asingh15's picture
Update FineProofs rollout distribution explorer (batch 1/1)
de181bd verified
Raw
History Blame Contribute Delete
11.2 kB
window.FINEPROOFS_DISTRIBUTION={meta:{"schema_version":"fineproofs_branch_distribution/v1","run_id":"fineproofs_qwen35_9b_resp81920_gptoss120b_m32_20260722","built_at_utc":"2026-07-29T21:11:59.012177+00:00","construction":{"source_problems":4164,"selected_base_trajectories":8328,"prefix_rows":103518,"semantic_problem_prefixes":99354,"continuations_per_prefix":32,"continuation_grades":3312576,"skipped_non_grade_journal_rows":2111},"continuation_grades":{"count":3312576,"mean":0.527504,"std":0.395719,"min":0.0,"p10":0.0,"p25":0.142857,"median":0.571429,"p75":1.0,"p90":1.0,"max":1.0,"zero_count":671905,"partial_count":1660817,"full_credit_count":979854,"at_least_half_credit_count":1700481,"truncated_count":36157,"histogram":{"edges":[0.0,0.05,0.1,0.15,0.2,0.25,0.3,0.35,0.4,0.45,0.5,0.55,0.6,0.65,0.7,0.75,0.8,0.85,0.9,0.95,1.0],"counts":[671906,796,387965,4,321,292349,7,297,258448,2,270,222826,225,6,253600,270,4,243170,256,979854]}},"prefix_values":{"count":103518,"mean":0.527504,"std":0.344247,"min":0.0,"p10":0.040178,"p25":0.196342,"median":0.544614,"p75":0.866062,"p90":0.977679,"max":1.0,"exact_zero_count":4072,"exact_one_count":4799,"histogram":{"edges":[0.0,0.025,0.05,0.075,0.1,0.125,0.15,0.175,0.2,0.225,0.25,0.275,0.3,0.325,0.35,0.375,0.4,0.425,0.45,0.475,0.5,0.525,0.55,0.575,0.6,0.625,0.65,0.675,0.7,0.725,0.75,0.775,0.8,0.825,0.85,0.875,0.9,0.925,0.95,0.975,1.0],"counts":[8585,3058,2120,2478,2264,3088,2768,1796,1974,1926,1975,2266,1760,1907,1932,1648,1922,1714,1777,1840,1664,1799,1563,1806,1886,1656,1965,1698,2340,2131,1979,2251,2066,2478,2651,2627,3194,3145,4691,11130]}},"base_endpoints":{"count":8328,"mean":0.444987,"std":0.398816,"min":0.0,"p10":0.0,"p25":0.0,"median":0.357071,"p75":0.857143,"p90":1.0,"max":1.0,"grade_at_least_half_count":3522,"grade_below_half_count":4806,"histogram":{"edges":[0.0,0.05,0.1,0.15,0.2,0.25,0.3,0.35,0.4,0.45,0.5,0.55,0.6,0.65,0.7,0.75,0.8,0.85,0.9,0.95,1.0],"counts":[2378,3,1077,1,2,702,0,2,641,0,1,489,0,1,582,0,0,490,0,1959]}},"depth_summary":[{"count":13470,"mean":0.516909,"std":0.340286,"min":0.0,"p10":0.040179,"p25":0.183037,"median":0.526786,"p75":0.848212,"p90":0.964286,"max":1.0,"group":"all","depth_bin":0,"progress_lo":0.0,"progress_hi":0.1},{"count":8966,"mean":0.565842,"std":0.330263,"min":0.0,"p10":0.075893,"p25":0.26781,"median":0.617203,"p75":0.87945,"p90":0.973214,"max":1.0,"group":"all","depth_bin":1,"progress_lo":0.1,"progress_hi":0.2},{"count":9379,"mean":0.54736,"std":0.33584,"min":0.0,"p10":0.058037,"p25":0.232145,"median":0.584795,"p75":0.870464,"p90":0.973214,"max":1.0,"group":"all","depth_bin":2,"progress_lo":0.2,"progress_hi":0.3},{"count":8932,"mean":0.547907,"std":0.335357,"min":0.0,"p10":0.06241,"p25":0.235555,"median":0.58033,"p75":0.870536,"p90":0.973214,"max":1.0,"group":"all","depth_bin":3,"progress_lo":0.3,"progress_hi":0.4},{"count":9195,"mean":0.539409,"std":0.338103,"min":0.0,"p10":0.053563,"p25":0.219868,"median":0.562507,"p75":0.866071,"p90":0.977679,"max":1.0,"group":"all","depth_bin":4,"progress_lo":0.4,"progress_hi":0.5},{"count":9549,"mean":0.534937,"std":0.341361,"min":0.0,"p10":0.044644,"p25":0.214284,"median":0.553571,"p75":0.870527,"p90":0.977679,"max":1.0,"group":"all","depth_bin":5,"progress_lo":0.5,"progress_hi":0.6},{"count":9418,"mean":0.527549,"std":0.342732,"min":0.0,"p10":0.040179,"p25":0.2052,"median":0.540164,"p75":0.861607,"p90":0.982143,"max":1.0,"group":"all","depth_bin":6,"progress_lo":0.6,"progress_hi":0.7},{"count":8483,"mean":0.530992,"std":0.342365,"min":0.0,"p10":0.044642,"p25":0.205359,"median":0.549096,"p75":0.866056,"p90":0.982143,"max":1.0,"group":"all","depth_bin":7,"progress_lo":0.7,"progress_hi":0.8},{"count":9984,"mean":0.515021,"std":0.350328,"min":0.0,"p10":0.026786,"p25":0.173986,"median":0.513393,"p75":0.866071,"p90":0.986607,"max":1.0,"group":"all","depth_bin":8,"progress_lo":0.8,"progress_hi":0.9},{"count":16142,"mean":0.486908,"std":0.362867,"min":0.0,"p10":0.013304,"p25":0.138393,"median":0.450866,"p75":0.861606,"p90":0.991071,"max":1.0,"group":"all","depth_bin":9,"progress_lo":0.9,"progress_hi":1.0},{"count":5867,"mean":0.767533,"std":0.228456,"min":0.0,"p10":0.410714,"p25":0.639509,"median":0.843746,"p75":0.950893,"p90":0.986607,"max":1.0,"group":"endpoint_correct","depth_bin":0,"progress_lo":0.0,"progress_hi":0.1},{"count":4083,"mean":0.804559,"std":0.196925,"min":0.0,"p10":0.513381,"p25":0.697567,"median":0.870531,"p75":0.964281,"p90":0.991071,"max":1.0,"group":"endpoint_correct","depth_bin":1,"progress_lo":0.1,"progress_hi":0.2},{"count":4169,"mean":0.803432,"std":0.201389,"min":0.0,"p10":0.508025,"p25":0.700892,"median":0.870522,"p75":0.964286,"p90":0.995536,"max":1.0,"group":"endpoint_correct","depth_bin":2,"progress_lo":0.2,"progress_hi":0.3},{"count":4004,"mean":0.807959,"std":0.1968,"min":0.008929,"p10":0.517814,"p25":0.705353,"median":0.874987,"p75":0.968746,"p90":0.995536,"max":1.0,"group":"endpoint_correct","depth_bin":3,"progress_lo":0.3,"progress_hi":0.4},{"count":4119,"mean":0.807061,"std":0.198506,"min":0.0,"p10":0.513394,"p25":0.703039,"median":0.874987,"p75":0.96875,"p90":0.995536,"max":1.0,"group":"endpoint_correct","depth_bin":4,"progress_lo":0.4,"progress_hi":0.5},{"count":4276,"mean":0.811629,"std":0.196226,"min":0.026786,"p10":0.522295,"p25":0.700894,"median":0.879464,"p75":0.973214,"p90":1.0,"max":1.0,"group":"endpoint_correct","depth_bin":5,"progress_lo":0.5,"progress_hi":0.6},{"count":4197,"mean":0.811937,"std":0.1973,"min":0.008929,"p10":0.52232,"p25":0.705357,"median":0.875001,"p75":0.973214,"p90":1.0,"max":1.0,"group":"endpoint_correct","depth_bin":6,"progress_lo":0.6,"progress_hi":0.7},{"count":3821,"mean":0.817813,"std":0.188801,"min":0.026786,"p10":0.540134,"p25":0.709821,"median":0.879465,"p75":0.977674,"p90":1.0,"max":1.0,"group":"endpoint_correct","depth_bin":7,"progress_lo":0.7,"progress_hi":0.8},{"count":4477,"mean":0.819862,"std":0.190263,"min":0.07143,"p10":0.538385,"p25":0.705357,"median":0.888377,"p75":0.982143,"p90":1.0,"max":1.0,"group":"endpoint_correct","depth_bin":8,"progress_lo":0.8,"progress_hi":0.9},{"count":7207,"mean":0.820753,"std":0.19431,"min":0.0,"p10":0.53568,"p25":0.700893,"median":0.892768,"p75":0.986607,"p90":1.0,"max":1.0,"group":"endpoint_correct","depth_bin":9,"progress_lo":0.9,"progress_hi":1.0},{"count":7603,"mean":0.323511,"std":0.281071,"min":0.0,"p10":0.013393,"p25":0.080365,"median":0.245536,"p75":0.51338,"p90":0.772322,"max":1.0,"group":"endpoint_incorrect","depth_bin":0,"progress_lo":0.0,"progress_hi":0.1},{"count":4883,"mean":0.366235,"std":0.283477,"min":0.0,"p10":0.022321,"p25":0.120537,"median":0.312487,"p75":0.584822,"p90":0.803571,"max":1.0,"group":"endpoint_incorrect","depth_bin":1,"progress_lo":0.1,"progress_hi":0.2},{"count":5210,"mean":0.342454,"std":0.275916,"min":0.0,"p10":0.017857,"p25":0.107145,"median":0.281251,"p75":0.540165,"p90":0.772319,"max":1.0,"group":"endpoint_incorrect","depth_bin":2,"progress_lo":0.2,"progress_hi":0.3},{"count":4928,"mean":0.336614,"std":0.269782,"min":0.0,"p10":0.017857,"p25":0.107058,"median":0.285701,"p75":0.526696,"p90":0.758925,"max":1.0,"group":"endpoint_incorrect","depth_bin":3,"progress_lo":0.3,"progress_hi":0.4},{"count":5076,"mean":0.322219,"std":0.264191,"min":0.0,"p10":0.013393,"p25":0.098214,"median":0.267856,"p75":0.501113,"p90":0.736594,"max":1.0,"group":"endpoint_incorrect","depth_bin":4,"progress_lo":0.4,"progress_hi":0.5},{"count":5273,"mean":0.31056,"std":0.259557,"min":0.0,"p10":0.008929,"p25":0.089291,"median":0.258929,"p75":0.48214,"p90":0.714286,"max":1.0,"group":"endpoint_incorrect","depth_bin":5,"progress_lo":0.5,"progress_hi":0.6},{"count":5221,"mean":0.298939,"std":0.251641,"min":0.0,"p10":0.008929,"p25":0.084821,"median":0.25,"p75":0.459819,"p90":0.687482,"max":1.0,"group":"endpoint_incorrect","depth_bin":6,"progress_lo":0.6,"progress_hi":0.7},{"count":4662,"mean":0.295912,"std":0.247748,"min":0.0,"p10":0.008929,"p25":0.085896,"median":0.245534,"p75":0.450894,"p90":0.674095,"max":1.0,"group":"endpoint_incorrect","depth_bin":7,"progress_lo":0.7,"progress_hi":0.8},{"count":5507,"mean":0.267196,"std":0.236876,"min":0.0,"p10":0.004464,"p25":0.066876,"median":0.209806,"p75":0.415178,"p90":0.624987,"max":1.0,"group":"endpoint_incorrect","depth_bin":8,"progress_lo":0.8,"progress_hi":0.9},{"count":8935,"mean":0.217627,"std":0.212168,"min":0.0,"p10":0.0,"p25":0.03237,"median":0.156162,"p75":0.340446,"p90":0.517855,"max":1.0,"group":"endpoint_incorrect","depth_bin":9,"progress_lo":0.9,"progress_hi":1.0}],"trajectory_patterns":{"falling":1390,"flat":5554,"rising":1187,"volatile":197},"trajectory_length_histogram":{"1":29,"2":157,"3":462,"4":629,"5":597,"6":615,"7":571,"8":590,"9":557,"10":493,"11":438,"12":334,"13":305,"14":259,"15":244,"16":226,"17":167,"18":175,"19":167,"20":126,"21":104,"22":99,"23":89,"24":108,"25":63,"26":61,"27":54,"28":57,"29":60,"30":42,"31":33,"32":24,"33":29,"34":21,"35":35,"36":25,"37":22,"38":22,"39":15,"40":13,"41":20,"42":8,"43":12,"44":13,"45":13,"46":13,"47":4,"48":13,"49":5,"50":7,"51":4,"52":2,"53":8,"54":7,"55":6,"56":3,"57":5,"58":3,"59":3,"60":4,"61":3,"62":3,"63":2,"64":4,"65":3,"66":7,"67":1,"68":3,"70":2,"74":3,"75":1,"76":1,"78":2,"80":2,"81":1,"82":2,"84":2,"86":1,"88":1,"93":1,"95":1,"97":1,"100":2,"111":1,"113":2,"114":1,"115":1,"125":1,"128":1,"132":1,"156":1},"point_fields":["step_idx","num_steps","progress","mean_grade","std_grade","q25_grade","median_grade","q75_grade","min_grade","max_grade","zero_count","full_credit_count","truncated_count"],"point_grade_histogram_bins":20,"reasoning_traces":{"schema_version":"fineproofs_reasoning_traces/v1","problem_count":4164,"trace_count":8328,"total_reasoning_chars":535315782,"max_reasoning_chars":747294,"shard_count":256,"incomplete_sampled_trace_count":321},"prm_provenance":{"aggregation":{"collapsed_branch_count":0,"dropped_step_zero_count":8328,"duplicate_group_count":0,"filtered_semantic_group_count":0,"input_observation_count":103518,"invalid_metadata_count":0,"kept_semantic_group_count":95190,"matched_observation_count":103518,"semantic_group_count_before_filter":95190,"unmatched_observation_count":0},"builder":"/storage/home/anikaitsingh/compressed_value_functions/inference/curate/build_prm_mc.py","dataset_rows":{"train":99037,"val":3282},"drop_step_zero":true,"hard_endpoints":{"base_responses":["/checkpoint/scientific-reasoning/anikaitsingh/compressed_value_functions/rollouts/curate/branch_runs/fineproofs_qwen35_9b_resp81920_gptoss120b_m32_20260722/base/mathcorpus_fineproofs_base.jsonl"],"enabled":true,"filtered_partial_prefixes":977,"filtered_trajectories":170,"incomplete_response_candidates":163,"metadata_correctness_labels":8295,"metadata_grade_threshold":0.5,"overlength_candidates":7,"represented_trajectories":8125,"rows":{"train":7874,"val":251},"skipped_length":0,"step_index_convention":"num_steps_plus_one","uses_trajectory_template":true},"partial_rows":{"train":91163,"val":3031},"prompt_length_filtered_prefixes":19,"provenance_version":"prm_mc_dataset/v2","target_aggregation":"pooled_numerator_denominator_by_problem_and_prefix","template":"/storage/home/anikaitsingh/compressed_value_functions/inference/math_corpus/prm_prompts/policy_prompt_mc_value.md"}},problems:[]};