Hugging Face's logo

seanfarrell
/
icd11_ner_model_small

Sentence Similarity
sentence-transformers
Safetensors
bert
feature-extraction
dense
Generated from Trainer
dataset_size:4096000
loss:MultipleNegativesRankingLoss
Eval Results (legacy)
text-embeddings-inference
Model card Files
xet
 Community
seanfarrell commited on
Commit
7083ed4
·
verified ·
1 Parent(s): ad40d7e

Add new SentenceTransformer model

Browse files
Files changed (9) hide show
  1. 1_Pooling/config.json +5 -0
  2. README.md +632 -0
  3. config.json +35 -0
  4. config_sentence_transformers.json +14 -0
  5. model.safetensors +3 -0
  6. modules.json +20 -0
  7. sentence_bert_config.json +10 -0
  8. tokenizer.json +0 -0
  9. tokenizer_config.json +25 -0
1_Pooling/config.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "embedding_dimension": 384,
3
+ "pooling_mode": "cls",
4
+ "include_prompt": true
5
+ }
README.md ADDED
@@ -0,0 +1,632 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - sentence-transformers
4
+ - sentence-similarity
5
+ - feature-extraction
6
+ - dense
7
+ - generated_from_trainer
8
+ - dataset_size:4096000
9
+ - loss:MultipleNegativesRankingLoss
10
+ base_model: BAAI/bge-small-en-v1.5
11
+ widget:
12
+ - source_sentence: vascular disorders of the stomach
13
+ sentences:
14
+ - vascular disorders of the stomach, unspecified
15
+ - ringworm of the hand
16
+ - anogenital herpes
17
+ - source_sentence: neonatal malabsorption secondary to defect in carbohydrate metabolism
18
+ sentences:
19
+ - overfeeding problem of newborn
20
+ - other specified neonatal malabsorption syndromes
21
+ - bronchial asthma
22
+ - source_sentence: white matter disorder due to carbon monoxide toxicity
23
+ sentences:
24
+ - white matter disorder due to nitrous oxide toxicity
25
+ - neonatal dehydration
26
+ - moerman lethal chondrodysplasia
27
+ - source_sentence: chronic myelogenous leukaemia, bcr-abl1-positive in complete remission
28
+ sentences:
29
+ - chronic myeloid leukaemia, bcr-abl-positive in complete remission
30
+ - lesion of sciatic nerve
31
+ - tooth injury
32
+ - source_sentence: metabolic or transporter liver disease, unspecified
33
+ sentences:
34
+ - metabolic or transporter liver disease
35
+ - staphyloma of cornea
36
+ - spontaneous rupture of synovium or tendon, unspecified
37
+ pipeline_tag: sentence-similarity
38
+ library_name: sentence-transformers
39
+ metrics:
40
+ - cosine_accuracy@1
41
+ - cosine_accuracy@5
42
+ - cosine_accuracy@10
43
+ - cosine_precision@1
44
+ - cosine_precision@3
45
+ - cosine_precision@5
46
+ - cosine_precision@10
47
+ - cosine_recall@1
48
+ - cosine_recall@3
49
+ - cosine_recall@5
50
+ - cosine_recall@10
51
+ - cosine_ndcg@10
52
+ - cosine_mrr@10
53
+ - cosine_map@10
54
+ model-index:
55
+ - name: SentenceTransformer based on BAAI/bge-small-en-v1.5
56
+ results:
57
+ - task:
58
+ type: information-retrieval
59
+ name: Information Retrieval
60
+ dataset:
61
+ name: icd11 val
62
+ type: icd11-val
63
+ metrics:
64
+ - type: cosine_accuracy@1
65
+ value: 0.7478448275862069
66
+ name: Cosine Accuracy@1
67
+ - type: cosine_accuracy@5
68
+ value: 0.8717672413793104
69
+ name: Cosine Accuracy@5
70
+ - type: cosine_accuracy@10
71
+ value: 0.896551724137931
72
+ name: Cosine Accuracy@10
73
+ - type: cosine_precision@1
74
+ value: 0.7478448275862069
75
+ name: Cosine Precision@1
76
+ - type: cosine_precision@3
77
+ value: 0.6253591954022988
78
+ name: Cosine Precision@3
79
+ - type: cosine_precision@5
80
+ value: 0.5258620689655173
81
+ name: Cosine Precision@5
82
+ - type: cosine_precision@10
83
+ value: 0.3786637931034483
84
+ name: Cosine Precision@10
85
+ - type: cosine_recall@1
86
+ value: 0.1588556265383415
87
+ name: Cosine Recall@1
88
+ - type: cosine_recall@3
89
+ value: 0.37728305808648516
90
+ name: Cosine Recall@3
91
+ - type: cosine_recall@5
92
+ value: 0.4796248471294223
93
+ name: Cosine Recall@5
94
+ - type: cosine_recall@10
95
+ value: 0.6031417673743621
96
+ name: Cosine Recall@10
97
+ - type: cosine_ndcg@10
98
+ value: 0.6649315033350626
99
+ name: Cosine Ndcg@10
100
+ - type: cosine_mrr@10
101
+ value: 0.8006516830870286
102
+ name: Cosine Mrr@10
103
+ - type: cosine_map@10
104
+ value: 0.5745241229374094
105
+ name: Cosine Map@10
106
+ ---
107
+
108
+ # SentenceTransformer based on BAAI/bge-small-en-v1.5
109
+
110
+ This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [BAAI/bge-small-en-v1.5](https://huggingface.co/BAAI/bge-small-en-v1.5). It maps sentences & paragraphs to a 384-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, classification, clustering, and more.
111
+
112
+ ## Model Details
113
+
114
+ ### Model Description
115
+ - **Model Type:** Sentence Transformer
116
+ - **Base model:** [BAAI/bge-small-en-v1.5](https://huggingface.co/BAAI/bge-small-en-v1.5) <!-- at revision 5c38ec7c405ec4b44b94cc5a9bb96e735b38267a -->
117
+ - **Maximum Sequence Length:** 30 tokens
118
+ - **Output Dimensionality:** 384 dimensions
119
+ - **Similarity Function:** Cosine Similarity
120
+ - **Supported Modality:** Text
121
+ <!-- - **Training Dataset:** Unknown -->
122
+ <!-- - **Language:** Unknown -->
123
+ <!-- - **License:** Unknown -->
124
+
125
+ ### Model Sources
126
+
127
+ - **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
128
+ - **Repository:** [Sentence Transformers on GitHub](https://github.com/huggingface/sentence-transformers)
129
+ - **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)
130
+
131
+ ### Full Model Architecture
132
+
133
+ ```
134
+ SentenceTransformer(
135
+ (0): Transformer({'transformer_task': 'feature-extraction', 'modality_config': {'text': {'method': 'forward', 'method_output_name': 'last_hidden_state'}}, 'module_output_name': 'token_embeddings', 'architecture': 'BertModel'})
136
+ (1): Pooling({'embedding_dimension': 384, 'pooling_mode': 'cls', 'include_prompt': True})
137
+ (2): Normalize({})
138
+ )
139
+ ```
140
+
141
+ ## Usage
142
+
143
+ ### Direct Usage (Sentence Transformers)
144
+
145
+ First install the Sentence Transformers library:
146
+
147
+ ```bash
148
+ pip install -U sentence-transformers
149
+ ```
150
+ Then you can load this model and run inference.
151
+ ```python
152
+ from sentence_transformers import SentenceTransformer
153
+
154
+ # Download from the 🤗 Hub
155
+ model = SentenceTransformer("seanfarrell/icd11_ner_model_small")
156
+ # Run inference
157
+ queries = [
158
+ 'metabolic or transporter liver disease, unspecified',
159
+ ]
160
+ documents = [
161
+ 'metabolic or transporter liver disease',
162
+ 'spontaneous rupture of synovium or tendon, unspecified',
163
+ 'staphyloma of cornea',
164
+ ]
165
+ query_embeddings = model.encode_query(queries)
166
+ document_embeddings = model.encode_document(documents)
167
+ print(query_embeddings.shape, document_embeddings.shape)
168
+ # [1, 384] [3, 384]
169
+
170
+ # Get the similarity scores for the embeddings
171
+ similarities = model.similarity(query_embeddings, document_embeddings)
172
+ print(similarities)
173
+ # tensor([[ 0.9840, -0.1296, -0.1292]])
174
+ ```
175
+ <!--
176
+ ### Direct Usage (Transformers)
177
+
178
+ <details><summary>Click to see the direct usage in Transformers</summary>
179
+
180
+ </details>
181
+ -->
182
+
183
+ <!--
184
+ ### Downstream Usage (Sentence Transformers)
185
+
186
+ You can finetune this model on your own dataset.
187
+
188
+ <details><summary>Click to expand</summary>
189
+
190
+ </details>
191
+ -->
192
+
193
+ <!--
194
+ ### Out-of-Scope Use
195
+
196
+ *List how the model may foreseeably be misused and address what users ought not to do with the model.*
197
+ -->
198
+
199
+ ## Evaluation
200
+
201
+ ### Metrics
202
+
203
+ #### Information Retrieval
204
+
205
+ * Dataset: `icd11-val`
206
+ * Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.sentence_transformer.evaluation.InformationRetrievalEvaluator)
207
+
208
+ | Metric | Value |
209
+ |:--------------------|:-----------|
210
+ | cosine_accuracy@1 | 0.7478 |
211
+ | cosine_accuracy@5 | 0.8718 |
212
+ | cosine_accuracy@10 | 0.8966 |
213
+ | cosine_precision@1 | 0.7478 |
214
+ | cosine_precision@3 | 0.6254 |
215
+ | cosine_precision@5 | 0.5259 |
216
+ | cosine_precision@10 | 0.3787 |
217
+ | cosine_recall@1 | 0.1589 |
218
+ | cosine_recall@3 | 0.3773 |
219
+ | cosine_recall@5 | 0.4796 |
220
+ | cosine_recall@10 | 0.6031 |
221
+ | **cosine_ndcg@10** | **0.6649** |
222
+ | cosine_mrr@10 | 0.8007 |
223
+ | cosine_map@10 | 0.5745 |
224
+
225
+ <!--
226
+ ## Bias, Risks and Limitations
227
+
228
+ *What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
229
+ -->
230
+
231
+ <!--
232
+ ### Recommendations
233
+
234
+ *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
235
+ -->
236
+
237
+ ## Training Details
238
+
239
+ ### Training Dataset
240
+
241
+ #### Unnamed Dataset
242
+
243
+ * Size: 4,096,000 training samples
244
+ * Columns: <code>anchor</code> and <code>positive</code>
245
+ * Approximate statistics based on the first 100 samples:
246
+ | | anchor | positive |
247
+ |:---------|:----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|
248
+ | type | string | string |
249
+ | modality | text | text |
250
+ | details | <ul><li>min: 4 tokens</li><li>mean: 11.07 tokens</li><li>max: 30 tokens</li></ul> | <ul><li>min: 4 tokens</li><li>mean: 11.07 tokens</li><li>max: 28 tokens</li></ul> |
251
+ * Samples:
252
+ | anchor | positive |
253
+ |:-------------------------------------------------------------|:-----------------------------------------------------------------|
254
+ | <code>duodenal adenoma in familial polyposis</code> | <code>benign duodenal neoplasm</code> |
255
+ | <code>congenital unilateral hip dislocation</code> | <code>developmental dysplasia of hip, unilateral</code> |
256
+ | <code>postprocedural pulmonary arterial tree disorder</code> | <code>postprocedural pulmonary arterial tree complication</code> |
257
+ * Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters:
258
+ ```json
259
+ {
260
+ "scale": 20.0,
261
+ "similarity_fct": "cos_sim",
262
+ "gather_across_devices": false,
263
+ "directions": [
264
+ "query_to_doc"
265
+ ],
266
+ "partition_mode": "joint",
267
+ "hardness_mode": null,
268
+ "hardness_strength": 0.0
269
+ }
270
+ ```
271
+
272
+ ### Training Hyperparameters
273
+ #### Non-Default Hyperparameters
274
+
275
+ - `per_device_train_batch_size`: 512
276
+ - `num_train_epochs`: 1
277
+ - `learning_rate`: 2e-05
278
+ - `lr_scheduler_type`: cosine
279
+ - `warmup_steps`: 800
280
+ - `weight_decay`: 0.01
281
+ - `bf16`: True
282
+ - `per_device_eval_batch_size`: 256
283
+ - `eval_on_start`: True
284
+ - `load_best_model_at_end`: True
285
+ - `dataloader_drop_last`: True
286
+
287
+ #### All Hyperparameters
288
+ <details><summary>Click to expand</summary>
289
+
290
+ - `per_device_train_batch_size`: 512
291
+ - `num_train_epochs`: 1
292
+ - `max_steps`: -1
293
+ - `learning_rate`: 2e-05
294
+ - `lr_scheduler_type`: cosine
295
+ - `lr_scheduler_kwargs`: None
296
+ - `warmup_steps`: 800
297
+ - `optim`: adamw_torch_fused
298
+ - `optim_args`: None
299
+ - `weight_decay`: 0.01
300
+ - `adam_beta1`: 0.9
301
+ - `adam_beta2`: 0.999
302
+ - `adam_epsilon`: 1e-08
303
+ - `optim_target_modules`: None
304
+ - `gradient_accumulation_steps`: 1
305
+ - `average_tokens_across_devices`: True
306
+ - `max_grad_norm`: 1.0
307
+ - `label_smoothing_factor`: 0.0
308
+ - `bf16`: True
309
+ - `fp16`: False
310
+ - `bf16_full_eval`: False
311
+ - `fp16_full_eval`: False
312
+ - `tf32`: None
313
+ - `gradient_checkpointing`: False
314
+ - `gradient_checkpointing_kwargs`: None
315
+ - `torch_compile`: False
316
+ - `torch_compile_backend`: None
317
+ - `torch_compile_mode`: None
318
+ - `use_liger_kernel`: False
319
+ - `liger_kernel_config`: None
320
+ - `use_cache`: False
321
+ - `neftune_noise_alpha`: None
322
+ - `torch_empty_cache_steps`: None
323
+ - `auto_find_batch_size`: False
324
+ - `log_on_each_node`: True
325
+ - `logging_nan_inf_filter`: True
326
+ - `include_num_input_tokens_seen`: no
327
+ - `log_level`: passive
328
+ - `log_level_replica`: warning
329
+ - `disable_tqdm`: False
330
+ - `project`: huggingface
331
+ - `trackio_space_id`: None
332
+ - `trackio_bucket_id`: None
333
+ - `trackio_static_space_id`: None
334
+ - `per_device_eval_batch_size`: 256
335
+ - `prediction_loss_only`: True
336
+ - `eval_on_start`: True
337
+ - `eval_do_concat_batches`: True
338
+ - `eval_use_gather_object`: False
339
+ - `eval_accumulation_steps`: None
340
+ - `include_for_metrics`: []
341
+ - `batch_eval_metrics`: False
342
+ - `save_only_model`: False
343
+ - `save_on_each_node`: False
344
+ - `enable_jit_checkpoint`: False
345
+ - `push_to_hub`: False
346
+ - `hub_private_repo`: None
347
+ - `hub_model_id`: None
348
+ - `hub_strategy`: every_save
349
+ - `hub_always_push`: False
350
+ - `hub_revision`: None
351
+ - `load_best_model_at_end`: True
352
+ - `ignore_data_skip`: False
353
+ - `restore_callback_states_from_checkpoint`: False
354
+ - `full_determinism`: False
355
+ - `seed`: 42
356
+ - `data_seed`: None
357
+ - `use_cpu`: False
358
+ - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
359
+ - `parallelism_config`: None
360
+ - `dataloader_drop_last`: True
361
+ - `dataloader_num_workers`: 0
362
+ - `dataloader_pin_memory`: True
363
+ - `dataloader_persistent_workers`: False
364
+ - `dataloader_prefetch_factor`: None
365
+ - `dataloader_multiprocessing_context`: None
366
+ - `dataloader_in_order`: True
367
+ - `remove_unused_columns`: True
368
+ - `label_names`: None
369
+ - `train_sampling_strategy`: random
370
+ - `length_column_name`: length
371
+ - `ddp_find_unused_parameters`: None
372
+ - `ddp_bucket_cap_mb`: None
373
+ - `ddp_broadcast_buffers`: False
374
+ - `ddp_static_graph`: None
375
+ - `ddp_backend`: None
376
+ - `ddp_timeout`: 1800
377
+ - `fsdp`: None
378
+ - `fsdp_config`: None
379
+ - `deepspeed`: None
380
+ - `debug`: []
381
+ - `skip_memory_metrics`: True
382
+ - `do_predict`: False
383
+ - `resume_from_checkpoint`: None
384
+ - `local_rank`: -1
385
+ - `prompts`: None
386
+ - `batch_sampler`: batch_sampler
387
+ - `multi_dataset_batch_sampler`: proportional
388
+ - `router_mapping`: {}
389
+ - `learning_rate_mapping`: {}
390
+ - `warmup_ratio`: None
391
+
392
+ </details>
393
+
394
+ ### Training Logs
395
+ <details><summary>Click to expand</summary>
396
+
397
+ | Epoch | Step | Training Loss | icd11-val_cosine_ndcg@10 |
398
+ |:----------:|:--------:|:-------------:|:------------------------:|
399
+ | 0 | 0 | - | 0.4871 |
400
+ | 0.0063 | 50 | 2.5057 | - |
401
+ | 0.0125 | 100 | 2.1921 | - |
402
+ | 0.0187 | 150 | 1.6707 | - |
403
+ | 0.025 | 200 | 1.4270 | - |
404
+ | 0.0312 | 250 | 1.3379 | - |
405
+ | 0.0375 | 300 | 1.2811 | - |
406
+ | 0.0437 | 350 | 1.2222 | - |
407
+ | 0.05 | 400 | 1.1848 | - |
408
+ | 0.0563 | 450 | 1.1484 | - |
409
+ | 0.0625 | 500 | 1.1260 | 0.5411 |
410
+ | 0.0688 | 550 | 1.0856 | - |
411
+ | 0.075 | 600 | 1.0654 | - |
412
+ | 0.0813 | 650 | 1.0247 | - |
413
+ | 0.0875 | 700 | 1.0050 | - |
414
+ | 0.0938 | 750 | 0.9669 | - |
415
+ | 0.1 | 800 | 0.9515 | - |
416
+ | 0.1062 | 850 | 0.9404 | - |
417
+ | 0.1125 | 900 | 0.8906 | - |
418
+ | 0.1187 | 950 | 0.8692 | - |
419
+ | 0.125 | 1000 | 0.8413 | 0.5746 |
420
+ | 0.1313 | 1050 | 0.8439 | - |
421
+ | 0.1375 | 1100 | 0.8270 | - |
422
+ | 0.1437 | 1150 | 0.8128 | - |
423
+ | 0.15 | 1200 | 0.7762 | - |
424
+ | 0.1562 | 1250 | 0.7666 | - |
425
+ | 0.1625 | 1300 | 0.7555 | - |
426
+ | 0.1688 | 1350 | 0.7562 | - |
427
+ | 0.175 | 1400 | 0.7180 | - |
428
+ | 0.1812 | 1450 | 0.6945 | - |
429
+ | 0.1875 | 1500 | 0.7032 | 0.6002 |
430
+ | 0.1938 | 1550 | 0.6747 | - |
431
+ | 0.2 | 1600 | 0.6729 | - |
432
+ | 0.2062 | 1650 | 0.6604 | - |
433
+ | 0.2125 | 1700 | 0.6644 | - |
434
+ | 0.2188 | 1750 | 0.6556 | - |
435
+ | 0.225 | 1800 | 0.6440 | - |
436
+ | 0.2313 | 1850 | 0.6284 | - |
437
+ | 0.2375 | 1900 | 0.6105 | - |
438
+ | 0.2437 | 1950 | 0.5935 | - |
439
+ | 0.25 | 2000 | 0.5923 | 0.6120 |
440
+ | 0.2562 | 2050 | 0.5850 | - |
441
+ | 0.2625 | 2100 | 0.5787 | - |
442
+ | 0.2687 | 2150 | 0.5764 | - |
443
+ | 0.275 | 2200 | 0.5643 | - |
444
+ | 0.2812 | 2250 | 0.5655 | - |
445
+ | 0.2875 | 2300 | 0.5463 | - |
446
+ | 0.2938 | 2350 | 0.5414 | - |
447
+ | 0.3 | 2400 | 0.5368 | - |
448
+ | 0.3063 | 2450 | 0.5236 | - |
449
+ | 0.3125 | 2500 | 0.5109 | 0.6230 |
450
+ | 0.3187 | 2550 | 0.5308 | - |
451
+ | 0.325 | 2600 | 0.5000 | - |
452
+ | 0.3312 | 2650 | 0.5069 | - |
453
+ | 0.3375 | 2700 | 0.5120 | - |
454
+ | 0.3438 | 2750 | 0.4853 | - |
455
+ | 0.35 | 2800 | 0.4877 | - |
456
+ | 0.3563 | 2850 | 0.4786 | - |
457
+ | 0.3625 | 2900 | 0.4869 | - |
458
+ | 0.3688 | 2950 | 0.4865 | - |
459
+ | 0.375 | 3000 | 0.4561 | 0.6332 |
460
+ | 0.3812 | 3050 | 0.4640 | - |
461
+ | 0.3875 | 3100 | 0.4629 | - |
462
+ | 0.3937 | 3150 | 0.4575 | - |
463
+ | 0.4 | 3200 | 0.4493 | - |
464
+ | 0.4062 | 3250 | 0.4503 | - |
465
+ | 0.4125 | 3300 | 0.4301 | - |
466
+ | 0.4188 | 3350 | 0.4410 | - |
467
+ | 0.425 | 3400 | 0.4415 | - |
468
+ | 0.4313 | 3450 | 0.4316 | - |
469
+ | 0.4375 | 3500 | 0.4359 | 0.6441 |
470
+ | 0.4437 | 3550 | 0.4147 | - |
471
+ | 0.45 | 3600 | 0.4028 | - |
472
+ | 0.4562 | 3650 | 0.4002 | - |
473
+ | 0.4625 | 3700 | 0.4132 | - |
474
+ | 0.4688 | 3750 | 0.4021 | - |
475
+ | 0.475 | 3800 | 0.4058 | - |
476
+ | 0.4813 | 3850 | 0.4023 | - |
477
+ | 0.4875 | 3900 | 0.4038 | - |
478
+ | 0.4938 | 3950 | 0.4007 | - |
479
+ | 0.5 | 4000 | 0.3894 | 0.6501 |
480
+ | 0.5062 | 4050 | 0.3856 | - |
481
+ | 0.5125 | 4100 | 0.3879 | - |
482
+ | 0.5188 | 4150 | 0.3758 | - |
483
+ | 0.525 | 4200 | 0.3796 | - |
484
+ | 0.5312 | 4250 | 0.3890 | - |
485
+ | 0.5375 | 4300 | 0.3831 | - |
486
+ | 0.5437 | 4350 | 0.3806 | - |
487
+ | 0.55 | 4400 | 0.3781 | - |
488
+ | 0.5563 | 4450 | 0.3761 | - |
489
+ | 0.5625 | 4500 | 0.3723 | 0.6551 |
490
+ | 0.5687 | 4550 | 0.3603 | - |
491
+ | 0.575 | 4600 | 0.3595 | - |
492
+ | 0.5813 | 4650 | 0.3524 | - |
493
+ | 0.5875 | 4700 | 0.3695 | - |
494
+ | 0.5938 | 4750 | 0.3552 | - |
495
+ | 0.6 | 4800 | 0.3521 | - |
496
+ | 0.6062 | 4850 | 0.3459 | - |
497
+ | 0.6125 | 4900 | 0.3472 | - |
498
+ | 0.6188 | 4950 | 0.3511 | - |
499
+ | 0.625 | 5000 | 0.3537 | 0.6589 |
500
+ | 0.6312 | 5050 | 0.3490 | - |
501
+ | 0.6375 | 5100 | 0.3428 | - |
502
+ | 0.6438 | 5150 | 0.3492 | - |
503
+ | 0.65 | 5200 | 0.3471 | - |
504
+ | 0.6562 | 5250 | 0.3348 | - |
505
+ | 0.6625 | 5300 | 0.3404 | - |
506
+ | 0.6687 | 5350 | 0.3442 | - |
507
+ | 0.675 | 5400 | 0.3494 | - |
508
+ | 0.6813 | 5450 | 0.3306 | - |
509
+ | 0.6875 | 5500 | 0.3349 | 0.6598 |
510
+ | 0.6937 | 5550 | 0.3401 | - |
511
+ | 0.7 | 5600 | 0.3374 | - |
512
+ | 0.7063 | 5650 | 0.3306 | - |
513
+ | 0.7125 | 5700 | 0.3277 | - |
514
+ | 0.7188 | 5750 | 0.3274 | - |
515
+ | 0.725 | 5800 | 0.3318 | - |
516
+ | 0.7312 | 5850 | 0.3316 | - |
517
+ | 0.7375 | 5900 | 0.3273 | - |
518
+ | 0.7438 | 5950 | 0.3252 | - |
519
+ | 0.75 | 6000 | 0.3306 | 0.6636 |
520
+ | 0.7562 | 6050 | 0.3347 | - |
521
+ | 0.7625 | 6100 | 0.3186 | - |
522
+ | 0.7688 | 6150 | 0.3301 | - |
523
+ | 0.775 | 6200 | 0.3203 | - |
524
+ | 0.7812 | 6250 | 0.3234 | - |
525
+ | 0.7875 | 6300 | 0.3134 | - |
526
+ | 0.7937 | 6350 | 0.3175 | - |
527
+ | 0.8 | 6400 | 0.3312 | - |
528
+ | 0.8063 | 6450 | 0.3302 | - |
529
+ | 0.8125 | 6500 | 0.3189 | 0.6644 |
530
+ | 0.8187 | 6550 | 0.3191 | - |
531
+ | 0.825 | 6600 | 0.3265 | - |
532
+ | 0.8313 | 6650 | 0.3192 | - |
533
+ | 0.8375 | 6700 | 0.3111 | - |
534
+ | 0.8438 | 6750 | 0.3124 | - |
535
+ | 0.85 | 6800 | 0.3101 | - |
536
+ | 0.8562 | 6850 | 0.3188 | - |
537
+ | 0.8625 | 6900 | 0.3180 | - |
538
+ | 0.8688 | 6950 | 0.3153 | - |
539
+ | 0.875 | 7000 | 0.3159 | 0.6646 |
540
+ | 0.8812 | 7050 | 0.3132 | - |
541
+ | 0.8875 | 7100 | 0.3168 | - |
542
+ | 0.8938 | 7150 | 0.3227 | - |
543
+ | 0.9 | 7200 | 0.3121 | - |
544
+ | 0.9062 | 7250 | 0.3212 | - |
545
+ | 0.9125 | 7300 | 0.3195 | - |
546
+ | 0.9187 | 7350 | 0.3177 | - |
547
+ | 0.925 | 7400 | 0.3228 | - |
548
+ | 0.9313 | 7450 | 0.3172 | - |
549
+ | **0.9375** | **7500** | **0.3045** | **0.6652** |
550
+ | 0.9437 | 7550 | 0.3052 | - |
551
+ | 0.95 | 7600 | 0.3273 | - |
552
+ | 0.9563 | 7650 | 0.3163 | - |
553
+ | 0.9625 | 7700 | 0.3138 | - |
554
+ | 0.9688 | 7750 | 0.3280 | - |
555
+ | 0.975 | 7800 | 0.3074 | - |
556
+ | 0.9812 | 7850 | 0.3266 | - |
557
+ | 0.9875 | 7900 | 0.3214 | - |
558
+ | 0.9938 | 7950 | 0.3240 | - |
559
+ | 1.0 | 8000 | 0.3141 | 0.6649 |
560
+
561
+ * The bold row denotes the saved checkpoint.
562
+ </details>
563
+
564
+ ### Training Time
565
+ - **Training**: 8.9 minutes
566
+ - **Evaluation**: 45.4 seconds
567
+ - **Total**: 9.7 minutes
568
+
569
+ ### Framework Versions
570
+ - Python: 3.13.15
571
+ - Sentence Transformers: 5.7.0
572
+ - Transformers: 5.18.0
573
+ - PyTorch: 2.11.0+cu130
574
+ - Accelerate: 1.15.0
575
+ - Datasets: 4.8.5
576
+ - Tokenizers: 0.23.2
577
+
578
+ ## Additional Resources
579
+
580
+ - [Training and Finetuning Embedding Models with Sentence Transformers](https://huggingface.co/blog/train-sentence-transformers): the end-to-end guide for training or finetuning Sentence Transformer models.
581
+ - [Introduction to Matryoshka Embedding Models](https://huggingface.co/blog/matryoshka): variable-size embeddings that can be truncated with minimal quality loss.
582
+ - [Binary and Scalar Embedding Quantization for Significantly Faster & Cheaper Retrieval](https://huggingface.co/blog/embedding-quantization): post-training compression of embedding vectors.
583
+ - [Multimodal Embedding & Reranker Models with Sentence Transformers](https://huggingface.co/blog/multimodal-sentence-transformers): use text, image, audio, and video models through the same API.
584
+ - [Training and Finetuning Multimodal Embedding & Reranker Models with Sentence Transformers](https://huggingface.co/blog/train-multimodal-sentence-transformers): train multimodal embedding models, with a Visual Document Retrieval walkthrough.
585
+
586
+ ## Citation
587
+
588
+ ### BibTeX
589
+
590
+ #### Sentence Transformers
591
+ ```bibtex
592
+ @inproceedings{reimers-2019-sentence-bert,
593
+ title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
594
+ author = "Reimers, Nils and Gurevych, Iryna",
595
+ booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
596
+ month = "11",
597
+ year = "2019",
598
+ publisher = "Association for Computational Linguistics",
599
+ url = "https://arxiv.org/abs/1908.10084",
600
+ }
601
+ ```
602
+
603
+ #### MultipleNegativesRankingLoss
604
+ ```bibtex
605
+ @misc{oord2019representationlearningcontrastivepredictive,
606
+ title={Representation Learning with Contrastive Predictive Coding},
607
+ author={Aaron van den Oord and Yazhe Li and Oriol Vinyals},
608
+ year={2019},
609
+ eprint={1807.03748},
610
+ archivePrefix={arXiv},
611
+ primaryClass={cs.LG},
612
+ url={https://arxiv.org/abs/1807.03748},
613
+ }
614
+ ```
615
+
616
+ <!--
617
+ ## Glossary
618
+
619
+ *Clearly define terms in order to be accessible across audiences.*
620
+ -->
621
+
622
+ <!--
623
+ ## Model Card Authors
624
+
625
+ *Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
626
+ -->
627
+
628
+ <!--
629
+ ## Model Card Contact
630
+
631
+ *Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
632
+ -->
config.json ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_cross_attention": false,
3
+ "architectures": [
4
+ "BertModel"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "bos_token_id": null,
8
+ "classifier_dropout": null,
9
+ "dtype": "float32",
10
+ "eos_token_id": null,
11
+ "hidden_act": "gelu",
12
+ "hidden_dropout_prob": 0.1,
13
+ "hidden_size": 384,
14
+ "id2label": {
15
+ "0": "LABEL_0"
16
+ },
17
+ "initializer_range": 0.02,
18
+ "intermediate_size": 1536,
19
+ "is_decoder": false,
20
+ "label2id": {
21
+ "LABEL_0": 0
22
+ },
23
+ "layer_norm_eps": 1e-12,
24
+ "max_position_embeddings": 512,
25
+ "model_type": "bert",
26
+ "num_attention_heads": 12,
27
+ "num_hidden_layers": 12,
28
+ "pad_token_id": 0,
29
+ "position_embedding_type": "absolute",
30
+ "tie_word_embeddings": true,
31
+ "transformers_version": "5.18.0",
32
+ "type_vocab_size": 2,
33
+ "use_cache": false,
34
+ "vocab_size": 30522
35
+ }
config_sentence_transformers.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "__version__": {
3
+ "pytorch": "2.11.0+cu130",
4
+ "sentence_transformers": "5.7.0",
5
+ "transformers": "5.18.0"
6
+ },
7
+ "default_prompt_name": null,
8
+ "model_type": "SentenceTransformer",
9
+ "prompts": {
10
+ "document": "",
11
+ "query": ""
12
+ },
13
+ "similarity_fn_name": "cosine"
14
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:65f3aa1f420dab2017d62bd3eacbc4882dbf70d631f8246d6fb0d5adae855c37
3
+ size 133462128
modules.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "idx": 0,
4
+ "name": "0",
5
+ "path": "",
6
+ "type": "sentence_transformers.base.modules.transformer.Transformer"
7
+ },
8
+ {
9
+ "idx": 1,
10
+ "name": "1",
11
+ "path": "1_Pooling",
12
+ "type": "sentence_transformers.sentence_transformer.modules.pooling.Pooling"
13
+ },
14
+ {
15
+ "idx": 2,
16
+ "name": "2",
17
+ "path": "2_Normalize",
18
+ "type": "sentence_transformers.sentence_transformer.modules.normalize.Normalize"
19
+ }
20
+ ]
sentence_bert_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "transformer_task": "feature-extraction",
3
+ "modality_config": {
4
+ "text": {
5
+ "method": "forward",
6
+ "method_output_name": "last_hidden_state"
7
+ }
8
+ },
9
+ "module_output_name": "token_embeddings"
10
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "backend": "tokenizers",
3
+ "clean_up_tokenization_spaces": true,
4
+ "cls_token": "[CLS]",
5
+ "do_basic_tokenize": true,
6
+ "do_lower_case": true,
7
+ "is_local": true,
8
+ "local_files_only": false,
9
+ "mask_token": "[MASK]",
10
+ "max_length": 30,
11
+ "model_max_length": 30,
12
+ "never_split": null,
13
+ "pad_to_multiple_of": null,
14
+ "pad_token": "[PAD]",
15
+ "pad_token_type_id": 0,
16
+ "padding_side": "right",
17
+ "sep_token": "[SEP]",
18
+ "stride": 0,
19
+ "strip_accents": null,
20
+ "tokenize_chinese_chars": true,
21
+ "tokenizer_class": "BertTokenizer",
22
+ "truncation_side": "right",
23
+ "truncation_strategy": "longest_first",
24
+ "unk_token": "[UNK]"
25
+ }