orukeet / report /current-benchmark-validation.json
NathanRoll's picture
Prepare the Orukeet v0.1.0 model card, report, benchmark scores and citations
769e303 verified
Raw
History Blame Contribute Delete
8.22 kB
{
"status": "passed",
"publication_authorized": false,
"models": {
"parakeet": "3cbdc85877e668ca7b82d0d56770eb1fac76691f55d6b97545e8d61ca588d10d",
"orukeet": "031c8ddab4845aeced904a7cde8e8aa57993b2e344716cf83a545b079c473b56"
},
"summaries": {
"fleurs": {
"splits": [
"fleurs_bg",
"fleurs_cs",
"fleurs_da",
"fleurs_de",
"fleurs_el",
"fleurs_en",
"fleurs_es",
"fleurs_et",
"fleurs_fi",
"fleurs_fr",
"fleurs_hr",
"fleurs_hu",
"fleurs_it",
"fleurs_lt",
"fleurs_lv",
"fleurs_mt",
"fleurs_nl",
"fleurs_pl",
"fleurs_pt",
"fleurs_ro",
"fleurs_ru",
"fleurs_sk",
"fleurs_sl",
"fleurs_sv",
"fleurs_uk"
],
"rows": 20146,
"models": {
"parakeet": {
"errors": 46442,
"words": 421870,
"substitutions": 38136,
"deletions": 5461,
"insertions": 2845,
"chars": 2757813,
"char_errors": 98352,
"utterance_error": 14637,
"wer": 11.008604546424253,
"cer": 3.5663041692819637
},
"orukeet": {
"errors": 41521,
"words": 421715,
"substitutions": 34613,
"deletions": 3955,
"insertions": 2953,
"chars": 2757813,
"char_errors": 86408,
"utterance_error": 13762,
"wer": 9.845748906251853,
"cer": 3.133207363951073
}
},
"wins": 23,
"ties": 0,
"losses": 2,
"relative_wer_reduction_percent": 10.563152080434325
},
"standard": {
"splits": [
"fleurs_bg",
"fleurs_cs",
"fleurs_da",
"fleurs_de",
"fleurs_el",
"fleurs_en",
"fleurs_es",
"fleurs_et",
"fleurs_fi",
"fleurs_fr",
"fleurs_hr",
"fleurs_hu",
"fleurs_it",
"fleurs_lt",
"fleurs_lv",
"fleurs_mt",
"fleurs_nl",
"fleurs_pl",
"fleurs_pt",
"fleurs_ro",
"fleurs_ru",
"fleurs_sk",
"fleurs_sl",
"fleurs_sv",
"fleurs_uk",
"librispeech_test_clean",
"librispeech_test_other"
],
"rows": 25705,
"models": {
"parakeet": {
"errors": 48914,
"words": 527711,
"substitutions": 40179,
"deletions": 5631,
"insertions": 3104,
"chars": 3309718,
"char_errors": 103604,
"utterance_error": 16201,
"wer": 9.269088573101564,
"cer": 3.130296901427856
},
"orukeet": {
"errors": 43804,
"words": 527556,
"substitutions": 36526,
"deletions": 4118,
"insertions": 3160,
"chars": 3309718,
"char_errors": 91222,
"utterance_error": 15227,
"wer": 8.303194352827,
"cer": 2.756186478727191
}
},
"wins": 25,
"ties": 0,
"losses": 2,
"relative_wer_reduction_percent": 10.420595430251279
},
"domains": {
"splits": [
"eurospeech_bg",
"eurospeech_de",
"eurospeech_el",
"eurospeech_en",
"eurospeech_et",
"eurospeech_fi",
"eurospeech_fr",
"eurospeech_hr",
"eurospeech_it",
"eurospeech_lt",
"eurospeech_lv",
"eurospeech_mt",
"eurospeech_pt",
"eurospeech_sk",
"eurospeech_sl",
"eurospeech_uk",
"gigaspeechbench_agr_en",
"gigaspeechbench_ait_en",
"gigaspeechbench_art_en",
"gigaspeechbench_bio_en",
"gigaspeechbench_chn_en",
"gigaspeechbench_ecm_en",
"gigaspeechbench_eng_en",
"gigaspeechbench_ent_en",
"gigaspeechbench_fin_en",
"gigaspeechbench_hum_en",
"gigaspeechbench_ind_en",
"gigaspeechbench_jpn_en",
"gigaspeechbench_law_en",
"gigaspeechbench_med_en",
"gigaspeechbench_mil_en",
"gigaspeechbench_phl_en",
"gigaspeechbench_sct_en",
"gigaspeechbench_sgp_en",
"golos_crowd_ru",
"golos_farfield_ru",
"lesbos_el",
"monsoon_en_in",
"nst_da_da",
"nst_sv_sv",
"voxpopuli_cs",
"voxpopuli_es",
"voxpopuli_hu",
"voxpopuli_it",
"voxpopuli_nl",
"voxpopuli_pl",
"voxpopuli_ro"
],
"rows": 12006,
"models": {
"parakeet": {
"errors": 43939,
"words": 262747,
"substitutions": 26343,
"deletions": 8233,
"insertions": 9363,
"chars": 1584470,
"char_errors": 141363,
"utterance_error": 8290,
"wer": 16.72293118475187,
"cer": 8.921784571497094
},
"orukeet": {
"errors": 40068,
"words": 262698,
"substitutions": 23458,
"deletions": 7145,
"insertions": 9465,
"chars": 1584470,
"char_errors": 129717,
"utterance_error": 8039,
"wer": 15.25249526071763,
"cer": 8.186775388615752
}
},
"wins": 36,
"ties": 0,
"losses": 11,
"relative_wer_reduction_percent": 8.792931740190369
},
"domain_english": {
"splits": [
"eurospeech_en",
"gigaspeechbench_agr_en",
"gigaspeechbench_ait_en",
"gigaspeechbench_art_en",
"gigaspeechbench_bio_en",
"gigaspeechbench_chn_en",
"gigaspeechbench_ecm_en",
"gigaspeechbench_eng_en",
"gigaspeechbench_ent_en",
"gigaspeechbench_fin_en",
"gigaspeechbench_hum_en",
"gigaspeechbench_ind_en",
"gigaspeechbench_jpn_en",
"gigaspeechbench_law_en",
"gigaspeechbench_med_en",
"gigaspeechbench_mil_en",
"gigaspeechbench_phl_en",
"gigaspeechbench_sct_en",
"gigaspeechbench_sgp_en",
"monsoon_en_in"
],
"rows": 5120,
"models": {
"parakeet": {
"errors": 9032,
"words": 94993,
"substitutions": 4421,
"deletions": 1646,
"insertions": 2965,
"chars": 498262,
"char_errors": 29760,
"utterance_error": 2806,
"wer": 9.508069015611676,
"cer": 5.972761318342559
},
"orukeet": {
"errors": 8399,
"words": 94993,
"substitutions": 4193,
"deletions": 1490,
"insertions": 2716,
"chars": 498262,
"char_errors": 27417,
"utterance_error": 2692,
"wer": 8.841704125567148,
"cer": 5.502526783098049
}
},
"wins": 20,
"ties": 0,
"losses": 0,
"relative_wer_reduction_percent": 7.008414526129303
}
},
"checks": [
"Fresh matched decoding of both checkpoints",
"All 74 paired split results recomputed from per-record counts",
"Pooled WER recomputed from summed errors and reference words",
"Independent upstream scoring matches all 148 model/partition pairs",
"Rendered abstract and all 74 benchmark rows match the current checkpoint evidence"
],
"inputs_sha256": {
"evidence/standard-asr-r3-20260908/comparison.json": "9b2702bfe9c7b107db1627df3cb71d24d25d4eb8da0ee8683b0bbf9f994ca7da",
"evidence/standard-asr-r3-20260908/hypotheses-audit.json": "a60177ac2ac1064b481cc85424d5ee989cca43112191ddae30d60b7f56ae57d3",
"evidence/standard-asr-r3-20260908/numeric-evidence.jsonl.gz": "3d1467e56efd6769da1f8cb74eef0c3eac7bf93fa8200c39538963a9d4b080c1",
"evidence/domains-r3-20260908/comparison.json": "617a2119027b9e706907fdeb8a5819babb55f1959456d677e744d2b57ddfd2be",
"evidence/domains-r3-20260908/hypotheses-audit.json": "66c748f9ea625cf38e9bf672628e95300a5919abbaf47ebfd7c4800f5b1f3116",
"evidence/domains-r3-20260908/numeric-evidence.jsonl.gz": "298f3164c745523f25b2e7df25492dab5b6607c149ea561d44af3720c29236df"
},
"pdf_sha256": "401d162a87d4454b0f16998bf3501b9c84da909a7b5be353705528477d2eef62"
}