| { |
| "manifest_version": 1, |
| "role": "reference_parser", |
| "name": "reference-parser-mmbert-small-v7-f1", |
| "version": "sha256-90111b0a27dd", |
| "task": "token_classification", |
| "runtime": { |
| "engine": "onnxruntime", |
| "requires": [ |
| "onnxruntime", |
| "tokenizers", |
| "numpy" |
| ], |
| "opset": 17, |
| "executes_code_on_load": false |
| }, |
| "files": [ |
| { |
| "path": "onnx/bibr_onnx.json", |
| "bytes": 34047, |
| "sha256": "d28762a27475c98532590295ff266d40161d9c162332ebeb82dd56214d865c81", |
| "kind": "legacy_manifest" |
| }, |
| { |
| "path": "onnx/model.onnx", |
| "bytes": 268711556, |
| "sha256": "90111b0a27ddb3fc2f7ce647c87c55f1ecb78dd102963d8634c3a9b309e600fa", |
| "kind": "model" |
| }, |
| { |
| "path": "onnx/tokenizer.json", |
| "bytes": 34363188, |
| "sha256": "609d8f4c067cd3950f88594c5a802616cea245823836ef5848ee4fc40aab5b6f", |
| "kind": "tokenizer" |
| } |
| ], |
| "encoder": { |
| "id": "jhu-clsp/mmBERT-small", |
| "revision": null |
| }, |
| "labels": { |
| "tags": [ |
| "O", |
| "B-AUTHOR", |
| "I-AUTHOR", |
| "B-TITLE", |
| "I-TITLE", |
| "B-CONTAINER", |
| "I-CONTAINER", |
| "B-YEAR", |
| "I-YEAR", |
| "B-VOLUME", |
| "I-VOLUME", |
| "B-ISSUE", |
| "I-ISSUE", |
| "B-PAGES", |
| "I-PAGES", |
| "B-DOI", |
| "I-DOI", |
| "B-URL", |
| "I-URL", |
| "B-ARXIV", |
| "I-ARXIV", |
| "B-PMID", |
| "I-PMID", |
| "B-PUBLISHER", |
| "I-PUBLISHER", |
| "B-EDITOR", |
| "I-EDITOR", |
| "B-EDITION", |
| "I-EDITION", |
| "B-SERIES", |
| "I-SERIES", |
| "B-ACCESS_DATE", |
| "I-ACCESS_DATE", |
| "B-NOTE", |
| "I-NOTE", |
| "B-PAGE_RANGE_START", |
| "I-PAGE_RANGE_START", |
| "B-PAGE_RANGE_END", |
| "I-PAGE_RANGE_END" |
| ] |
| }, |
| "input": { |
| "layout": "single_text", |
| "truncation": "right", |
| "add_special_tokens": false, |
| "max_length": 256, |
| "special_tokens": { |
| "pad_token": "<pad>", |
| "pad_token_id": 0 |
| }, |
| "tokenizer_file": "onnx/tokenizer.json", |
| "window": null, |
| "stride": null, |
| "template": null, |
| "normalization": null, |
| "pair": null, |
| "features": null, |
| "model_inputs": [ |
| { |
| "name": "input_ids", |
| "dtype": "int64", |
| "shape": [ |
| "batch", |
| "sequence" |
| ] |
| }, |
| { |
| "name": "attention_mask", |
| "dtype": "int64", |
| "shape": [ |
| "batch", |
| "sequence" |
| ] |
| } |
| ], |
| "model_outputs": [ |
| { |
| "name": "emissions", |
| "dtype": "float32", |
| "shape": [ |
| "batch", |
| "sequence", |
| 39 |
| ] |
| } |
| ] |
| }, |
| "decode": { |
| "method": "crf_viterbi", |
| "crf": { |
| "file": "onnx/bibr_onnx.json", |
| "key": "crf" |
| }, |
| "thresholds": {}, |
| "when_nothing_found": "all_outside", |
| "calibration": null |
| }, |
| "quantization": { |
| "weights": "int8_embeddings", |
| "note": "int8, per-row float scale" |
| }, |
| "provenance": { |
| "ledger_key": "reference-parser-mmbert-small-v7", |
| "run_record_sha256": "f2eebc39be8d540f1154fa328607543a6e0f1446e59d926074a050ed16f1a52b", |
| "platform_commit": "83ba758659302b123a0597448cf4e9f7728a53ee", |
| "registry_build": "2026-11-10T21:38:59Z#d8d427de3ea7", |
| "split_id": "corpus-split-v1@c6fc246571f4" |
| }, |
| "licence": "mit", |
| "compat": { |
| "bibr_min_version": null, |
| "export_schema": null |
| }, |
| "self_test": { |
| "tolerance": { |
| "probability_abs": 0.01, |
| "logit_abs": 0.05, |
| "reason": "onnxruntime builds, CPUs and BLAS kernels give slightly different float results; the largest difference measured between onnxruntime 1.27.0 and 1.31.0 on the eight bundles normalised so far was 8e-05 (stored values are rounded to 5 places), so these limits are about two orders of magnitude looser than the drift seen, and far tighter than any change of a decision; tags and label lists are compared exactly" |
| }, |
| "cases": [ |
| { |
| "name": "reference 1", |
| "input": { |
| "text": "Marlowe, J. K., & Okonkwo, T. (2018). Tidal patterns in invented estuaries. Journal of Fictional Hydrology, 14(2), 101-117. https://doi.org/10.0000/example.2018.14" |
| }, |
| "expect": { |
| "tokens": 69, |
| "tags": [ |
| "B-AUTHOR", |
| "I-AUTHOR", |
| "I-AUTHOR", |
| "I-AUTHOR", |
| "I-AUTHOR", |
| "I-AUTHOR", |
| "I-AUTHOR", |
| "I-AUTHOR", |
| "I-AUTHOR", |
| "I-AUTHOR", |
| "I-AUTHOR", |
| "I-AUTHOR", |
| "O", |
| "O", |
| "B-YEAR", |
| "I-YEAR", |
| "I-YEAR", |
| "I-YEAR", |
| "O", |
| "B-TITLE", |
| "I-TITLE", |
| "I-TITLE", |
| "I-TITLE", |
| "I-TITLE", |
| "O", |
| "B-CONTAINER", |
| "I-CONTAINER", |
| "I-CONTAINER", |
| "I-CONTAINER", |
| "I-CONTAINER", |
| "O", |
| "O", |
| "B-VOLUME", |
| "I-VOLUME", |
| "O", |
| "B-ISSUE", |
| "O", |
| "O", |
| "B-PAGE_RANGE_START", |
| "I-PAGE_RANGE_START", |
| "I-PAGE_RANGE_START", |
| "O", |
| "B-PAGE_RANGE_END", |
| "I-PAGE_RANGE_END", |
| "I-PAGE_RANGE_END", |
| "O", |
| "B-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL", |
| "I-URL" |
| ], |
| "logits": [ |
| [ |
| -213.23517, |
| -214.81645, |
| -207.72357, |
| -223.91331, |
| -209.72394, |
| -222.92453, |
| -210.36705, |
| -221.49272, |
| -211.90256, |
| -220.74118, |
| -211.88834, |
| -220.79301, |
| -212.97563, |
| -221.15988, |
| -211.10245, |
| -233.41255, |
| -210.36565, |
| -228.62802, |
| -210.66747, |
| -232.88083, |
| -211.13853, |
| -226.67072, |
| -211.25479, |
| -220.94482, |
| -211.11005, |
| -221.70401, |
| -210.74956, |
| -226.28813, |
| -212.46799, |
| -227.15686, |
| -211.37155, |
| -223.93962, |
| -212.32109, |
| -223.19925, |
| -211.5863, |
| -221.28754, |
| -211.35732, |
| -221.90063, |
| -212.64578 |
| ], |
| [ |
| -210.89044, |
| -226.61595, |
| -210.88315, |
| -227.88895, |
| -211.82895, |
| -226.43167, |
| -212.69856, |
| -223.20265, |
| -213.75299, |
| -224.10251, |
| -213.91539, |
| -222.56773, |
| -213.87148, |
| -224.31747, |
| -212.55251, |
| -235.98375, |
| -212.32094, |
| -231.16647, |
| -211.89937, |
| -237.86696, |
| -212.82118, |
| -227.32922, |
| -211.88637, |
| -224.14027, |
| -212.93463, |
| -227.45416, |
| -213.06108, |
| -226.21512, |
| -212.7144, |
| -230.35287, |
| -212.64566, |
| -226.30988, |
| -211.58527, |
| -225.53059, |
| -212.0025, |
| -223.73209, |
| -213.07523, |
| -223.98383, |
| -214.14764 |
| ] |
| ] |
| } |
| }, |
| { |
| "name": "reference 2", |
| "input": { |
| "text": "Brandt, E. Notes on the Ostrava lighthouse ledger. Imaginary Press, Lisbon, 1977." |
| }, |
| "expect": { |
| "tokens": 24, |
| "tags": [ |
| "B-AUTHOR", |
| "I-AUTHOR", |
| "I-AUTHOR", |
| "O", |
| "B-TITLE", |
| "I-TITLE", |
| "I-TITLE", |
| "I-TITLE", |
| "I-TITLE", |
| "I-TITLE", |
| "I-TITLE", |
| "O", |
| "B-PUBLISHER", |
| "I-PUBLISHER", |
| "I-PUBLISHER", |
| "O", |
| "O", |
| "O", |
| "O", |
| "B-YEAR", |
| "I-YEAR", |
| "I-YEAR", |
| "I-YEAR", |
| "O" |
| ], |
| "logits": [ |
| [ |
| -212.99773, |
| -211.18358, |
| -206.37553, |
| -221.91766, |
| -208.38661, |
| -222.02475, |
| -209.13133, |
| -221.18004, |
| -210.62602, |
| -220.11, |
| -210.98463, |
| -220.18121, |
| -211.76031, |
| -220.50638, |
| -209.86336, |
| -232.37166, |
| -208.87422, |
| -227.93137, |
| -209.41269, |
| -231.90529, |
| -209.78856, |
| -226.26849, |
| -210.24641, |
| -219.94896, |
| -209.94037, |
| -220.83743, |
| -209.59392, |
| -226.59508, |
| -211.48244, |
| -227.07428, |
| -209.93088, |
| -223.89554, |
| -211.40326, |
| -223.59013, |
| -210.3266, |
| -221.1261, |
| -211.50865, |
| -220.23767, |
| -211.63181 |
| ], |
| [ |
| -210.83913, |
| -226.33218, |
| -211.28046, |
| -226.18927, |
| -212.05051, |
| -225.98215, |
| -212.60696, |
| -223.65367, |
| -213.82381, |
| -223.65247, |
| -213.90894, |
| -222.28218, |
| -213.66322, |
| -224.25191, |
| -212.56088, |
| -235.02065, |
| -211.59062, |
| -231.3573, |
| -211.33801, |
| -237.86525, |
| -212.58578, |
| -227.12978, |
| -211.84647, |
| -224.0826, |
| -212.99342, |
| -227.82146, |
| -213.24168, |
| -226.36017, |
| -212.55092, |
| -231.03183, |
| -212.46497, |
| -226.61258, |
| -211.36986, |
| -225.78674, |
| -211.59822, |
| -223.5401, |
| -213.85986, |
| -222.71497, |
| -213.9064 |
| ] |
| ] |
| } |
| } |
| ] |
| }, |
| "spec": { |
| "input_unit": "reference_string", |
| "list_number_stripped": true |
| } |
| } |
|
|
