Hugging Face's logo

failed09
/
bashkir-roberta-token-lid

Token Classification
ONNX
Bashkir
Russian
Tatar
onnxruntime
bashkir
tatar
russian
code-switching
language-identification
low-resource
experimental
Model card Files
xet
 Community
bashkir-roberta-token-lid
File size: 4,956 Bytes
818b1b2
 
 
 
 
 
 
 
 
 
 
 
 
 
fcc552c
818b1b2
 
 
 
 
 
 
 
 
 
 
 
 
fcc552c
818b1b2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
{
  "id": "bashkir-roberta-token-lid",
  "version": "v1.0-experimental",
  "status": "experimental_release_candidate",
  "type": "token_classification",
  "task": "Word-level Bashkir, Russian, Tatar and other-language identification",
  "language": [
    "ba",
    "ru",
    "tt"
  ],
  "license": "apache-2.0",
  "base_model": "failed09/bashkir-roberta",
  "source": "Fine-tuned BashkirRoBERTa encoder with multilingual and conversational token annotations; original texts are not redistributed.",
  "annotation_provenance": "Rule-based annotations, curated multilingual conversational samples and linguistic review; not independently human-verified gold.",
  "created_at": "2026-10-01T18:08:35.195621+00:00",
  "labels": [
    "O",
    "BA",
    "RU",
    "TAT",
    "OTHER"
  ],
  "default_artifact": "onnx/model_int8.onnx",
  "source_policy": "Only inference models, tokenizer, runtime, aggregate metrics and illustrative usage examples are included. Training data, annotation batches and evaluation corpora are excluded.",
  "tokenizer_metadata": "Training input paths and output prefix removed; vocabulary and inference parameters preserved.",
  "evaluation": {
    "summary_file": "benchmark_summary.json",
    "annotation_status": "Linguistically curated evaluation suites and reviewed conversational rubrics; not independent expert gold",
    "primary_authored_suite": {
      "fp32": {
        "cases": 40,
        "strict_lexical_tokens": 344,
        "ambiguous_tokens": 0,
        "strict_lexical_accuracy": 0.936046511627907,
        "exact_case_accuracy": 0.75,
        "high_confidence_errors": 16,
        "per_class": {
          "O": {
            "precision": 1.0,
            "recall": 1.0,
            "f1": 1.0,
            "support": 104
          },
          "BA": {
            "precision": 0.925,
            "recall": 0.9098360655737705,
            "f1": 0.9173553719008265,
            "support": 122
          },
          "RU": {
            "precision": 0.9736842105263158,
            "recall": 0.9946236559139785,
            "f1": 0.9840425531914894,
            "support": 186
          },
          "TAT": {
            "precision": 0.8125,
            "recall": 0.7222222222222222,
            "f1": 0.7647058823529411,
            "support": 36
          },
          "OTHER": {
            "precision": 0.0,
            "recall": 0,
            "f1": 0,
            "support": 0
          }
        }
      },
      "int8": {
        "cases": 40,
        "strict_lexical_tokens": 344,
        "ambiguous_tokens": 0,
        "strict_lexical_accuracy": 0.9476744186046512,
        "exact_case_accuracy": 0.825,
        "high_confidence_errors": 16,
        "per_class": {
          "O": {
            "precision": 1.0,
            "recall": 1.0,
            "f1": 1.0,
            "support": 104
          },
          "BA": {
            "precision": 0.9333333333333333,
            "recall": 0.9180327868852459,
            "f1": 0.9256198347107439,
            "support": 122
          },
          "RU": {
            "precision": 0.9789473684210527,
            "recall": 1.0,
            "f1": 0.9893617021276596,
            "support": 186
          },
          "TAT": {
            "precision": 0.8235294117647058,
            "recall": 0.7777777777777778,
            "f1": 0.7999999999999999,
            "support": 36
          },
          "OTHER": {
            "precision": 0,
            "recall": 0,
            "f1": 0,
            "support": 0
          }
        }
      }
    }
  },
  "artifacts": {
    "LICENSE": {
      "bytes": 11358,
      "sha256": "cfc7749b96f63bd31c3c42b5c471bf756814053e847c10f3eb003417bc523d30"
    },
    "benchmark_summary.json": {
      "bytes": 8774,
      "sha256": "4aa76ec74c6eb668aec0d11091b738f55adb97263edb6c200162c56c9ebaea35"
    },
    "config.json": {
      "bytes": 993,
      "sha256": "8fd417e386da064d89425613062a4780b76df9edbf9911fcc34a3f6d7f3b5181"
    },
    "example.py": {
      "bytes": 718,
      "sha256": "339991327f3f6411b0553da42ff79e60e685f0f982fb9ebf48279e30f7337738"
    },
    "onnx/model_fp32.onnx": {
      "bytes": 200280944,
      "sha256": "5d8fdd658aa3a26f3d926542b8fc52b8662738369d841382de0f817022381e45"
    },
    "onnx/model_int8.onnx": {
      "bytes": 82388631,
      "sha256": "5240d1f1dddb7973e93581ad55fe53b5d0d7f6195ccd7f91087279a5a95a7d50"
    },
    "requirements.txt": {
      "bytes": 87,
      "sha256": "e1b0f7409a8c01fc591b70f26fc2dad1a2f1cfeec57cfd7989acc4530bf517f9"
    },
    "runtime.py": {
      "bytes": 5257,
      "sha256": "787c70621d1dd8e33aa07dcda50ebc46b8a8b43374eea0738c360b572f67edf8"
    },
    "spm_bashkir_bert_16k.model": {
      "bytes": 586856,
      "sha256": "0f0b8ebe2da787dd6adafc6d74148c7726ff85156970227e5ddbc7f6f4b532af"
    }
  }
}