Hugging Face's logo

gliner-community
/
gliner_xxl-v2.5

Token Classification
GLiNER
PyTorch
multilingual
Model card Files Community
2
Ihor commited on
Commit
bc6836e
•
1 Parent(s): c1f1f58

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +287 -3
README.md CHANGED
@@ -1,3 +1,287 @@
1
- ---
2
- license: apache-2.0
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ language:
4
+ - multilingual
5
+ library_name: gliner
6
+ datasets:
7
+ - urchade/pile-mistral-v0.1
8
+ pipeline_tag: token-classification
9
+ ---
10
+
11
+ # About
12
+
13
+ GLiNER is a Named Entity Recognition (NER) model capable of identifying any entity type using a bidirectional transformer encoder (BERT-like). It provides a practical alternative to traditional NER models, which are limited to predefined entities, and Large Language Models (LLMs) that, despite their flexibility, are costly and large for resource-constrained scenarios.
14
+
15
+
16
+ ## Links
17
+
18
+ * Paper: https://arxiv.org/abs/2311.08526
19
+ * Repository: https://github.com/urchade/GLiNER
20
+
21
+ ## Installation
22
+ To use this model, you must install the GLiNER Python library:
23
+ ```
24
+ !pip install gliner -U
25
+ ```
26
+
27
+ ## Usage
28
+ Once you've downloaded the GLiNER library, you can import the GLiNER class. You can then load this model using `GLiNER.from_pretrained` and predict entities with `predict_entities`.
29
+
30
+ ```python
31
+ from gliner import GLiNER
32
+
33
+ model = GLiNER.from_pretrained("gliner-community/gliner_xxl-v2.5", load_tokenizer=True)
34
+
35
+ text = """
36
+ Cristiano Ronaldo dos Santos Aveiro (Portuguese pronunciation: [kɾiʃˈtjɐnu ʁɔˈnaldu]; born 5 February 1985) is a Portuguese professional footballer who plays as a forward for and captains both Saudi Pro League club Al Nassr and the Portugal national team. Widely regarded as one of the greatest players of all time, Ronaldo has won five Ballon d'Or awards,[note 3] a record three UEFA Men's Player of the Year Awards, and four European Golden Shoes, the most by a European player. He has won 33 trophies in his career, including seven league titles, five UEFA Champions Leagues, the UEFA European Championship and the UEFA Nations League. Ronaldo holds the records for most appearances (183), goals (140) and assists (42) in the Champions League, goals in the European Championship (14), international goals (128) and international appearances (205). He is one of the few players to have made over 1,200 professional career appearances, the most by an outfield player, and has scored over 850 official senior career goals for club and country, making him the top goalscorer of all time.
37
+ """
38
+
39
+ labels = ["person", "award", "date", "competitions", "teams"]
40
+
41
+ entities = model.predict_entities(text, labels)
42
+
43
+ for entity in entities:
44
+ print(entity["text"], "=>", entity["label"])
45
+ ```
46
+
47
+ ```
48
+ Cristiano Ronaldo dos Santos Aveiro => person
49
+ 5 February 1985 => date
50
+ Al Nassr => teams
51
+ Portugal national team => teams
52
+ Ballon d'Or => award
53
+ UEFA Men's Player of the Year Awards => award
54
+ European Golden Shoes => award
55
+ UEFA Champions Leagues => competitions
56
+ UEFA European Championship => competitions
57
+ UEFA Nations League => competitions
58
+ Champions League => competitions
59
+ European Championship => competitions
60
+ ```
61
+
62
+ ## Named Entity Recognition benchmark result
63
+ Below is a comparison of results between previous versions of the model and the current one:
64
+ ![Models performance](models_comparison.png)
65
+
66
+ ### Results on other datasets
67
+ | Model | Dataset | Precision | Recall | F1 Score |
68
+ |------------------------------------|---------------------|-----------|--------|----------|
69
+ | gliner-community/gliner_small-v2.5 | ACE 2004 | 35.18% | 22.81% | 27.67% |
70
+ | | ACE 2005 | 35.89% | 22.39% | 27.58% |
71
+ | | AnatEM | 49.12% | 31.31% | 38.24% |
72
+ | | Broad Tweet Corpus | 59.51% | 77.85% | 67.46% |
73
+ | | CoNLL 2003 | 63.16% | 70.43% | 66.60% |
74
+ | | FabNER | 23.78% | 22.55% | 23.15% |
75
+ | | FindVehicle | 37.46% | 40.06% | 38.72% |
76
+ | | GENIA_NER | 45.90% | 54.11% | 49.67% |
77
+ | | HarveyNER | 13.20% | 32.58% | 18.78% |
78
+ | | MultiNERD | 45.87% | 87.01% | 60.07% |
79
+ | | Ontonotes | 23.05% | 41.16% | 29.55% |
80
+ | | PolyglotNER | 31.88% | 67.22% | 43.25% |
81
+ | | TweetNER7 | 40.98% | 39.91% | 40.44% |
82
+ | | WikiANN en | 55.35% | 60.06% | 57.61% |
83
+ | | WikiNeural | 64.52% | 86.24% | 73.81% |
84
+ | | bc2gm | 51.70% | 49.99% | 50.83% |
85
+ | | bc4chemd | 30.78% | 57.56% | 40.11% |
86
+ | | bc5cdr | 63.48% | 69.65% | 66.42% |
87
+ | | ncbi | 63.36% | 66.67% | 64.97% |
88
+ | | **Average** | | | **46.58%** |
89
+ |------------------------------------|---------------------|-----------|--------|----------|
90
+ | urchade/gliner_small-v2.1 | ACE 2004 | 38.89% | 23.53% | 29.32% |
91
+ | | ACE 2005 | 42.09% | 26.82% | 32.76% |
92
+ | | AnatEM | 63.71% | 19.45% | 29.80% |
93
+ | | Broad Tweet Corpus | 57.01% | 70.49% | 63.04% |
94
+ | | CoNLL 2003 | 57.11% | 62.66% | 59.76% |
95
+ | | FabNER | 32.41% | 12.33% | 17.87% |
96
+ | | FindVehicle | 43.47% | 33.02% | 37.53% |
97
+ | | GENIA_NER | 61.03% | 37.25% | 46.26% |
98
+ | | HarveyNER | 23.12% | 15.16% | 18.32% |
99
+ | | MultiNERD | 43.63% | 83.60% | 57.34% |
100
+ | | Ontonotes | 23.25% | 35.41% | 28.07% |
101
+ | | PolyglotNER | 29.47% | 64.41% | 40.44% |
102
+ | | TweetNER7 | 44.78% | 30.83% | 36.52% |
103
+ | | WikiANN en | 52.58% | 58.31% | 55.30% |
104
+ | | WikiNeural | 53.38% | 82.19% | 64.72% |
105
+ | | bc2gm | 66.64% | 30.56% | 41.90% |
106
+ | | bc4chemd | 42.01% | 56.03% | 48.02% |
107
+ | | bc5cdr | 72.03% | 58.58% | 64.61% |
108
+ | | ncbi | 68.88% | 46.71% | 55.67% |
109
+ | | **Average** | | | **43.54%** |
110
+ |------------------------------------|---------------------|-----------|--------|----------|
111
+ | EmergentMethods/gliner_small-v2.1 | ACE 2004 | 39.92% | 17.50% | 24.34% |
112
+ | | ACE 2005 | 38.53% | 16.58% | 23.18% |
113
+ | | AnatEM | 55.95% | 25.69% | 35.22% |
114
+ | | Broad Tweet Corpus | 66.63% | 72.00% | 69.21% |
115
+ | | CoNLL 2003 | 62.89% | 58.96% | 60.86% |
116
+ | | FabNER | 32.76% | 13.33% | 18.95% |
117
+ | | FindVehicle | 42.93% | 43.20% | 43.06% |
118
+ | | GENIA_NER | 51.28% | 43.75% | 47.22% |
119
+ | | HarveyNER | 24.82% | 21.52% | 23.05% |
120
+ | | MultiNERD | 59.27% | 80.69% | 68.34% |
121
+ | | Ontonotes | 32.97% | 37.59% | 35.13% |
122
+ | | PolyglotNER | 33.60% | 63.30% | 43.90% |
123
+ | | TweetNER7 | 46.90% | 28.66% | 35.58% |
124
+ | | WikiANN en | 51.91% | 55.43% | 53.61% |
125
+ | | WikiNeural | 70.65% | 82.21% | 75.99% |
126
+ | | bc2gm | 49.95% | 43.13% | 46.29% |
127
+ | | bc4chemd | 35.88% | 71.64% | 47.81% |
128
+ | | bc5cdr | 68.41% | 68.90% | 68.65% |
129
+ | | ncbi | 55.31% | 59.87% | 57.50% |
130
+ | | **Average** | | | **46.20%** |
131
+ |-----------------------------------------|---------------------|-----------|--------|----------|
132
+ | gliner-community/gliner_medium-v2.5 | ACE 2004 | 33.06% | 20.96% | 25.66% |
133
+ | | ACE 2005 | 33.65% | 19.65% | 24.81% |
134
+ | | AnatEM | 52.03% | 35.28% | 42.05% |
135
+ | | Broad Tweet Corpus | 60.57% | 79.09% | 68.60% |
136
+ | | CoNLL 2003 | 63.80% | 68.31% | 65.98% |
137
+ | | FabNER | 26.20% | 22.26% | 24.07% |
138
+ | | FindVehicle | 41.95% | 40.68% | 41.30% |
139
+ | | GENIA_NER | 51.83% | 62.34% | 56.60% |
140
+ | | HarveyNER | 14.04% | 32.17% | 19.55% |
141
+ | | MultiNERD | 47.63% | 88.78% | 62.00% |
142
+ | | Ontonotes | 21.68% | 38.41% | 27.71% |
143
+ | | PolyglotNER | 32.73% | 68.27% | 44.24% |
144
+ | | TweetNER7 | 40.39% | 37.64% | 38.97% |
145
+ | | WikiANN en | 56.41% | 59.90% | 58.10% |
146
+ | | WikiNeural | 65.61% | 86.28% | 74.54% |
147
+ | | bc2gm | 55.20% | 56.71% | 55.95% |
148
+ | | bc4chemd | 35.94% | 63.67% | 45.94% |
149
+ | | bc5cdr | 63.50% | 70.09% | 66.63% |
150
+ | | ncbi | 62.96% | 68.55% | 65.63% |
151
+ | | **Average** | | | **47.81%** |
152
+ |-----------------------------------------|---------------------|-----------|--------|----------|
153
+ | urchade/gliner_medium-v2.1 | ACE 2004 | 36.33% | 22.74% | 27.97% |
154
+ | | ACE 2005 | 40.49% | 25.46% | 31.27% |
155
+ | | AnatEM | 59.75% | 16.87% | 26.31% |
156
+ | | Broad Tweet Corpus | 60.89% | 67.25% | 63.91% |
157
+ | | CoNLL 2003 | 60.62% | 62.39% | 61.50% |
158
+ | | FabNER | 27.72% | 12.24% | 16.98% |
159
+ | | FindVehicle | 41.55% | 31.31% | 35.71% |
160
+ | | GENIA_NER | 60.86% | 43.93% | 51.03% |
161
+ | | HarveyNER | 23.20% | 23.16% | 23.18% |
162
+ | | MultiNERD | 41.25% | 83.74% | 55.27% |
163
+ | | Ontonotes | 20.58% | 34.11% | 25.67% |
164
+ | | PolyglotNER | 31.32% | 64.22% | 42.11% |
165
+ | | TweetNER7 | 44.52% | 33.42% | 38.18% |
166
+ | | WikiANN en | 54.57% | 56.47% | 55.51% |
167
+ | | WikiNeural | 57.60% | 81.57% | 67.52% |
168
+ | | bc2gm | 67.98% | 33.45% | 44.84% |
169
+ | | bc4chemd | 45.66% | 52.00% | 48.62% |
170
+ | | bc5cdr | 72.20% | 58.12% | 64.40% |
171
+ | | ncbi | 73.12% | 49.74% | 59.20% |
172
+ | | **Average** | | | **44.17%** |
173
+ |-----------------------------------------|---------------------|-----------|--------|----------|
174
+ | EmergentMethods/gliner_news_medium-v2.1 | ACE 2004 | 39.21% | 17.24% | 23.95% |
175
+ | | ACE 2005 | 39.82% | 16.48% | 23.31% |
176
+ | | AnatEM | 57.67% | 23.57% | 33.46% |
177
+ | | Broad Tweet Corpus | 69.52% | 65.94% | 67.69% |
178
+ | | CoNLL 2003 | 68.26% | 58.45% | 62.97% |
179
+ | | FabNER | 30.74% | 15.51% | 20.62% |
180
+ | | FindVehicle | 40.33% | 37.37% | 38.79% |
181
+ | | GENIA_NER | 53.70% | 47.73% | 50.54% |
182
+ | | HarveyNER | 26.29% | 27.05% | 26.67% |
183
+ | | MultiNERD | 56.78% | 81.96% | 67.08% |
184
+ | | Ontonotes | 30.90% | 35.86% | 33.19% |
185
+ | | PolyglotNER | 35.98% | 60.96% | 45.25% |
186
+ | | TweetNER7 | 52.37% | 30.50% | 38.55% |
187
+ | | WikiANN en | 53.81% | 52.29% | 53.04% |
188
+ | | WikiNeural | 76.84% | 78.92% | 77.86% |
189
+ | | bc2gm | 62.97% | 44.24% | 51.96% |
190
+ | | bc4chemd | 44.90% | 65.56% | 53.30% |
191
+ | | bc5cdr | 73.93% | 67.03% | 70.31% |
192
+ | | ncbi | 69.53% | 60.82% | 64.88% |
193
+ | | **Average** | | | **47.55%** |
194
+ |-----------------------------------------|---------------------|-----------|--------|----------|
195
+ | gliner-community/gliner_large-v2.5 | ACE 2004 | 31.64% | 22.81% | 26.51% |
196
+ | | ACE 2005 | 32.10% | 22.56% | 26.49% |
197
+ | | AnatEM | 53.64% | 27.82% | 36.64% |
198
+ | | Broad Tweet Corpus | 61.93% | 76.85% | 68.59% |
199
+ | | CoNLL 2003 | 62.83% | 67.71% | 65.18% |
200
+ | | FabNER | 24.54% | 27.03% | 25.73% |
201
+ | | FindVehicle | 40.71% | 56.24% | 47.23% |
202
+ | | GENIA_NER | 43.56% | 52.56% | 47.64% |
203
+ | | HarveyNER | 14.85% | 27.05% | 19.17% |
204
+ | | MultiNERD | 38.04% | 89.17% | 53.33% |
205
+ | | Ontonotes | 17.28% | 40.16% | 24.16% |
206
+ | | PolyglotNER | 32.88% | 63.31% | 43.28% |
207
+ | | TweetNER7 | 38.03% | 41.43% | 39.66% |
208
+ | | WikiANN en | 57.80% | 60.54% | 59.14% |
209
+ | | WikiNeural | 67.72% | 83.94% | 74.96% |
210
+ | | bc2gm | 54.74% | 48.54% | 51.45% |
211
+ | | bc4chemd | 40.20% | 58.66% | 47.71% |
212
+ | | bc5cdr | 66.27% | 71.95% | 69.00% |
213
+ | | ncbi | 68.09% | 61.55% | 64.65% |
214
+ | | **Average** | | | **46.87%** |
215
+ |-----------------------------------------|---------------------|-----------|--------|----------|
216
+ | urchade/gliner_large-v2.1 | ACE 2004 | 37.52% | 25.38% | 30.28% |
217
+ | | ACE 2005 | 39.02% | 29.00% | 33.27% |
218
+ | | AnatEM | 52.86% | 13.64% | 21.68% |
219
+ | | Broad Tweet Corpus | 51.44% | 71.73% | 59.91% |
220
+ | | CoNLL 2003 | 54.86% | 64.98% | 59.49% |
221
+ | | FabNER | 23.98% | 16.00% | 19.19% |
222
+ | | FindVehicle | 47.04% | 57.53% | 51.76% |
223
+ | | GENIA_NER | 58.10% | 49.98% | 53.74% |
224
+ | | HarveyNER | 16.29% | 21.93% | 18.69% |
225
+ | | MultiNERD | 34.09% | 85.43% | 48.74% |
226
+ | | Ontonotes | 14.02% | 32.01% | 19.50% |
227
+ | | PolyglotNER | 28.53% | 64.92% | 39.64% |
228
+ | | TweetNER7 | 38.00% | 34.34% | 36.08% |
229
+ | | WikiANN en | 51.69% | 59.92% | 55.50% |
230
+ | | WikiNeural | 50.94% | 82.08% | 62.87% |
231
+ | | bc2gm | 64.48% | 32.47% | 43.19% |
232
+ | | bc4chemd | 48.66% | 57.52% | 52.72% |
233
+ | | bc5cdr | 72.19% | 64.27% | 68.00% |
234
+ | | ncbi | 69.54% | 52.25% | 59.67% |
235
+ | | **Average** | | | **43.89%** |
236
+ |-----------------------------------------|---------------------|-----------|--------|----------|
237
+ | EmergenMethods/fliner_news_large-v2.1 | ACE 2004 | 43.19% | 18.39% | 25.80% |
238
+ | | ACE 2005 | 45.24% | 21.20% | 28.87% |
239
+ | | AnatEM | 61.51% | 21.66% | 32.04% |
240
+ | | Broad Tweet Corpus | 69.38% | 68.99% | 69.18% |
241
+ | | CoNLL 2003 | 61.47% | 52.18% | 56.45% |
242
+ | | FabNER | 27.42% | 19.11% | 22.52% |
243
+ | | FindVehicle | 46.30% | 62.48% | 53.19% |
244
+ | | GENIA_NER | 54.13% | 54.02% | 54.07% |
245
+ | | HarveyNER | 15.91% | 15.78% | 15.84% |
246
+ | | MultiNERD | 53.73% | 79.07% | 63.98% |
247
+ | | Ontonotes | 26.78% | 39.77% | 32.01% |
248
+ | | PolyglotNER | 34.28% | 55.87% | 42.49% |
249
+ | | TweetNER7 | 48.06% | 28.18% | 35.53% |
250
+ | | WikiANN en | 53.66% | 51.34% | 52.47% |
251
+ | | WikiNeural | 69.81% | 70.75% | 70.28% |
252
+ | | bc2gm | 59.83% | 37.62% | 46.20% |
253
+ | | bc4chemd | 46.24% | 69.15% | 55.42% |
254
+ | | bc5cdr | 71.94% | 70.37% | 71.15% |
255
+ | | ncbi | 70.17% | 61.44% | 65.52% |
256
+ | | **Average** | | | **47.00%** |
257
+ |-----------------------------------------|---------------------|-----------|--------|----------|
258
+
259
+ ## Other available models
260
+
261
+ | Release | Model Name | # of Parameters | Language | License |
262
+ | - | - | - | - | - |
263
+ | v0 | [urchade/gliner_base](https://huggingface.co/urchade/gliner_base)<br>[urchade/gliner_multi](https://huggingface.co/urchade/gliner_multi) | 209M<br>209M | English<br>Multilingual | cc-by-nc-4.0 |
264
+ | v1 | [urchade/gliner_small-v1](https://huggingface.co/urchade/gliner_small-v1)<br>[urchade/gliner_medium-v1](https://huggingface.co/urchade/gliner_medium-v1)<br>[urchade/gliner_large-v1](https://huggingface.co/urchade/gliner_large-v1) | 166M<br>209M<br>459M | English <br> English <br> English | cc-by-nc-4.0 |
265
+ | v2 | [urchade/gliner_small-v2](https://huggingface.co/urchade/gliner_small-v2)<br>[urchade/gliner_medium-v2](https://huggingface.co/urchade/gliner_medium-v2)<br>[urchade/gliner_large-v2](https://huggingface.co/urchade/gliner_large-v2) | 166M<br>209M<br>459M | English <br> English <br> English | apache-2.0 |
266
+ | v2.1 | [urchade/gliner_small-v2.1](https://huggingface.co/urchade/gliner_small-v2.1)<br>[urchade/gliner_medium-v2.1](https://huggingface.co/urchade/gliner_medium-v2.1)<br>[urchade/gliner_large-v2.1](https://huggingface.co/urchade/gliner_large-v2.1) <br>[urchade/gliner_multi-v2.1](https://huggingface.co/urchade/gliner_multi-v2.1) | 166M<br>209M<br>459M<br>209M | English <br> English <br> English <br> Multilingual | apache-2.0 |
267
+
268
+
269
+ ## Model Authors
270
+ The model authors are:
271
+ * [Urchade Zaratiana](https://huggingface.co/urchade)
272
+ * [Ihor Stepanov](https://huggingface.co/Ihor)
273
+ * Nadi Tomeh
274
+ * Pierre Holat
275
+ * Thierry Charnois
276
+
277
+ ## Citation
278
+ ```bibtex
279
+ @misc{zaratiana2023gliner,
280
+ title={GLiNER: Generalist Model for Named Entity Recognition using Bidirectional Transformer},
281
+ author={Urchade Zaratiana and Nadi Tomeh and Pierre Holat and Thierry Charnois},
282
+ year={2023},
283
+ eprint={2311.08526},
284
+ archivePrefix={arXiv},
285
+ primaryClass={cs.CL}
286
+ }
287
+ ```