Publish onnx-android artifact

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+model.onnx.data filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,47 @@
+---
+license: apache-2.0
+pipeline_tag: token-classification
+library_name: openmed
+tags:
+- openmed
+- medical-nlp
+---
+
+# OpenMed-NER-ProteinDetect-SnowMed-568M-v1-onnx-android
+
+This model card is rendered from the OpenMed model manifest. Update `models.jsonl` and rerun the publish step instead of editing this file directly.
+
+## Manifest Summary
+
+| Field | Value |
+|---|---|
+| Repository | `OpenMed/OpenMed-NER-ProteinDetect-SnowMed-568M-v1-onnx-android` |
+| Family | NER |
+| Task | token-classification |
+| Languages | en |
+| Tier | Not specified |
+| Parameters | 568M (568,000,000) |
+| Architecture | xlm-roberta |
+| Base model | OpenMed/OpenMed-NER-ProteinDetect-SnowMed-568M |
+| Formats | onnx-android, int8 |
+| License | apache-2.0 |
+| arXiv | [arXiv:2508.01630](https://arxiv.org/abs/2508.01630) |
+| Reproducibility hash | `sha256:52980c11e54355be05351d881b8e8fd8f2ef8f99d2d2ebc0f5ee7cb86e67ee68` |
+| Released | 2026-07-12 |
+
+## Benchmark
+
+| Dataset | Micro F1 | Recall |
+|---|---:|---:|
+| Not reported | Not reported | Not reported |
+
+## Canonical Labels
+
+`protein`, `protein_complex`, `protein_enum`, `protein_familiy_or_group`, `protein_variant`
+
+## Artifact Format
+
+| Field | Value |
+|---|---|
+| Runtime artifacts | onnx-android |
+| Quantization | int8 |

config.json

@@ -0,0 +1,67 @@
+{
+  "transformers_version": "5.12.1",
+  "architectures": [
+    "XLMRobertaForTokenClassification"
+  ],
+  "output_hidden_states": false,
+  "return_dict": true,
+  "dtype": "bfloat16",
+  "chunk_size_feed_forward": 0,
+  "is_encoder_decoder": false,
+  "id2label": {
+    "0": "B-protein",
+    "1": "B-protein_complex",
+    "2": "B-protein_enum",
+    "3": "B-protein_familiy_or_group",
+    "4": "B-protein_variant",
+    "5": "I-protein",
+    "6": "I-protein_complex",
+    "7": "I-protein_enum",
+    "8": "I-protein_familiy_or_group",
+    "9": "I-protein_variant",
+    "10": "O"
+  },
+  "label2id": {
+    "B-protein": 0,
+    "B-protein_complex": 1,
+    "B-protein_enum": 2,
+    "B-protein_familiy_or_group": 3,
+    "B-protein_variant": 4,
+    "I-protein": 5,
+    "I-protein_complex": 6,
+    "I-protein_enum": 7,
+    "I-protein_familiy_or_group": 8,
+    "I-protein_variant": 9,
+    "O": 10
+  },
+  "problem_type": null,
+  "vocab_size": 250002,
+  "hidden_size": 1024,
+  "num_hidden_layers": 24,
+  "num_attention_heads": 16,
+  "intermediate_size": 4096,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.2,
+  "attention_probs_dropout_prob": 0.2,
+  "max_position_embeddings": 8194,
+  "type_vocab_size": 1,
+  "initializer_range": 0.02,
+  "layer_norm_eps": 1e-07,
+  "pad_token_id": 1,
+  "bos_token_id": 0,
+  "eos_token_id": 2,
+  "use_cache": true,
+  "classifier_dropout": 0.2,
+  "is_decoder": false,
+  "add_cross_attention": false,
+  "tie_word_embeddings": true,
+  "_name_or_path": "OpenMed/OpenMed-NER-ProteinDetect-SnowMed-568M",
+  "matryoshka_dimensions": [
+    256
+  ],
+  "model_type": "xlm-roberta",
+  "output_past": true,
+  "position_embedding_type": "absolute",
+  "output_attentions": false,
+  "max_sequence_length": 512
+}

id2label.json

@@ -0,0 +1,13 @@
+{
+  "0": "B-protein",
+  "1": "B-protein_complex",
+  "2": "B-protein_enum",
+  "3": "B-protein_familiy_or_group",
+  "4": "B-protein_variant",
+  "5": "I-protein",
+  "6": "I-protein_complex",
+  "7": "I-protein_enum",
+  "8": "I-protein_familiy_or_group",
+  "9": "I-protein_variant",
+  "10": "O"
+}

model.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91d46e6a4ecce8a64d03c6909f5b924ff9de31012a34dd03c4a73148276c7483
+size 633271

model.onnx.data

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1029ee5408de4e4649bd0c0a9a8e921acc6aca598ca6a4e6148750ba44518a71
+size 2266865708

model_fp16.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23d149e2a5fa465b3a9de14b6a29ee7a71a5b4e7c186b38eace3b27a39f66f43
+size 1134233581

model_int8.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2fc3e84263b50e65e7234180996df8b036bedb864d4548169446f89d91bcef9a
+size 1361883438

openmed-onnx.json

@@ -0,0 +1,430 @@
+{
+  "format": "openmed-onnx",
+  "format_version": 1,
+  "formats": [
+    "onnx-android",
+    "onnx-int8"
+  ],
+  "task": "token-classification",
+  "family": "xlm-roberta",
+  "source_model_id": "OpenMed/OpenMed-NER-ProteinDetect-SnowMed-568M",
+  "config_path": "config.json",
+  "label_map_path": "id2label.json",
+  "artifacts": [
+    {
+      "format": "onnx-android",
+      "path": "model.onnx",
+      "precision": "float32",
+      "metadata": {
+        "profile": "android",
+        "opset": 18,
+        "execution_providers": [
+          "NNAPI",
+          "XNNPACK"
+        ],
+        "inputs": [
+          {
+            "name": "input_ids",
+            "dtype": "int64",
+            "axes": [
+              "batch",
+              "sequence"
+            ],
+            "shape": [
+              {
+                "kind": "dynamic",
+                "name": "batch"
+              },
+              {
+                "kind": "dynamic",
+                "name": "sequence"
+              }
+            ]
+          },
+          {
+            "name": "attention_mask",
+            "dtype": "int64",
+            "axes": [
+              "batch",
+              "sequence"
+            ],
+            "shape": [
+              {
+                "kind": "dynamic",
+                "name": "batch"
+              },
+              {
+                "kind": "dynamic",
+                "name": "sequence"
+              }
+            ]
+          }
+        ],
+        "outputs": [
+          {
+            "name": "logits",
+            "dtype": "float32",
+            "axes": [
+              "batch",
+              "sequence",
+              "labels"
+            ],
+            "shape": [
+              {
+                "kind": "dynamic",
+                "name": "batch"
+              },
+              {
+                "kind": "dynamic",
+                "name": "sequence"
+              },
+              {
+                "kind": "static",
+                "value": 11
+              }
+            ]
+          }
+        ],
+        "operators": [
+          "Constant",
+          "Equal",
+          "Not",
+          "Cast",
+          "CumSum",
+          "Mul",
+          "Add",
+          "Shape",
+          "Gather",
+          "Unsqueeze",
+          "Concat",
+          "Reshape",
+          "ConstantOfShape",
+          "Where",
+          "Expand",
+          "GatherElements",
+          "LayerNormalization",
+          "Range",
+          "GreaterOrEqual",
+          "And",
+          "Flatten",
+          "MatMul",
+          "Transpose",
+          "Softmax",
+          "IsNaN",
+          "Div",
+          "Erf"
+        ],
+        "unsupported_ops": [
+          "CumSum",
+          "GreaterOrEqual",
+          "Flatten",
+          "IsNaN"
+        ],
+        "warnings": [
+          "CumSum is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set",
+          "GreaterOrEqual is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set",
+          "Flatten is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set",
+          "IsNaN is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set"
+        ]
+      }
+    },
+    {
+      "format": "onnx-android",
+      "path": "model_fp16.onnx",
+      "precision": "float16",
+      "metadata": {
+        "profile": "android",
+        "opset": 18,
+        "execution_providers": [
+          "NNAPI",
+          "XNNPACK"
+        ],
+        "inputs": [
+          {
+            "name": "input_ids",
+            "dtype": "int64",
+            "axes": [
+              "batch",
+              "sequence"
+            ],
+            "shape": [
+              {
+                "kind": "dynamic",
+                "name": "batch"
+              },
+              {
+                "kind": "dynamic",
+                "name": "sequence"
+              }
+            ]
+          },
+          {
+            "name": "attention_mask",
+            "dtype": "int64",
+            "axes": [
+              "batch",
+              "sequence"
+            ],
+            "shape": [
+              {
+                "kind": "dynamic",
+                "name": "batch"
+              },
+              {
+                "kind": "dynamic",
+                "name": "sequence"
+              }
+            ]
+          }
+        ],
+        "outputs": [
+          {
+            "name": "logits",
+            "dtype": "float32",
+            "axes": [
+              "batch",
+              "sequence",
+              "labels"
+            ],
+            "shape": [
+              {
+                "kind": "dynamic",
+                "name": "batch"
+              },
+              {
+                "kind": "dynamic",
+                "name": "sequence"
+              },
+              {
+                "kind": "static",
+                "value": 11
+              }
+            ]
+          }
+        ],
+        "operators": [
+          "Constant",
+          "Equal",
+          "Not",
+          "Cast",
+          "CumSum",
+          "Mul",
+          "Add",
+          "Shape",
+          "Gather",
+          "Unsqueeze",
+          "Concat",
+          "Reshape",
+          "ConstantOfShape",
+          "Where",
+          "Expand",
+          "GatherElements",
+          "LayerNormalization",
+          "Range",
+          "GreaterOrEqual",
+          "And",
+          "Flatten",
+          "MatMul",
+          "Transpose",
+          "Softmax",
+          "IsNaN",
+          "Div",
+          "Erf"
+        ],
+        "unsupported_ops": [
+          "CumSum",
+          "GreaterOrEqual",
+          "Flatten",
+          "IsNaN"
+        ],
+        "warnings": [
+          "CumSum is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set",
+          "GreaterOrEqual is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set",
+          "Flatten is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set",
+          "IsNaN is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set"
+        ]
+      }
+    },
+    {
+      "format": "onnx-int8",
+      "path": "model_int8.onnx",
+      "precision": "int8",
+      "metadata": {
+        "format": "onnx-int8",
+        "quantization": {
+          "scheme": "dynamic",
+          "weight_type": "qint8",
+          "op_types": [
+            "MatMul",
+            "Gemm"
+          ],
+          "calibration": "none",
+          "target": "arm64-cpu",
+          "output": "model_int8.onnx"
+        },
+        "profile": "android",
+        "opset": 18,
+        "execution_providers": [
+          "NNAPI",
+          "XNNPACK"
+        ],
+        "inputs": [
+          {
+            "name": "input_ids",
+            "dtype": "int64",
+            "axes": [
+              "batch",
+              "sequence"
+            ],
+            "shape": [
+              {
+                "kind": "dynamic",
+                "name": "batch"
+              },
+              {
+                "kind": "dynamic",
+                "name": "sequence"
+              }
+            ]
+          },
+          {
+            "name": "attention_mask",
+            "dtype": "int64",
+            "axes": [
+              "batch",
+              "sequence"
+            ],
+            "shape": [
+              {
+                "kind": "dynamic",
+                "name": "batch"
+              },
+              {
+                "kind": "dynamic",
+                "name": "sequence"
+              }
+            ]
+          }
+        ],
+        "outputs": [
+          {
+            "name": "logits",
+            "dtype": "float32",
+            "axes": [
+              "batch",
+              "sequence",
+              "labels"
+            ],
+            "shape": [
+              {
+                "kind": "dynamic",
+                "name": "batch"
+              },
+              {
+                "kind": "dynamic",
+                "name": "sequence"
+              },
+              {
+                "kind": "static",
+                "value": 11
+              }
+            ]
+          }
+        ],
+        "operators": [
+          "Constant",
+          "Cast",
+          "Shape",
+          "Gather",
+          "Equal",
+          "ConstantOfShape",
+          "Flatten",
+          "Not",
+          "Unsqueeze",
+          "Concat",
+          "CumSum",
+          "Range",
+          "Reshape",
+          "Mul",
+          "Add",
+          "Where",
+          "Expand",
+          "GatherElements",
+          "LayerNormalization",
+          "DynamicQuantizeLinear",
+          "MatMulInteger",
+          "Transpose",
+          "GreaterOrEqual",
+          "MatMul",
+          "And",
+          "Softmax",
+          "IsNaN",
+          "Div",
+          "Erf"
+        ],
+        "unsupported_ops": [
+          "Flatten",
+          "CumSum",
+          "DynamicQuantizeLinear",
+          "MatMulInteger",
+          "GreaterOrEqual",
+          "IsNaN"
+        ],
+        "warnings": [
+          "Flatten is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set",
+          "CumSum is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set",
+          "DynamicQuantizeLinear is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set",
+          "MatMulInteger is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set",
+          "GreaterOrEqual is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set",
+          "IsNaN is not in the Android ONNX Runtime Mobile NNAPI/XNNPACK safe set"
+        ]
+      }
+    }
+  ],
+  "minimum_opset": 18,
+  "max_sequence_length": 512,
+  "dynamic_shapes": {
+    "batch_axis": "dynamic",
+    "sequence_axis": "dynamic",
+    "shape_buckets": {
+      "strategy": "next_configured_bucket_or_exact",
+      "axes": {
+        "input": [
+          "batch",
+          "sequence"
+        ],
+        "output": [
+          "batch",
+          "sequence"
+        ]
+      },
+      "buckets": [
+        8,
+        16,
+        32,
+        64,
+        128,
+        256,
+        512,
+        1024,
+        2048
+      ],
+      "min_length": 8,
+      "max_length": 2048,
+      "overflow": "exact_length"
+    }
+  },
+  "optimization": {
+    "enabled": false,
+    "minimum_opset": 18
+  },
+  "validation": {},
+  "operator_fallbacks": [],
+  "tokenizer": {
+    "path": ".",
+    "files": [
+      "tokenizer.json",
+      "tokenizer_config.json"
+    ]
+  }
+}

tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ffb37461c391f096759f4a9bbbc329da0f36952f88bab061fcf84940c022e98
+size 17082999

tokenizer_config.json

@@ -0,0 +1,23 @@
+{
+  "add_prefix_space": true,
+  "backend": "tokenizers",
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "<s>",
+  "eos_token": "</s>",
+  "is_local": false,
+  "local_files_only": false,
+  "mask_token": "<mask>",
+  "max_length": 512,
+  "model_max_length": 8192,
+  "pad_to_multiple_of": null,
+  "pad_token": "<pad>",
+  "pad_token_type_id": 0,
+  "padding_side": "right",
+  "sep_token": "</s>",
+  "stride": 0,
+  "tokenizer_class": "XLMRobertaTokenizer",
+  "truncation_side": "right",
+  "truncation_strategy": "longest_first",
+  "unk_token": "<unk>"
+}