548253) - rollout_corr/kl:np.float64(0.0003705817951065171) - rollout_corr/k3_kl:np.float64(0.00037091223762786285) - rollout_corr/rollout_ppl:np.float64(1.5277292910031974) - rollout_corr/rollout_log_ppl:np.float64(0.4143283190205693) - rollout_corr/log_ppl_diff:np.float64(0.00037058553425595164) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012861145660281181) - rollout_corr/log_ppl_diff_max:np.float64(0.004396498203277588) - rollout_corr/log_ppl_diff_min:np.float64(-0.005613982677459717) - rollout_corr/ppl_ratio:np.float64(1.000371980946511) - rollout_corr/chi2_token:np.float64(0.0007458962500095367) - rollout_corr/chi2_seq:np.float64(0.24556456739082932) - actor/pg_loss:np.float64(0.026858666129555786) - 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(0.12543518841266632) - perf/mfu/actor:np.float64(0.04267720705794013) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.51401138305664) - actor/lr:np.float64(1e-05) - training/global_step:93 - training/epoch:1 - critic/score/mean:0.89453125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.89453125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.00040849673678167164 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.00040849673678167164 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:308.390625 - response_length/max:764.0 - response_length/min:153.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:308.390625 - response_length_non_aborted/max:764.0 - response_length_non_aborted/min:153.0 - 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:4.4304877519607544e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(2.0778703466057777) - timing_s/agent_loop/generate_sequences/max:np.float64(6.9205704890191555) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.8023154061083915) - timing_s/agent_loop/t
|
|
step:94 - global_seqlen/min:15894 - global_seqlen/max:24399 - global_seqlen/minmax_diff:8505 - global_seqlen/balanced_min:19132 - global_seqlen/balanced_max:19142 - global_seqlen/mean:19138.125 - actor/entropy:0.3750686049461365 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.18785905838012695 - training/rollout_probs_diff_mean:0.0031683756969869137 - training/rollout_probs_diff_std:0.007266761269420385 - training/rollout_actor_probs_pearson_corr:0.9996273517608643 - self_distillation/success_group_fraction:1.0 - self_distillation/success_sample_fraction:0.99609375 - self_distillation/correct_sample_fraction:0.81640625 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.99609375 - rollout_corr/rollout_is_mean:1.0000969171524048 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.3875888586044312 - rollout_corr/rollout_is_min:0.7326339483261108 - rollout_corr/rollout_is_std:0.025707904249429703 - rollout_corr/rollout_is_eff_sample_size:0.9993396592460632 - rollout_corr/rollout_is_seq_mean:1.0000383853912354 - rollout_corr/rollout_is_seq_std:0.0016439437167719007 - rollout_corr/rollout_is_seq_max:1.0065339803695679 - rollout_corr/rollout_is_seq_min:0.9953336119651794 - rollout_corr/rollout_is_seq_max_deviation:0.006533980369567871 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4934561057016253) - rollout_corr/training_log_ppl:np.float64(0.3906862249423284) - rollout_corr/kl:np.float64(0.0003073524479724199) - rollout_corr/k3_kl:np.float64(0.0003456278131750423) - rollout_corr/rollout_ppl:np.float64(1.4929668521508574) - rollout_corr/rollout_log_ppl:np.float64(0.3903788701281883) - rollout_corr/log_ppl_diff:np.float64(0.0003073548141401261) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012503058824222535) - rollout_corr/log_ppl_diff_max:np.float64(0.0051155686378479) - rollout_corr/log_ppl_diff_min:np.float64(-0.006019532680511475) - rollout_corr/ppl_ratio:np.float64(1.0003087604418397) - rollout_corr/chi2_token:np.float64(0.0007715593092143536) - rollout_corr/chi2_seq:np.float64(0.6320373849011958) - actor/pg_loss:np.float64(0.024574471541200182) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.00390625) - actor/grad_norm:np.float64(0.11671514809131622) - perf/mfu/actor:np.float64(0.04427980230577799) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.3998794555664) - actor/lr:np.float64(1e-05) - training/global_step:94 - training/epoch:1 - critic/score/mean:0.81640625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.81640625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:0.004353114403784275 - critic/advantages/max:0.875 - critic/advantages/min:-0.75 - critic/returns/mean:0.004353114403784275 - critic/returns/max:0.875 - critic/returns/min:-0.75 - response_length/mean:345.25390625 - response_length/max:969.0 - response_length/min:144.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:345.25390625 - response_length_non_aborted/max:969.0 - response_length_non_aborted/min:144.0 - 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:0.00014358758926391602 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9073866792023182) - timing_s/agent_loop/generate_sequences/max:np.float64(7.9768695048987865) - timing_s/agent_loop/generate_sequences/mean:np.float64(4.207158071636513) - timing_s/agent_loop/tool_calls
|
|
step:95 - global_seqlen/min:15816 - global_seqlen/max:23140 - global_seqlen/minmax_diff:7324 - global_seqlen/balanced_min:19096 - global_seqlen/balanced_max:19099 - global_seqlen/mean:19097.625 - actor/entropy:0.3676719665527344 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.22370457649230957 - training/rollout_probs_diff_mean:0.0032451103907078505 - training/rollout_probs_diff_std:0.007513775490224361 - training/rollout_actor_probs_pearson_corr:0.999595046043396 - self_distillation/success_group_fraction:0.96875 - self_distillation/success_sample_fraction:0.9609375 - self_distillation/correct_sample_fraction:0.71875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.9609375 - rollout_corr/rollout_is_mean:1.0000975131988525 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.416880488395691 - rollout_corr/rollout_is_min:0.7036675810813904 - rollout_corr/rollout_is_std:0.025719493627548218 - rollout_corr/rollout_is_eff_sample_size:0.9993390639868975 - rollout_corr/rollout_is_seq_mean:1.0001386404037476 - rollout_corr/rollout_is_seq_std:0.0017106679733842611 - rollout_corr/rollout_is_seq_max:1.0050138235092163 - rollout_corr/rollout_is_seq_min:0.9917443990707397 - rollout_corr/rollout_is_seq_max_deviation:0.008255600929260254 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.4875444285571575) - rollout_corr/training_log_ppl:np.float64(0.3878075061365962) - rollout_corr/kl:np.float64(0.00021361277267573087) - rollout_corr/k3_kl:np.float64(0.0003522702571672198) - rollout_corr/rollout_ppl:np.float64(1.4872421850450337) - rollout_corr/rollout_log_ppl:np.float64(0.38759389059850946) - rollout_corr/log_ppl_diff:np.float64(0.00021361553808674216) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012911753146909177) - rollout_corr/log_ppl_diff_max:np.float64(0.009041011333465576) - rollout_corr/log_ppl_diff_min:np.float64(-0.004711031913757324) - rollout_corr/ppl_ratio:np.float64(1.00021508987993) - rollout_corr/chi2_token:np.float64(0.0009871532674878836) - rollout_corr/chi2_seq:np.float64(0.3058506976813078) - actor/pg_loss:np.float64(0.02389417673475691) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.0390625) - actor/grad_norm:np.float64(0.09887026250362396) - perf/mfu/actor:np.float64(0.0442911035402508) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.27938842773438) - actor/lr:np.float64(1e-05) - training/global_step:95 - training/epoch:1 - critic/score/mean:0.71875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.71875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.01118201483041048 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.01118201483041048 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:333.61328125 - response_length/max:914.0 - response_length/min:147.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:333.61328125 - response_length_non_aborted/max:914.0 - response_length_non_aborted/min:147.0 - 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:0.00013998709619045258 - timing_s/agent_loop/generate_sequences/min:np.float64(1.9920523762702942) - timing_s/agent_loop/generate_sequences/max:np.float64(7.638837767764926) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.9653719907728373) - timing_s/agent_loop/tool_calls/min:n
|
|
step:96 - global_seqlen/min:16506 - global_seqlen/max:20734 - global_seqlen/minmax_diff:4228 - global_seqlen/balanced_min:18562 - global_seqlen/balanced_max:18570 - global_seqlen/mean:18568.125 - actor/entropy:0.3837437331676483 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1243370771408081 - training/rollout_probs_diff_mean:0.0033954009413719177 - training/rollout_probs_diff_std:0.007726112846285105 - training/rollout_actor_probs_pearson_corr:0.9995816349983215 - self_distillation/success_group_fraction:0.96875 - self_distillation/success_sample_fraction:0.9609375 - 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:0.9609375 - rollout_corr/rollout_is_mean:0.999896764755249 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.521704077720642 - rollout_corr/rollout_is_min:0.5504812598228455 - rollout_corr/rollout_is_std:0.026567615568637848 - rollout_corr/rollout_is_eff_sample_size:0.9992945406115649 - rollout_corr/rollout_is_seq_mean:0.9999316930770874 - rollout_corr/rollout_is_seq_std:0.0016884636133909225 - rollout_corr/rollout_is_seq_max:1.0046794414520264 - rollout_corr/rollout_is_seq_min:0.9952861666679382 - rollout_corr/rollout_is_seq_max_deviation:0.004713833332061768 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5298391045071185) - rollout_corr/training_log_ppl:np.float64(0.41313516756054014) - rollout_corr/kl:np.float64(0.0004468855342381062) - rollout_corr/k3_kl:np.float64(0.00037861409984429883) - rollout_corr/rollout_ppl:np.float64(1.5291804745793343) - rollout_corr/rollout_log_ppl:np.float64(0.412688278069254) - rollout_corr/log_ppl_diff:np.float64(0.00044688949128612876) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013512258301489055) - rollout_corr/log_ppl_diff_max:np.float64(0.005292952060699463) - rollout_corr/log_ppl_diff_min:np.float64(-0.004189729690551758) - rollout_corr/ppl_ratio:np.float64(1.0004484120290726) - rollout_corr/chi2_token:np.float64(0.0006225025281310081) - rollout_corr/chi2_seq:np.float64(0.11581976874731481) - actor/pg_loss:np.float64(0.024461124197841855) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.0390625) - actor/grad_norm:np.float64(0.1165952980518341) - perf/mfu/actor:np.float64(0.04281109867655259) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.39710235595703) - actor/lr:np.float64(1e-05) - training/global_step:96 - 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:0.0016348825301975012 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:0.0016348825301975012 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:313.59765625 - response_length/max:958.0 - response_length/min:125.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:313.59765625 - response_length_non_aborted/max:958.0 - response_length_non_aborted/min:125.0 - response_length_non_aborted/clip_ratio:0.0 - 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:0.00017161108553409576 - timing_s/agent_loop/generate_sequences/min:np.float64(1.6753948535770178) - timing_s/agent_loop/generate_sequences/max:np.float64(7.938263056799769) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.7899505308741936) - timing_s/agent_loop/tool_cal
|
|
step:97 - global_seqlen/min:17567 - global_seqlen/max:22811 - global_seqlen/minmax_diff:5244 - global_seqlen/balanced_min:19731 - global_seqlen/balanced_max:19736 - global_seqlen/mean:19734.125 - actor/entropy:0.3824337422847748 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.21496814489364624 - training/rollout_probs_diff_mean:0.003402416128665209 - training/rollout_probs_diff_std:0.007751376833766699 - training/rollout_actor_probs_pearson_corr:0.999586820602417 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.921875 - self_distillation/correct_sample_fraction:0.6953125 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.921875 - rollout_corr/rollout_is_mean:0.9998922348022461 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.2838637828826904 - rollout_corr/rollout_is_min:0.7233055233955383 - rollout_corr/rollout_is_std:0.026276608929038048 - rollout_corr/rollout_is_eff_sample_size:0.9993098971564499 - rollout_corr/rollout_is_seq_mean:0.9998603463172913 - rollout_corr/rollout_is_seq_std:0.001723793218843639 - rollout_corr/rollout_is_seq_max:1.0054033994674683 - rollout_corr/rollout_is_seq_min:0.994937002658844 - rollout_corr/rollout_is_seq_max_deviation:0.005403399467468262 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5242640036158264) - rollout_corr/training_log_ppl:np.float64(0.41170950725791045) - rollout_corr/kl:np.float64(0.000516180335338845) - rollout_corr/k3_kl:np.float64(0.00037659904211295725) - rollout_corr/rollout_ppl:np.float64(1.523464354686439) - rollout_corr/rollout_log_ppl:np.float64(0.4111933222739026) - rollout_corr/log_ppl_diff:np.float64(0.0005161849840078503) - rollout_corr/log_ppl_abs_diff:np.float64(0.0013714819506276399) - rollout_corr/log_ppl_diff_max:np.float64(0.00566476583480835) - rollout_corr/log_ppl_diff_min:np.float64(-0.004933863878250122) - rollout_corr/ppl_ratio:np.float64(1.0005178111605346) - rollout_corr/chi2_token:np.float64(0.0004749374929815531) - rollout_corr/chi2_seq:np.float64(0.256080090533942) - actor/pg_loss:np.float64(0.023126466827307013) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.078125) - actor/grad_norm:np.float64(0.11917595565319061) - perf/mfu/actor:np.float64(0.043733904196214396) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.33081007003784) - actor/lr:np.float64(1e-05) - training/global_step:97 - training/epoch:1 - critic/score/mean:0.6953125 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.6953125 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.005037573631852865 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.005037573631852865 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:322.28515625 - response_length/max:857.0 - response_length/min:132.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:322.28515625 - response_length_non_aborted/max:857.0 - response_length_non_aborted/min:132.0 - response_length_non_aborted/clip_ratio:0.0 - 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:0.00012538768351078033 - timing_s/agent_loop/generate_sequences/min:np.float64(1.7283506728708744) - timing_s/agent_loop/generate_sequences/max:np.float64(7.300098922103643) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.7889412216027267) - timing_s/agent_loop/tool_calls/mi
|
|
step:98 - global_seqlen/min:15055 - global_seqlen/max:23063 - global_seqlen/minmax_diff:8008 - global_seqlen/balanced_min:18964 - global_seqlen/balanced_max:18967 - global_seqlen/mean:18965.375 - actor/entropy:0.4025936424732208 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.13014081120491028 - training/rollout_probs_diff_mean:0.0034873203840106726 - training/rollout_probs_diff_std:0.007587778847664595 - training/rollout_actor_probs_pearson_corr:0.9995974898338318 - self_distillation/success_group_fraction:0.875 - self_distillation/success_sample_fraction:0.875 - self_distillation/correct_sample_fraction:0.69921875 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.875 - rollout_corr/rollout_is_mean:0.9998480081558228 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.408286690711975 - rollout_corr/rollout_is_min:0.6845983862876892 - rollout_corr/rollout_is_std:0.026697421446442604 - rollout_corr/rollout_is_eff_sample_size:0.999287517232827 - rollout_corr/rollout_is_seq_mean:0.9998010396957397 - rollout_corr/rollout_is_seq_std:0.0018041635630652308 - rollout_corr/rollout_is_seq_max:1.0051735639572144 - rollout_corr/rollout_is_seq_min:0.9945611953735352 - rollout_corr/rollout_is_seq_max_deviation:0.005438804626464844 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5353243323042989) - rollout_corr/training_log_ppl:np.float64(0.420910959423054) - rollout_corr/kl:np.float64(0.0005718736656437073) - rollout_corr/k3_kl:np.float64(0.0003728994470293401) - rollout_corr/rollout_ppl:np.float64(1.5344101637601852) - rollout_corr/rollout_log_ppl:np.float64(0.4203390850452706) - rollout_corr/log_ppl_diff:np.float64(0.0005718743777833879) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014747647219337523) - rollout_corr/log_ppl_diff_max:np.float64(0.006104141473770142) - rollout_corr/log_ppl_diff_min:np.float64(-0.004441380500793457) - rollout_corr/ppl_ratio:np.float64(1.000573659548536) - rollout_corr/chi2_token:np.float64(0.00035082665272057056) - rollout_corr/chi2_seq:np.float64(0.1691306286957115) - actor/pg_loss:np.float64(0.022095311349403346) - 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(0.10847803950309753) - perf/mfu/actor:np.float64(0.044107948037882636) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.31708526611328) - actor/lr:np.float64(1e-05) - training/global_step:98 - training/epoch:1 - critic/score/mean:0.69921875 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.69921875 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.007903549820184708 - critic/advantages/max:0.75 - critic/advantages/min:-0.875 - critic/returns/mean:-0.007903549820184708 - critic/returns/max:0.75 - critic/returns/min:-0.875 - response_length/mean:293.94921875 - response_length/max:778.0 - response_length/min:133.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:293.94921875 - response_length_non_aborted/max:778.0 - response_length_non_aborted/min:133.0 - response_length_non_aborted/clip_ratio:0.0 - 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:6.200186908245087e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.7341085374355316) - timing_s/agent_loop/generate_sequences/max:np.float64(6.66490295343101) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.5941180460504256) - timing_s/agent_loop/tool_calls/min:np.flo
|
|
step:99 - global_seqlen/min:15629 - global_seqlen/max:23569 - global_seqlen/minmax_diff:7940 - global_seqlen/balanced_min:19284 - global_seqlen/balanced_max:19289 - global_seqlen/mean:19286.75 - actor/entropy:0.4128725528717041 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.1383599489927292 - training/rollout_probs_diff_mean:0.0036423271521925926 - training/rollout_probs_diff_std:0.007705996744334698 - training/rollout_actor_probs_pearson_corr:0.9995896816253662 - self_distillation/success_group_fraction:0.90625 - self_distillation/success_sample_fraction:0.8828125 - 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:0.8828125 - rollout_corr/rollout_is_mean:0.9999335408210754 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.758552074432373 - rollout_corr/rollout_is_min:0.5887811779975891 - rollout_corr/rollout_is_std:0.02760840766131878 - rollout_corr/rollout_is_eff_sample_size:0.9992382373180808 - rollout_corr/rollout_is_seq_mean:0.9999943971633911 - rollout_corr/rollout_is_seq_std:0.0018893217202275991 - rollout_corr/rollout_is_seq_max:1.006414532661438 - rollout_corr/rollout_is_seq_min:0.9945664405822754 - rollout_corr/rollout_is_seq_max_deviation:0.006414532661437988 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5488653075881302) - rollout_corr/training_log_ppl:np.float64(0.4293285302410368) - rollout_corr/kl:np.float64(0.0004041061194555162) - rollout_corr/k3_kl:np.float64(0.000398535114356946) - rollout_corr/rollout_ppl:np.float64(1.5482591893523932) - rollout_corr/rollout_log_ppl:np.float64(0.42892442041193135) - rollout_corr/log_ppl_diff:np.float64(0.0004041098291054368) - rollout_corr/log_ppl_abs_diff:np.float64(0.0014884080737829208) - rollout_corr/log_ppl_diff_max:np.float64(0.005733281373977661) - rollout_corr/log_ppl_diff_min:np.float64(-0.005734086036682129) - rollout_corr/ppl_ratio:np.float64(1.0004059437196702) - rollout_corr/chi2_token:np.float64(0.0007911187130957842) - rollout_corr/chi2_seq:np.float64(0.32431633002124727) - actor/pg_loss:np.float64(0.02191116502035584) - actor/kl_loss:np.float64(0.0) - self_distillation/dynamic_weighting_enabled:np.float64(0.0) - self_distillation/empty_target_batch:np.float64(0.1171875) - actor/grad_norm:np.float64(0.10866109281778336) - perf/mfu/actor:np.float64(0.04494325981092019) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(160.94140625) - perf/cpu_memory_used_gb:np.float64(198.1120834350586) - actor/lr:np.float64(1e-05) - training/global_step:99 - training/epoch:1 - critic/score/mean:0.625 - critic/score/max:1.0 - critic/score/min:0.0 - critic/rewards/mean:0.625 - critic/rewards/max:1.0 - critic/rewards/min:0.0 - critic/advantages/mean:-0.0004708984633907676 - critic/advantages/max:0.875 - critic/advantages/min:-0.875 - critic/returns/mean:-0.0004708984633907676 - critic/returns/max:0.875 - critic/returns/min:-0.875 - response_length/mean:290.3359375 - response_length/max:839.0 - response_length/min:127.0 - response_length/clip_ratio:0.0 - response_length_non_aborted/mean:290.3359375 - response_length_non_aborted/max:839.0 - response_length_non_aborted/min:127.0 - 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:9.31471586227417e-05 - timing_s/agent_loop/generate_sequences/min:np.float64(1.5497951079159975) - timing_s/agent_loop/generate_sequences/max:np.float64(7.139094682410359) - timing_s/agent_loop/generate_sequences/mean:np.float64(3.533284538832959) - timing_s/agent_loop/tool_calls/min:np.float64(0
|
|
test_gen_batch meta info: {'eos_token_id': 151645, 'pad_token_id': 151643, 'recompute_log_prob': False, 'do_sample': True, 'validate': True, 'global_steps': 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>
|
|
To determine the correct reactant for the synthesis of the product "Nc1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1", we need to analyze the molecular structure of the product and compare it with the options provided.
|
|
|
|
The product contains:
|
|
- A nitrogen atom bonded to a ring system (likely a pyridine-like structure).
|
|
- A substituted aromatic ring with a chlorine atom.
|
|
- A fluorinated aromatic ring (C(F)(F)F).
|
|
- A carbon chain connecting the nitrogen to the aromatic ring.
|
|
|
|
Now, evaluating each option:
|
|
|
|
- **Option A**: "O=C1c2ccccc2C(=O)N1c1ccn(Cc2ccc(Cl)cc2C(F)(F)F)n1"
|
|
This option has a carbonyl group (C=O), a nitrogen-containing ring, a chlorine-substituted aromatic ring, and a fluorinated aromatic ring, all of which match the product's structure.
|
|
|
|
- **Option B**: "O=C1c2ccccc2C(Cl)N1c1ccc2ccc(Cl)nc2n1"
|
|
This option has a chlorine atom but the ring structure and fluorinated group are not correctly positioned or present in the right configuration.
|
|
|
|
- **Option C**: "O=C1c2ccccc2C(=O)N1Cc1ccc(Oc2ccc(Cl)c(C(F)(F)F)c2)cc1"
|
|
This option includes an oxygen atom and a different arrangement of substituents that do not match the product.
|
|
|
|
- **Option D**: "O=C1c2ccccc2C(=O)N1CC=C(Cc1ccccc1)C(F)(F)F"
|
|
This option has a different ring structure and does not include the correct nitrogen-containing ring system.
|
|
|
|
Therefore, **Option A** is the only one that matches the structure of the product.
|
|
</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-Qwen3-8B-mbs32-tr0.1-train32-rollout8-lr1e-5-vllm0.8/global_step_100
|
|
step:100 - global_seqlen/min:18884 - global_seqlen/max:41519 - global_seqlen/minmax_diff:22635 - global_seqlen/balanced_min:20404 - global_seqlen/balanced_max:27699 - global_seqlen/mean:23139.75 - actor/entropy:0.2955058515071869 - perf/mfu/actor_infer:0 - training/rollout_probs_diff_valid:1 - training/rollout_probs_diff_max:0.12101149559020996 - training/rollout_probs_diff_mean:0.002571779303252697 - training/rollout_probs_diff_std:0.006634686142206192 - training/rollout_actor_probs_pearson_corr:0.9996629357337952 - self_distillation/success_group_fraction:0.9375 - self_distillation/success_sample_fraction:0.93359375 - self_distillation/correct_sample_fraction:0.74609375 - self_distillation/feedback_available_fraction:0.0 - self_distillation/feedback_used_fraction:0.0 - self_distillation/reprompt_sample_fraction:0.93359375 - rollout_corr/rollout_is_mean:1.0000677108764648 - rollout_corr/rollout_is_ratio_fraction_high:0.0 - rollout_corr/rollout_is_ratio_fraction_low:0.0 - rollout_corr/rollout_is_max:1.4396264553070068 - rollout_corr/rollout_is_min:0.7055256962776184 - rollout_corr/rollout_is_std:0.022850310429930687 - rollout_corr/rollout_is_eff_sample_size:0.999478016711329 - rollout_corr/rollout_is_seq_mean:1.0000479221343994 - rollout_corr/rollout_is_seq_std:0.0016783673781901598 - rollout_corr/rollout_is_seq_max:1.005556344985962 - rollout_corr/rollout_is_seq_min:0.9953861832618713 - rollout_corr/rollout_is_seq_max_deviation:0.005556344985961914 - rollout_corr/rollout_is_seq_fraction_high:0.0 - rollout_corr/rollout_is_seq_fraction_low:0.0 - rollout_corr/training_ppl:np.float64(1.5015285341069102) - rollout_corr/training_log_ppl:np.float64(0.3963073937320587) - rollout_corr/kl:np.float64(0.0003049511138506489) - rollout_corr/k3_kl:np.float64(0.0003528280437024023) - rollout_corr/rollout_ppl:np.float64(1.5010556285269558) - rollout_corr/rollout_log_ppl:np.float64(0.396002438999858) - rollout_corr/log_ppl_diff:np.float64(0.0003049547322007129) - rollout_corr/log_ppl_abs_diff:np.float64(0.0012992577630939195) - rollout_corr/log_ppl_diff_max:np.float64(0.00518268346786499) - rollout_corr/log_ppl_diff_min:np.float64(-0.00502324104309082) - rollout_corr/ppl_ratio:np.float64(1.0003064163029194) - rollout_corr/chi2_token:np.float64(0.0008032168261706829) - rollout_corr/chi2_seq:np.float64(0.38230782863684) - actor/pg_loss:np.float64(0.023008882750787052) - 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(0.10394406318664551) - perf/mfu/actor:np.float64(0.05015731000074418) - perf/max_memory_allocated_gb:np.float64(150.15545225143433) - perf/max_memory_reserved_gb:np.float64(161.068359375) - perf/cpu_memory_used_gb:np.float64(198.26213359832764) - actor/lr:np.float64(1e-05) - val-aux/sciknoweval/reward/mean@16:np.float64(0.7535714285714286) - val-aux/sciknoweval/reward/std@16:np.float64(0.12122068576933157) - val-aux/sciknoweval/reward/best@2/mean:np.float64(0.8051285714285714) - val-aux/sciknoweval/reward/best@2/std:np.float64(0.10681452617322888) - val-aux/sciknoweval/reward/worst@2/mean:np.float64(0.7031761904761905) - val-aux/sciknoweval/reward/worst@2/std:np.float64(0.09916615120061645) - val-aux/sciknoweval/reward/maj@2/mean:np.float64(0.7537380952380953) - val-aux/sciknoweval/reward/maj@2/std:np.float64(0.12035272201748512) - val-aux/sciknoweval/reward/best@4/mean:np.float64(0.8485619047619049) - val-aux/sciknoweval/reward/best@4/std:np.float64(0.08027809906180897) - val-aux/sciknoweval/reward/worst@4/mean:np.float64(0.6645857142857142) - val-aux/sciknoweval/reward/worst@4/std:np.float64(0.06781975137735766) - val-aux/sciknoweval/reward/maj@4/mean:np.float64(0.7588809523809524) - val-aux/sciknoweval/reward/maj@4/std:np.float64(0.10127914193936502) - val-aux/sciknoweval/reward/best@8/mean:np.float64(0.8807) - val-aux/sciknoweval/reward/best@8/std:np.float64(0.05177293853368067) - val-aux/sciknoweval/reward/worst@8/mean:np.float64(0.6388952380952381) - val-aux/sciknoweval/reward/worst@8/std
|
|
("Final validation metrics: {'val-aux/sciknoweval/reward/mean@16': "
|
|
"np.float64(0.7535714285714286), 'val-aux/sciknoweval/reward/std@16': "
|
|
"np.float64(0.12122068576933157), 'val-aux/sciknoweval/reward/best@2/mean': "
|
|
"np.float64(0.8051285714285714), 'val-aux/sciknoweval/reward/best@2/std': "
|
|
"np.float64(0.10681452617322888), 'val-aux/sciknoweval/reward/worst@2/mean': "
|
|
"np.float64(0.7031761904761905), 'val-aux/sciknoweval/reward/worst@2/std': "
|
|
"np.float64(0.09916615120061645), 'val-aux/sciknoweval/reward/maj@2/mean': "
|
|
"np.float64(0.7537380952380953), 'val-aux/sciknoweval/reward/maj@2/std': "
|
|
"np.float64(0.12035272201748512), 'val-aux/sciknoweval/reward/best@4/mean': "
|
|
"np.float64(0.8485619047619049), 'val-aux/sciknoweval/reward/best@4/std': "
|
|
"np.float64(0.08027809906180897), 'val-aux/sciknoweval/reward/worst@4/mean': "
|
|
"np.float64(0.6645857142857142), 'val-aux/sciknoweval/reward/worst@4/std': "
|
|
"np.float64(0.06781975137735766), 'val-aux/sciknoweval/reward/maj@4/mean': "
|
|
"np.float64(0.7588809523809524), 'val-aux/sciknoweval/reward/maj@4/std': "
|
|
"np.float64(0.10127914193936502), 'val-aux/sciknoweval/reward/best@8/mean': "
|
|
"np.float64(0.8807), 'val-aux/sciknoweval/reward/best@8/std': "
|
|
"np.float64(0.05177293853368067), 'val-aux/sciknoweval/reward/worst@8/mean': "
|
|
"np.float64(0.6388952380952381), 'val-aux/sciknoweval/reward/worst@8/std': "
|
|
"np.float64(0.04212482341518254), 'val-aux/sciknoweval/reward/maj@8/mean': "
|
|
"np.float64(0.7632142857142856), 'val-aux/sciknoweval/reward/maj@8/std': "
|
|
"np.float64(0.07759319570379096), 'val-aux/sciknoweval/reward/best@16/mean': "
|
|
"np.float64(0.9000428571428573), 'val-aux/sciknoweval/reward/best@16/std': "
|
|
"np.float64(0.02643611186706914), 'val-aux/sciknoweval/reward/worst@16/mean': "
|
|
"np.float64(0.623247619047619), 'val-aux/sciknoweval/reward/worst@16/std': "
|
|
"np.float64(0.02453756899133995), 'val-aux/sciknoweval/reward/maj@16/mean': "
|
|
"np.float64(0.7659095238095238), 'val-aux/sciknoweval/reward/maj@16/std': "
|
|
"np.float64(0.05543722468721845), 'val-aux/sciknoweval/score/mean@16': "
|
|
"np.float64(0.7535714285714286), 'val-aux/sciknoweval/score/std@16': "
|
|
"np.float64(0.12122068576933157), 'val-aux/sciknoweval/score/best@2/mean': "
|
|
"np.float64(0.8051285714285714), 'val-aux/sciknoweval/score/best@2/std': "
|
|
"np.float64(0.10681452617322888), 'val-aux/sciknoweval/score/worst@2/mean': "
|
|
"np.float64(0.7031761904761905), 'val-aux/sciknoweval/score/worst@2/std': "
|
|
"np.float64(0.09916615120061645), 'val-aux/sciknoweval/score/maj@2/mean': "
|
|
"np.float64(0.7537380952380953), 'val-aux/sciknoweval/score/maj@2/std': "
|
|
"np.float64(0.12035272201748512), 'val-aux/sciknoweval/score/best@4/mean': "
|
|
"np.float64(0.8485619047619049), 'val-aux/sciknoweval/score/best@4/std': "
|
|
"np.float64(0.08027809906180897), 'val-aux/sciknoweval/score/worst@4/mean': "
|
|
"np.float64(0.6645857142857142), 'val-aux/sciknoweval/score/worst@4/std': "
|
|
"np.float64(0.06781975137735766), 'val-aux/sciknoweval/score/maj@4/mean': "
|
|
"np.float64(0.7588809523809524), 'val-aux/sciknoweval/score/maj@4/std': "
|
|
"np.float64(0.10127914193936502), 'val-aux/sciknoweval/score/best@8/mean': "
|
|
"np.float64(0.8807), 'val-aux/sciknoweval/score/best@8/std': "
|
|
"np.float64(0.05177293853368067), 'val-aux/sciknoweval/score/worst@8/mean': "
|
|
"np.float64(0.6388952380952381), 'val-aux/sciknoweval/score/worst@8/std': "
|
|
"np.float64(0.04212482341518254), 'val-aux/sciknoweval/score/maj@8/mean': "
|
|
"np.float64(0.7632142857142856), 'val-aux/sciknoweval/score/maj@8/std': "
|
|
"np.float64(0.07759319570379096), 'val-aux/sciknoweval/score/best@16/mean': "
|
|
"np.float64(0.9000428571428573), 'val-aux/sciknoweval/score/best@16/std': "
|
|
"np.float64(0.02643611186706914), 'val-aux/sciknoweval/score/worst@16/mean': "
|
|
"np.float64(0.623247619047619), 'val-aux/sciknoweval/score/worst@16/std': "
|
|
"np.float64(0.02453756899133995), 'val-aux/sciknoweval/score/maj@16/mean': "
|
|
"np.float64(0.7659095238095238), 'val-aux/sciknoweval/score/maj@16/std': "
|
|
"np.float64(0.05543722468721845), 'val-core/sciknoweval/acc/mean@16': "
|
|
"np.float64(0.7535714285714286), 'val-aux/sciknoweval/acc/std@16': "
|
|
"np.float64(0.12122068576933157), 'val-aux/sciknoweval/acc/best@2/mean': "
|
|
"np.float64(0.8051285714285714), 'val-aux/sciknoweval/acc/best@2/std': "
|
|
"np.float64(0.10681452617322888), 'val-aux/sciknoweval/acc/worst@2/mean': "
|
|
"np.float64(0.7031761904761905), 'val-aux/sciknoweval/acc/worst@2/std': "
|
|
"np.float64(0.09916615120061645), 'val-aux/sciknoweval/acc/maj@2/mean': "
|
|
"np.float64(0.7537380952380953), 'val-aux/sciknoweval/acc/maj@2/std': "
|
|
"np.float64(0.12035272201748512), 'val-aux/sciknoweval/acc/best@4/mean': "
|
|
"np.float64(0.8485619047619049), 'val-aux/sciknoweval/acc/best@4/std': "
|
|
"np.float64(0.08027809906180897), 'val-aux/sciknoweval/acc/worst@4/mean': "
|
|
"np.float64(0.6645857142857142), 'val-aux/sciknoweval/acc/worst@4/std': "
|
|
"np.float64(0.06781975137735766), 'val-aux/sciknoweval/acc/maj@4/mean': "
|
|
"np.float64(0.7588809523809524), 'val-aux/sciknoweval/acc/maj@4/std': "
|
|
"np.float64(0.10127914193936502), 'val-aux/sciknoweval/acc/best@8/mean': "
|
|
"np.float64(0.8807), 'val-aux/sciknoweval/acc/best@8/std': "
|
|
"np.float64(0.05177293853368067), 'val-aux/sciknoweval/acc/worst@8/mean': "
|
|
"np.float64(0.6388952380952381), 'val-aux/sciknoweval/acc/worst@8/std': "
|
|
"np.float64(0.04212482341518254), 'val-aux/sciknoweval/acc/maj@8/mean': "
|
|
"np.float64(0.7632142857142856), 'val-aux/sciknoweval/acc/maj@8/std': "
|
|
"np.float64(0.07759319570379096), 'val-core/sciknoweval/acc/best@16/mean': "
|
|
"np.float64(0.9000428571428573), 'val-core/sciknoweval/acc/best@16/std': "
|
|
"np.float64(0.02643611186706914), 'val-aux/sciknoweval/acc/worst@16/mean': "
|
|
"np.float64(0.623247619047619), 'val-aux/sciknoweval/acc/worst@16/std': "
|
|
"np.float64(0.02453756899133995), 'val-core/sciknoweval/acc/maj@16/mean': "
|
|
"np.float64(0.7659095238095238), 'val-core/sciknoweval/acc/maj@16/std': "
|
|
'np.float64(0.05543722468721845), '
|
|
"'val-aux/sciknoweval/incorrect_format/mean@16': "
|
|
'np.float64(0.9994047619047619), '
|
|
"'val-aux/sciknoweval/incorrect_format/std@16': "
|
|
'np.float64(0.0023053472298853674), '
|
|
"'val-aux/sciknoweval/incorrect_format/best@2/mean': "
|
|
'np.float64(0.9999666666666667), '
|
|
"'val-aux/sciknoweval/incorrect_format/best@2/std': "
|
|
'np.float64(0.0005486204557414609), '
|
|
"'val-aux/sciknoweval/incorrect_format/worst@2/mean': "
|
|
'np.float64(0.9989095238095239), '
|
|
"'val-aux/sciknoweval/incorrect_format/worst@2/std': "
|
|
'np.float64(0.0030314321460128213), '
|
|
"'val-aux/sciknoweval/incorrect_format/maj@2/mean': "
|
|
'np.float64(0.9994619047619048), '
|
|
"'val-aux/sciknoweval/incorrect_format/maj@2/std': "
|
|
'np.float64(0.0021986843400458376), '
|
|
"'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(0.9978380952380952), '
|
|
"'val-aux/sciknoweval/incorrect_format/worst@4/std': "
|
|
'np.float64(0.003989388429962291), '
|
|
"'val-aux/sciknoweval/incorrect_format/maj@4/mean': "
|
|
'np.float64(0.9997333333333334), '
|
|
"'val-aux/sciknoweval/incorrect_format/maj@4/std': "
|
|
'np.float64(0.0015374570050248517), '
|
|
"'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(0.9963047619047618), '
|
|
"'val-aux/sciknoweval/incorrect_format/worst@8/std': "
|
|
'np.float64(0.004640859533041082), '
|
|
"'val-aux/sciknoweval/incorrect_format/maj@8/mean': "
|
|
'np.float64(0.9999476190476191), '
|
|
"'val-aux/sciknoweval/incorrect_format/maj@8/std': "
|
|
'np.float64(0.0004966783087710074), '
|
|
"'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(0.994052380952381), '
|
|
"'val-aux/sciknoweval/incorrect_format/worst@16/std': "
|
|
'np.float64(0.004610765367099621), '
|
|
"'val-aux/sciknoweval/incorrect_format/maj@16/mean': "
|
|
'np.float64(0.9999857142857144), '
|
|
"'val-aux/sciknoweval/incorrect_format/maj@16/std': "
|
|
"np.float64(0.0002604287412166727), '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)}')
|