Datasetpublic

anirudhb11/rebase_gpt-oss-20b_rg_cognition_ns32_md4_bt0_1_seed65_rg_cognition__v0

Aggregated Metrics Weighted mean of per-shard wandb summary values (weights = shard rows). See _meta/aggregated_metrics.json for raw values and _meta/splits.json for the per-shard wandb run names. Aggregated from 10 shards. Metric Value avg_response_tokens 5610.97 generation_phase_time_s 523.591 judge_output_tokens_level_1 114132 judge_output_tokens_level_2 42085.3 judge_output_tokens_level_3 26250.5 judge_output_tokens_level_4 18430.2… See the full description on the dataset page: https://huggingface.co/datasets/anirudhb11/rebase_gpt-oss-20b_rg_cognition_ns32_md4_bt0_1_seed65_rg_cognition__v0.

sourceHugging Faceupdated 5mo agoView on Hugging Face

0likes4downloads

Dataset Card

Aggregated Metrics

Weighted mean of per-shard wandb summary values (weights = shard rows). See `meta/aggregatedmetrics.json` for raw values and `meta/splits.json` for the per-shard wandb run names._

Aggregated from 10 shards.

Metric	Value
`avg_response_tokens`	5610.97
`generation_phase_time_s`	523.591
`judge_output_tokens_level_1`	114132
`judge_output_tokens_level_2`	42085.3
`judge_output_tokens_level_3`	26250.5
`judge_output_tokens_level_4`	18430.2
`judge_output_tokens_search`	200898
`judge_skipped_completed_level_1`	183.9
`judge_skipped_completed_level_2`	66.8
`judge_skipped_completed_level_3`	27.8
`judge_skipped_completed_level_4`	12.6
`judge_token_cdf_level_1`	1
`judge_token_cdf_level_2`	1
`judge_token_cdf_level_3`	1
`judge_token_cdf_level_4`	1
`judge_token_count_level_1`	11950
`judge_token_count_level_2`	5670
`judge_token_count_level_3`	4110
`judge_token_count_level_4`	3615
`maj@1`	0.505558
`maj@16`	0.531873
`maj@16_std`	0.0411912
`maj@1_std`	0.05645
`maj@2`	0.498297
`maj@2_std`	0.0638959
`maj@32`	0.531255
`maj@32_std`	0.0359006
`maj@4`	0.526578
`maj@4_std`	0.0486834
`maj@8`	0.528526
`maj@8_std`	0.0394321
`median_response_tokens`	2913.9
`num_unique_answers@1`	0.908
`num_unique_answers@16`	5.132
`num_unique_answers@2`	1.314
`num_unique_answers@32`	8.63
`num_unique_answers@4`	1.962
`num_unique_answers@8`	3.097
`pass@1`	0.472813
`pass@16`	0.618465
`pass@2`	0.543165
`pass@32`	0.63
`pass@4`	0.576253
`pass@8`	0.599146
`policy_output_tokens_final_completion`	285292
`policy_output_tokens_level_1`	759761
`policy_output_tokens_level_2`	392014
`policy_output_tokens_level_3`	217022
`policy_output_tokens_level_4`	141699
`policy_output_tokens_search_phases`	1.5105e+06
`policy_output_tokens_total_so_far`	1.79579e+06
`token_cdf`	1
`token_count`	29400
`total_generated_output_tokens`	2.17385e+06
`total_judge_output_tokens`	378060
`total_judge_output_tokens_search`	200898
`total_judge_output_tokens_weighted_mv`	177162
`total_policy_output_tokens`	1.79579e+06
`total_time_s`	672.63
`truncated_count_level_1`	0.2
`truncated_count_level_2`	0
`truncated_count_level_3`	0
`truncated_count_level_4`	0
`truncated_rate_level_1`	0.00149916
`truncated_rate_level_2`	0
`truncated_rate_level_3`	0
`truncated_rate_level_4`	0
`w_best@1`	0.513212
`w_best@16`	0.490008
`w_best@16_std`	0.070197
`w_best@1_std`	0.0685179
`w_best@2`	0.51572
`w_best@2_std`	0.0626417
`w_best@32`	0.496805
`w_best@32_std`	0.0600436
`w_best@4`	0.497089
`w_best@4_std`	0.069911
`w_best@8`	0.487709
`w_best@8_std`	0.0663563
`w_maj@1`	0.505558
`w_maj@16`	0.525707
`w_maj@16_std`	0.0369956
`w_maj@1_std`	0.05645
`w_maj@2`	0.506912
`w_maj@2_std`	0.0709063
`w_maj@32`	0.525007
`w_maj@32_std`	0.0297644
`w_maj@4`	0.526805
`w_maj@4_std`	0.047325
`w_maj@8`	0.528917
`w_maj@8_std`	0.0357681
`w_num_unique_answers@1`	0.908
`w_num_unique_answers@16`	5.132
`w_num_unique_answers@2`	1.314
`w_num_unique_answers@32`	8.63
`w_num_unique_answers@4`	1.962
`w_num_unique_answers@8`	3.097