diff --git a/docs/catalogue/models-catalogue.json b/docs/catalogue/models-catalogue.json new file mode 100644 index 00000000..657d607a --- /dev/null +++ b/docs/catalogue/models-catalogue.json @@ -0,0 +1,806 @@ +{ + "generated_note": "generated from models.yaml + index-v12; regenerate, never edit", + "models": [ + { + "id": "ara-diac-1.0", + "task": "diacritization", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 2780311342, + "sha256": "9e6d9777fe2474b150450774aaa1c9e8a9051b502b184fde298b91636dcfdac6", + "metrics": [ + { + "name": "der_total_greedy", + "value": 2.5793, + "protocol": null, + "anchor": null + }, + { + "name": "der_morph_greedy", + "value": 1.5317, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-2.0", + "task": "diacritization", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 2780299051, + "sha256": "872c655ec6750bf44e5986f958de26767eafa13f8e0a7c64340bc35fd8a4550e", + "metrics": [ + { + "name": "der_total_greedy", + "value": 2.2864, + "protocol": null, + "anchor": null + }, + { + "name": "der_morph_greedy", + "value": 1.3343, + "protocol": null, + "anchor": null + }, + { + "name": "wer_wikinews_multiref", + "value": 17.3794, + "protocol": null, + "anchor": null + }, + { + "name": "der_wikinews_multiref", + "value": 11.8273, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-2.0-int8", + "task": "diacritization", + "precision": "int8", + "license": "BSD-3-Clause", + "size_bytes": 519207340, + "sha256": "7f1d55d43f97564ad62fc14a7e0cabe52c01d8298b747677a75900de748bb6c6", + "metrics": [ + { + "name": "der_total_greedy", + "value": 2.2864, + "protocol": null, + "anchor": null + }, + { + "name": "der_morph_greedy", + "value": 1.3343, + "protocol": null, + "anchor": null + }, + { + "name": "wer_wikinews_multiref", + "value": 17.3794, + "protocol": null, + "anchor": null + }, + { + "name": "der_wikinews_multiref", + "value": 11.8273, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-layerdrop-1.0", + "task": "diacritization", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 1015218371, + "sha256": "8f297bdc0e7eba5160c6ed8a2a437abbd399687fd14f2f2b640a4b4c639a3d16", + "metrics": [ + { + "name": "der_teacher_fullset", + "value": 2.2921, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_fullset", + "value": 5.784, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-layerdrop-1.0-int4", + "task": "diacritization", + "precision": "int4", + "license": "BSD-3-Clause", + "size_bytes": 145195702, + "sha256": "352e02105645a6e283e2c3602505bea0e90085b741666fd863e478e4835e7bc9", + "metrics": [ + { + "name": "der_teacher_fullset", + "value": 2.2921, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_fullset", + "value": 5.784, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-layerdrop-1.0-int8", + "task": "diacritization", + "precision": "int8", + "license": "BSD-3-Clause", + "size_bytes": 191603048, + "sha256": "68dcc06cd8861e017c4d4912ce3e1fe057d78b79b0a7b6246ae8603c99930963", + "metrics": [ + { + "name": "der_teacher_fullset", + "value": 2.2921, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_fullset", + "value": 5.784, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-news-1.0", + "task": "diacritization", + "precision": "int8", + "license": "BSD-3-Clause", + "size_bytes": 399018923, + "sha256": "e5a52b1530dfa6d6c7ef8970443206ce33af865d0337a58d145945adab33484e", + "metrics": [ + { + "name": "wer_wikinews2024_multiref", + "value": 4.0976, + "protocol": null, + "anchor": null + }, + { + "name": "der_wikinews2024_multiref", + "value": 1.5606, + "protocol": null, + "anchor": null + }, + { + "name": "der_total_greedy", + "value": 5.4814, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-news-1.1", + "task": "diacritization", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 2160091218, + "sha256": "2f55e2a3c9e97da3666ef236e8ba31a24399257c9cc448363cc19f96541eb3ba", + "metrics": [ + { + "name": "wer_wikinews2024_multiref", + "value": 8.85, + "protocol": "greedy decode, first-alternate protocol, WikiNews-2024 multiref (WO35)", + "anchor": null + }, + { + "name": "der_total_greedy", + "value": 2.84, + "protocol": "greedy decode, first-alternate protocol, WikiNews-2024 multiref (WO35)", + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-plane-1.0", + "task": "diacritization", + "precision": "int8", + "license": "BSD-3-Clause", + "size_bytes": 219076965, + "sha256": "1310204e36acc4fcbb988e9f9c46cf6422e7ea3860bd467fb469ea0ad4d39ef3", + "metrics": [ + { + "name": "der_total_greedy", + "value": 3.5905, + "protocol": null, + "anchor": null + }, + { + "name": "der_morph_greedy", + "value": 2.2149, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-plane-large-1.0", + "task": "diacritization", + "precision": "int8", + "license": "BSD-3-Clause", + "size_bytes": 866052488, + "sha256": "a66df2e12ddc935e8ab59836fd45486f17ac40b764c1619c03b2b0e191608a52", + "metrics": [ + { + "name": "der_total_greedy", + "value": 2.7397, + "protocol": null, + "anchor": null + }, + { + "name": "der_morph_greedy", + "value": 1.6717, + "protocol": null, + "anchor": null + }, + { + "name": "wer_total_greedy", + "value": 8.6083, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-small-1.0", + "task": "diacritization", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 1418126835, + "sha256": "8b149beb04294a2e482af876a34ae0fca3e87412606116fae1a376845fd4111a", + "metrics": [ + { + "name": "der_teacher_greedy", + "value": 1.3205, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_greedy", + "value": 3.658, + "protocol": null, + "anchor": null + }, + { + "name": "der_teacher_fullset", + "value": 2.5815, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_fullset", + "value": 8.259, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-small-1.0-int8", + "task": "diacritization", + "precision": "int8", + "license": "BSD-3-Clause", + "size_bytes": 256959713, + "sha256": "af247d837bee7e755b5287102cfb715c48dc90c2cf41ec1c41e7238e1a72b678", + "metrics": [ + { + "name": "der_teacher_greedy", + "value": 1.3205, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_greedy", + "value": 3.658, + "protocol": null, + "anchor": null + }, + { + "name": "der_teacher_fullset", + "value": 2.5815, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_fullset", + "value": 8.259, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-small-2.0", + "task": "diacritization", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 1418516306, + "sha256": "d9aa95d0e8fd0d7da80fd2f7dabedf0ac65f55ba984de565d252f5e14906d98c", + "metrics": [ + { + "name": "der_teacher_fullset", + "value": 2.289, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_fullset", + "value": 5.0821, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-small-2.0-int8", + "task": "diacritization", + "precision": "int8", + "license": "BSD-3-Clause", + "size_bytes": 264043016, + "sha256": "f4d07acdaf9c18cded97e15e0c5636e6cee90de2fa32c5a373a44c0cc912bf50", + "metrics": [ + { + "name": "der_teacher_fullset", + "value": 2.289, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_fullset", + "value": 4.8218, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-small-2.1", + "task": "diacritization", + "precision": "fp32", + "license": null, + "size_bytes": 1419756840, + "sha256": "a69918de9f87664fd97cdab3bf68b655966f608211dbeadd7b6fb3f7b03e5ffb", + "metrics": [ + { + "name": "der_teacher_fullset", + "value": 2.289, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_fullset", + "value": 4.5701, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-small-2.1-fp16", + "task": "diacritization", + "precision": "fp16", + "license": "BSD-3-Clause", + "size_bytes": 706351903, + "sha256": "395ee72137d1cdfa9cd53bf167ad0d2f5fe5cbf9a5b62f118e0c9e0fa63c972d", + "metrics": [ + { + "name": "der_teacher_fullset", + "value": 2.289, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_fullset", + "value": 4.5701, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-small-2.1-int8", + "task": "diacritization", + "precision": "int8", + "license": "BSD-3-Clause", + "size_bytes": 267965657, + "sha256": "9b4e4aad2ab53a8cfbec432a78e27b409376627243aec3fd043f06574552a32b", + "metrics": [ + { + "name": "der_teacher_fullset", + "value": 2.289, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_fullset", + "value": 4.5701, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "ara-diac-small-int8static-2.1", + "task": "diacritization", + "precision": "int8", + "license": "BSD-3-Clause", + "size_bytes": 515362212, + "sha256": "7836bfd5582ad054ee419bfbd45a2b4a87fe8dc91e1f849898428faefd0c95da", + "metrics": [ + { + "name": "der_teacher_fullset", + "value": 2.289, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_fullset", + "value": 4.5701, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "fas-diac-base-1.0", + "task": "diac", + "precision": "fp16", + "license": "BSD-3-Clause", + "size_bytes": 711254463, + "sha256": "620b31ab96e6d7588d583fa5119406ea7c08506d483cc8ee253b3c479c6d87f0", + "metrics": [ + { + "name": "cer", + "value": 0.603, + "protocol": "greedy decode, 9288-row docs split, row-mean", + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "fas-g2p-1.0", + "task": "g2p", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 2780254270, + "sha256": "b53fd69196127efcc596935d24e135b3c897006872c1646c674e70f3c0d9a0e3", + "metrics": [ + { + "name": "cer", + "value": 1.6, + "protocol": null, + "anchor": null + }, + { + "name": "sb_homograph_ezafe_norm", + "value": 77.34, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "heb-diac-1.0", + "task": "diacritization", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 2781139753, + "sha256": "b32fa239e0c56708b6004942de4b9e6e9f265e7cc3828f41d560e96386e20dd8", + "metrics": [ + { + "name": "der_greedy", + "value": 29.0, + "protocol": null, + "anchor": null + }, + { + "name": "der_beam4", + "value": 17.46, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "heb-diac-1.1", + "task": "diacritization", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 2780987905, + "sha256": "f0022dde8d06e46e8f08cf0b2f7bbb887a05ed686b90f1d5aa82ca3ef690a0b0", + "metrics": [ + { + "name": "der_greedy", + "value": 16.44, + "protocol": null, + "anchor": null + }, + { + "name": "der_beam4", + "value": 16.43, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "heb-diac-plane-1.0", + "task": "diacritization", + "precision": "int8", + "license": "BSD-3-Clause", + "size_bytes": 219322007, + "sha256": "dd9ae811c60c38f28ead57bdefa022b281867401ee13e38563b990eaf7f81f95", + "metrics": [ + { + "name": "der_total_greedy", + "value": 12.48, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "heb-diac-plane-2.0", + "task": "diacritization", + "precision": "int8", + "license": "BSD-3-Clause", + "size_bytes": 415901226, + "sha256": "9c8f0432294ebae6abc847ddb79572c1ab979f4851f03d44a9bb6495ba767a9e", + "metrics": [ + { + "name": "der_total_greedy", + "value": 8.18, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "heb-diac-small-1.0", + "task": "diacritization", + "precision": "fp32", + "license": null, + "size_bytes": 1418663344, + "sha256": "905c53873d5a62bbe6252c6ec0ecef5f37e96e78cb01e8cd99b6b24c4d4d74ad", + "metrics": [ + { + "name": "der_teacher_greedy", + "value": 24.79, + "protocol": null, + "anchor": null + }, + { + "name": "der_student_greedy", + "value": 30.37, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "khm-latn-1.0", + "task": "translit", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 1418009977, + "sha256": "55993d473a2ed9489058779cad7e115db05e4020085b71616468cfae4f2f65cb", + "metrics": [ + { + "name": "cer", + "value": 27.42, + "protocol": null, + "anchor": null + }, + { + "name": "em", + "value": 59.66, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "tha-g2p-base-1.0", + "task": "g2p", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 2781174343, + "sha256": "f341a5868867a9847d3f897d1fcc4d1d63b9a59b8913871bcf4ecab48f4fa895", + "metrics": [ + { + "name": "per_teacher", + "value": 1.25, + "protocol": null, + "anchor": null + }, + { + "name": "per_student", + "value": 3.53, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "tha-g2p-small-1.0", + "task": "g2p", + "precision": "int8", + "license": null, + "size_bytes": 257468517, + "sha256": "f2cfa93a42d98aabf08fdc82dfc6bd1574e21993f36558aa581fbe29e0588a94", + "metrics": [ + { + "name": "per_teacher", + "value": 4.43, + "protocol": null, + "anchor": null + }, + { + "name": "per_student_beam4", + "value": 12.06, + "protocol": null, + "anchor": null + }, + { + "name": "per_student", + "value": 2.85, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "tha-g2p-small-1.0-int4", + "task": "g2p", + "precision": "int4", + "license": "BSD-3-Clause", + "size_bytes": 201958073, + "sha256": "e5d12a3d65c6ac39fa08f52c975c99ff7a51c3e7b7aca7fbb461d2de2345e4cb", + "metrics": [ + { + "name": "per_teacher", + "value": 4.43, + "protocol": null, + "anchor": null + }, + { + "name": "per_student_beam4", + "value": 12.06, + "protocol": null, + "anchor": null + }, + { + "name": "per_student", + "value": 2.85, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "urd-diac-1.0", + "task": "diacritization", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 1417851421, + "sha256": "c299ea44a4da4851a52283f8e42d2fb6f8c7a9c443ba7eb0270f97372e10b8ea", + "metrics": [ + { + "name": "cer", + "value": 3.74, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "urd-diac-base-1.0", + "task": "diacritization", + "precision": "fp16", + "license": "BSD-3-Clause", + "size_bytes": 1347488093, + "sha256": "6818469cbeeb58b822279c06e921669e4e79dc174d0b54c37a1be34687068252", + "metrics": [ + { + "name": "cer", + "value": 3.57, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "urd-g2p-1.0", + "task": "g2p", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 1418039605, + "sha256": "9f0a248c7116aa7f6b9ac125fb9684a3556f3b8ce8b64567c05a5e7e7a018ebf", + "metrics": [ + { + "name": "cer", + "value": 14.77, + "protocol": null, + "anchor": null + }, + { + "name": "em", + "value": 33.6, + "protocol": null, + "anchor": null + } + ], + "status": "shipped" + }, + { + "id": "urd-g2p-base-1.0", + "task": "g2p", + "precision": "fp32", + "license": "BSD-3-Clause", + "size_bytes": 2159598918, + "sha256": "d3e35dd435bf18fedec03d4a15bba097fca743737b729312fa08f4fdcac24282", + "metrics": [ + { + "name": "cer", + "value": 15.65, + "protocol": "greedy decode, full 12699-row test split, row-mean", + "anchor": null + }, + { + "name": "cer_normalized", + "value": 10.75, + "protocol": "greedy decode, full 12699-row test split, row-mean", + "anchor": null + } + ], + "status": "research-only" + } + ] +}