fjmgAI commited on
Commit
034ff5d
·
verified ·
1 Parent(s): bdf11cb

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +80 -369
README.md CHANGED
@@ -1,6 +1,7 @@
1
  ---
2
  language:
3
  - en
 
4
  license: apache-2.0
5
  tags:
6
  - sentence-transformers
@@ -98,99 +99,29 @@ model-index:
98
  - type: ndcg@10
99
  value: 0.464
100
  name: Ndcg@10
 
 
101
  ---
 
102
 
103
- # EuroBERT-210m trained on GooAQ
104
 
105
- This is a [Cross Encoder](https://www.sbert.net/docs/cross_encoder/usage/usage.html) model finetuned from [EuroBERT/EuroBERT-210m](https://huggingface.co/EuroBERT/EuroBERT-210m) using the [sentence-transformers](https://www.SBERT.net) library. It computes scores for pairs of texts, which can be used for text reranking and semantic search.
106
-
107
- ## Model Details
108
-
109
- ### Model Description
110
- - **Model Type:** Cross Encoder
111
- - **Base model:** [EuroBERT/EuroBERT-210m](https://huggingface.co/EuroBERT/EuroBERT-210m) <!-- at revision 5a0c63d3e255a4f2005d3591d5508b7fd07cae94 -->
112
- - **Maximum Sequence Length:** 8192 tokens
113
- - **Number of Output Labels:** 1 label
114
- <!-- - **Training Dataset:** Unknown -->
115
- - **Language:** en
116
- - **License:** apache-2.0
117
-
118
- ### Model Sources
119
-
120
- - **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
121
- - **Documentation:** [Cross Encoder Documentation](https://www.sbert.net/docs/cross_encoder/usage/usage.html)
122
- - **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers)
123
- - **Hugging Face:** [Cross Encoders on Hugging Face](https://huggingface.co/models?library=sentence-transformers&other=cross-encoder)
124
-
125
- ## Usage
126
-
127
- ### Direct Usage (Sentence Transformers)
128
-
129
- First install the Sentence Transformers library:
130
-
131
- ```bash
132
- pip install -U sentence-transformers
133
- ```
134
-
135
- Then you can load this model and run inference.
136
- ```python
137
- from sentence_transformers import CrossEncoder
138
 
139
- # Download from the 🤗 Hub
140
- model = CrossEncoder("cross_encoder_model_id")
141
- # Get scores for pairs of texts
142
- pairs = [
143
- ['what are the risks with taking statins?', "['Muscle pain and damage. One of the most common complaints of people taking statins is muscle pain. ... ', 'Liver damage. Occasionally, statin use could cause an increase in the level of enzymes that signal liver inflammation. ... ', 'Increased blood sugar or type 2 diabetes. ... ', 'Neurological side effects.']"],
144
- ['what are the risks with taking statins?', 'Doctors discovered that statins can help lower blood pressure, as well as lower cholesterol. Statins are often prescribed to people with high cholesterol. Too much cholesterol in your blood increases your risk of heart attacks and strokes.'],
145
- ['what are the risks with taking statins?', 'Lipitor and Crestor are both effective statins that lower levels of “bad” cholesterol and increase levels of “good” cholesterol. While Crestor is the more potent statin, both medications are effective and have slightly different side effects and drug interactions.'],
146
- ['what are the risks with taking statins?', "About simvastatin Simvastatin belongs to a group of medicines called statins. It's used to lower cholesterol if you've been diagnosed with high blood cholesterol. It's also taken to prevent heart disease, including heart attacks and strokes."],
147
- ['what are the risks with taking statins?', 'Zetia works to lower cholesterol in a new way different from the statins: it inhibits the absorption of cholesterol in the small intestine, whereas the statins work by blocking cholesterol production in the liver.'],
148
- ]
149
- scores = model.predict(pairs)
150
- print(scores.shape)
151
- # (5,)
152
 
153
- # Or rank different texts based on similarity to a single text
154
- ranks = model.rank(
155
- 'what are the risks with taking statins?',
156
- [
157
- "['Muscle pain and damage. One of the most common complaints of people taking statins is muscle pain. ... ', 'Liver damage. Occasionally, statin use could cause an increase in the level of enzymes that signal liver inflammation. ... ', 'Increased blood sugar or type 2 diabetes. ... ', 'Neurological side effects.']",
158
- 'Doctors discovered that statins can help lower blood pressure, as well as lower cholesterol. Statins are often prescribed to people with high cholesterol. Too much cholesterol in your blood increases your risk of heart attacks and strokes.',
159
- 'Lipitor and Crestor are both effective statins that lower levels of “bad” cholesterol and increase levels of “good” cholesterol. While Crestor is the more potent statin, both medications are effective and have slightly different side effects and drug interactions.',
160
- "About simvastatin Simvastatin belongs to a group of medicines called statins. It's used to lower cholesterol if you've been diagnosed with high blood cholesterol. It's also taken to prevent heart disease, including heart attacks and strokes.",
161
- 'Zetia works to lower cholesterol in a new way different from the statins: it inhibits the absorption of cholesterol in the small intestine, whereas the statins work by blocking cholesterol production in the liver.',
162
- ]
163
- )
164
- # [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...]
165
- ```
166
-
167
- <!--
168
- ### Direct Usage (Transformers)
169
-
170
- <details><summary>Click to see the direct usage in Transformers</summary>
171
-
172
- </details>
173
- -->
174
-
175
- <!--
176
- ### Downstream Usage (Sentence Transformers)
177
-
178
- You can finetune this model on your own dataset.
179
-
180
- <details><summary>Click to expand</summary>
181
-
182
- </details>
183
- -->
184
-
185
- <!--
186
- ### Out-of-Scope Use
187
 
188
- *List how the model may foreseeably be misused and address what users ought not to do with the model.*
189
- -->
190
 
191
- ## Evaluation
 
192
 
193
- ### Metrics
 
194
 
195
  #### Cross Encoder Reranking
196
 
@@ -261,305 +192,85 @@ You can finetune this model on your own dataset.
261
  *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
262
  -->
263
 
264
- ## Training Details
265
-
266
- ### Training Dataset
267
-
268
- #### Unnamed Dataset
269
-
270
- * Size: 578,402 training samples
271
- * Columns: <code>question</code>, <code>answer</code>, and <code>label</code>
272
- * Approximate statistics based on the first 1000 samples:
273
- | | question | answer | label |
274
- |:--------|:-----------------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------|:------------------------------------------------|
275
- | type | string | string | int |
276
- | details | <ul><li>min: 20 characters</li><li>mean: 45.09 characters</li><li>max: 84 characters</li></ul> | <ul><li>min: 51 characters</li><li>mean: 253.23 characters</li><li>max: 374 characters</li></ul> | <ul><li>0: ~82.60%</li><li>1: ~17.40%</li></ul> |
277
- * Samples:
278
- | question | answer | label |
279
- |:-----------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:---------------|
280
- | <code>what are the risks with taking statins?</code> | <code>['Muscle pain and damage. One of the most common complaints of people taking statins is muscle pain. ... ', 'Liver damage. Occasionally, statin use could cause an increase in the level of enzymes that signal liver inflammation. ... ', 'Increased blood sugar or type 2 diabetes. ... ', 'Neurological side effects.']</code> | <code>1</code> |
281
- | <code>what are the risks with taking statins?</code> | <code>Doctors discovered that statins can help lower blood pressure, as well as lower cholesterol. Statins are often prescribed to people with high cholesterol. Too much cholesterol in your blood increases your risk of heart attacks and strokes.</code> | <code>0</code> |
282
- | <code>what are the risks with taking statins?</code> | <code>Lipitor and Crestor are both effective statins that lower levels of “bad” cholesterol and increase levels of “good” cholesterol. While Crestor is the more potent statin, both medications are effective and have slightly different side effects and drug interactions.</code> | <code>0</code> |
283
- * Loss: [<code>BinaryCrossEntropyLoss</code>](https://sbert.net/docs/package_reference/cross_encoder/losses.html#binarycrossentropyloss) with these parameters:
284
- ```json
285
- {
286
- "activation_fn": "torch.nn.modules.linear.Identity",
287
- "pos_weight": 5
288
- }
289
- ```
290
 
291
- ### Training Hyperparameters
292
- #### Non-Default Hyperparameters
293
-
294
- - `eval_strategy`: steps
295
- - `per_device_train_batch_size`: 64
296
- - `per_device_eval_batch_size`: 64
297
- - `learning_rate`: 2e-05
298
- - `num_train_epochs`: 2
299
- - `warmup_ratio`: 0.1
300
- - `seed`: 12
301
- - `fp16`: True
302
- - `dataloader_num_workers`: 4
303
- - `load_best_model_at_end`: True
304
-
305
- #### All Hyperparameters
306
- <details><summary>Click to expand</summary>
307
 
308
- - `overwrite_output_dir`: False
309
- - `do_predict`: False
310
- - `eval_strategy`: steps
311
- - `prediction_loss_only`: True
312
- - `per_device_train_batch_size`: 64
313
- - `per_device_eval_batch_size`: 64
314
- - `per_gpu_train_batch_size`: None
315
- - `per_gpu_eval_batch_size`: None
316
- - `gradient_accumulation_steps`: 1
317
- - `eval_accumulation_steps`: None
318
- - `torch_empty_cache_steps`: None
319
- - `learning_rate`: 2e-05
320
- - `weight_decay`: 0.0
321
- - `adam_beta1`: 0.9
322
- - `adam_beta2`: 0.999
323
- - `adam_epsilon`: 1e-08
324
- - `max_grad_norm`: 1.0
325
- - `num_train_epochs`: 2
326
- - `max_steps`: -1
327
- - `lr_scheduler_type`: linear
328
- - `lr_scheduler_kwargs`: {}
329
- - `warmup_ratio`: 0.1
330
- - `warmup_steps`: 0
331
- - `log_level`: passive
332
- - `log_level_replica`: warning
333
- - `log_on_each_node`: True
334
- - `logging_nan_inf_filter`: True
335
- - `save_safetensors`: True
336
- - `save_on_each_node`: False
337
- - `save_only_model`: False
338
- - `restore_callback_states_from_checkpoint`: False
339
- - `no_cuda`: False
340
- - `use_cpu`: False
341
- - `use_mps_device`: False
342
- - `seed`: 12
343
- - `data_seed`: None
344
- - `jit_mode_eval`: False
345
- - `use_ipex`: False
346
- - `bf16`: False
347
- - `fp16`: True
348
- - `fp16_opt_level`: O1
349
- - `half_precision_backend`: auto
350
- - `bf16_full_eval`: False
351
- - `fp16_full_eval`: False
352
- - `tf32`: None
353
- - `local_rank`: 0
354
- - `ddp_backend`: None
355
- - `tpu_num_cores`: None
356
- - `tpu_metrics_debug`: False
357
- - `debug`: []
358
- - `dataloader_drop_last`: False
359
- - `dataloader_num_workers`: 4
360
- - `dataloader_prefetch_factor`: None
361
- - `past_index`: -1
362
- - `disable_tqdm`: False
363
- - `remove_unused_columns`: True
364
- - `label_names`: None
365
- - `load_best_model_at_end`: True
366
- - `ignore_data_skip`: False
367
- - `fsdp`: []
368
- - `fsdp_min_num_params`: 0
369
- - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
370
- - `tp_size`: 0
371
- - `fsdp_transformer_layer_cls_to_wrap`: None
372
- - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
373
- - `deepspeed`: None
374
- - `label_smoothing_factor`: 0.0
375
- - `optim`: adamw_torch
376
- - `optim_args`: None
377
- - `adafactor`: False
378
- - `group_by_length`: False
379
- - `length_column_name`: length
380
- - `ddp_find_unused_parameters`: None
381
- - `ddp_bucket_cap_mb`: None
382
- - `ddp_broadcast_buffers`: False
383
- - `dataloader_pin_memory`: True
384
- - `dataloader_persistent_workers`: False
385
- - `skip_memory_metrics`: True
386
- - `use_legacy_prediction_loop`: False
387
- - `push_to_hub`: False
388
- - `resume_from_checkpoint`: None
389
- - `hub_model_id`: None
390
- - `hub_strategy`: every_save
391
- - `hub_private_repo`: None
392
- - `hub_always_push`: False
393
- - `gradient_checkpointing`: False
394
- - `gradient_checkpointing_kwargs`: None
395
- - `include_inputs_for_metrics`: False
396
- - `include_for_metrics`: []
397
- - `eval_do_concat_batches`: True
398
- - `fp16_backend`: auto
399
- - `push_to_hub_model_id`: None
400
- - `push_to_hub_organization`: None
401
- - `mp_parameters`:
402
- - `auto_find_batch_size`: False
403
- - `full_determinism`: False
404
- - `torchdynamo`: None
405
- - `ray_scope`: last
406
- - `ddp_timeout`: 1800
407
- - `torch_compile`: False
408
- - `torch_compile_backend`: None
409
- - `torch_compile_mode`: None
410
- - `include_tokens_per_second`: False
411
- - `include_num_input_tokens_seen`: False
412
- - `neftune_noise_alpha`: None
413
- - `optim_target_modules`: None
414
- - `batch_eval_metrics`: False
415
- - `eval_on_start`: False
416
- - `use_liger_kernel`: False
417
- - `eval_use_gather_object`: False
418
- - `average_tokens_across_devices`: False
419
- - `prompts`: None
420
- - `batch_sampler`: batch_sampler
421
- - `multi_dataset_batch_sampler`: proportional
422
 
423
- </details>
 
 
424
 
425
- ### Training Logs
426
- | Epoch | Step | Training Loss | gooaq-dev_ndcg@10 | NanoMSMARCO_R100_ndcg@10 | NanoNFCorpus_R100_ndcg@10 | NanoNQ_R100_ndcg@10 | NanoBEIR_R100_mean_ndcg@10 |
427
- |:----------:|:---------:|:-------------:|:--------------------:|:------------------------:|:-------------------------:|:--------------------:|:--------------------------:|
428
- | -1 | -1 | - | 0.1245 (-0.4667) | 0.0299 (-0.5105) | 0.2478 (-0.0773) | 0.0258 (-0.4748) | 0.1012 (-0.3542) |
429
- | 0.0001 | 1 | 1.1115 | - | - | - | - | - |
430
- | 0.0221 | 200 | 1.1756 | - | - | - | - | - |
431
- | 0.0443 | 400 | 1.0297 | - | - | - | - | - |
432
- | 0.0664 | 600 | 0.828 | - | - | - | - | - |
433
- | 0.0885 | 800 | 0.7511 | - | - | - | - | - |
434
- | 0.1106 | 1000 | 0.7224 | 0.6667 (+0.0754) | 0.4106 (-0.1299) | 0.2898 (-0.0353) | 0.4147 (-0.0860) | 0.3717 (-0.0837) |
435
- | 0.1328 | 1200 | 0.7315 | - | - | - | - | - |
436
- | 0.1549 | 1400 | 0.7403 | - | - | - | - | - |
437
- | 0.1770 | 1600 | 0.731 | - | - | - | - | - |
438
- | 0.1992 | 1800 | 0.7495 | - | - | - | - | - |
439
- | 0.2213 | 2000 | 0.7227 | 0.6639 (+0.0727) | 0.4653 (-0.0751) | 0.4152 (+0.0901) | 0.4534 (-0.0472) | 0.4446 (-0.0107) |
440
- | 0.2434 | 2200 | 0.7002 | - | - | - | - | - |
441
- | 0.2655 | 2400 | 0.6803 | - | - | - | - | - |
442
- | 0.2877 | 2600 | 0.6812 | - | - | - | - | - |
443
- | 0.3098 | 2800 | 0.6788 | - | - | - | - | - |
444
- | 0.3319 | 3000 | 0.6792 | 0.6908 (+0.0996) | 0.5115 (-0.0289) | 0.3722 (+0.0472) | 0.3895 (-0.1112) | 0.4244 (-0.0310) |
445
- | 0.3541 | 3200 | 0.6701 | - | - | - | - | - |
446
- | 0.3762 | 3400 | 0.6442 | - | - | - | - | - |
447
- | 0.3983 | 3600 | 0.6488 | - | - | - | - | - |
448
- | 0.4204 | 3800 | 0.6252 | - | - | - | - | - |
449
- | 0.4426 | 4000 | 0.6368 | 0.6973 (+0.1061) | 0.5330 (-0.0074) | 0.4015 (+0.0765) | 0.5109 (+0.0103) | 0.4818 (+0.0264) |
450
- | 0.4647 | 4200 | 0.6455 | - | - | - | - | - |
451
- | 0.4868 | 4400 | 0.6164 | - | - | - | - | - |
452
- | 0.5090 | 4600 | 0.6436 | - | - | - | - | - |
453
- | 0.5311 | 4800 | 0.6094 | - | - | - | - | - |
454
- | 0.5532 | 5000 | 0.6287 | 0.6933 (+0.1021) | 0.5572 (+0.0168) | 0.3381 (+0.0131) | 0.4847 (-0.0159) | 0.4600 (+0.0046) |
455
- | 0.5753 | 5200 | 0.6131 | - | - | - | - | - |
456
- | 0.5975 | 5400 | 0.6179 | - | - | - | - | - |
457
- | 0.6196 | 5600 | 0.6007 | - | - | - | - | - |
458
- | 0.6417 | 5800 | 0.5961 | - | - | - | - | - |
459
- | 0.6639 | 6000 | 0.6156 | 0.7059 (+0.1146) | 0.5424 (+0.0020) | 0.3565 (+0.0315) | 0.5356 (+0.0349) | 0.4782 (+0.0228) |
460
- | 0.6860 | 6200 | 0.5884 | - | - | - | - | - |
461
- | 0.7081 | 6400 | 0.5824 | - | - | - | - | - |
462
- | 0.7303 | 6600 | 0.5692 | - | - | - | - | - |
463
- | 0.7524 | 6800 | 0.5979 | - | - | - | - | - |
464
- | 0.7745 | 7000 | 0.5843 | 0.7206 (+0.1293) | 0.5639 (+0.0235) | 0.3664 (+0.0414) | 0.5308 (+0.0301) | 0.4870 (+0.0317) |
465
- | 0.7966 | 7200 | 0.5864 | - | - | - | - | - |
466
- | 0.8188 | 7400 | 0.5852 | - | - | - | - | - |
467
- | 0.8409 | 7600 | 0.5826 | - | - | - | - | - |
468
- | 0.8630 | 7800 | 0.5817 | - | - | - | - | - |
469
- | 0.8852 | 8000 | 0.5666 | 0.7267 (+0.1355) | 0.5411 (+0.0007) | 0.3818 (+0.0567) | 0.5273 (+0.0266) | 0.4834 (+0.0280) |
470
- | 0.9073 | 8200 | 0.5776 | - | - | - | - | - |
471
- | 0.9294 | 8400 | 0.5667 | - | - | - | - | - |
472
- | 0.9515 | 8600 | 0.5587 | - | - | - | - | - |
473
- | 0.9737 | 8800 | 0.5593 | - | - | - | - | - |
474
- | 0.9958 | 9000 | 0.5567 | 0.7298 (+0.1385) | 0.5091 (-0.0313) | 0.3617 (+0.0366) | 0.4834 (-0.0173) | 0.4514 (-0.0040) |
475
- | 1.0179 | 9200 | 0.4478 | - | - | - | - | - |
476
- | 1.0401 | 9400 | 0.4039 | - | - | - | - | - |
477
- | 1.0622 | 9600 | 0.4101 | - | - | - | - | - |
478
- | 1.0843 | 9800 | 0.3935 | - | - | - | - | - |
479
- | 1.1064 | 10000 | 0.396 | 0.7338 (+0.1426) | 0.5604 (+0.0200) | 0.3518 (+0.0268) | 0.5027 (+0.0021) | 0.4716 (+0.0163) |
480
- | 1.1286 | 10200 | 0.4077 | - | - | - | - | - |
481
- | 1.1507 | 10400 | 0.4245 | - | - | - | - | - |
482
- | 1.1728 | 10600 | 0.3953 | - | - | - | - | - |
483
- | 1.1950 | 10800 | 0.3912 | - | - | - | - | - |
484
- | 1.2171 | 11000 | 0.3943 | 0.7319 (+0.1407) | 0.5296 (-0.0108) | 0.3645 (+0.0395) | 0.4739 (-0.0268) | 0.4560 (+0.0006) |
485
- | 1.2392 | 11200 | 0.4032 | - | - | - | - | - |
486
- | 1.2613 | 11400 | 0.4063 | - | - | - | - | - |
487
- | 1.2835 | 11600 | 0.3909 | - | - | - | - | - |
488
- | 1.3056 | 11800 | 0.3759 | - | - | - | - | - |
489
- | 1.3277 | 12000 | 0.3944 | 0.7440 (+0.1528) | 0.4978 (-0.0426) | 0.3138 (-0.0112) | 0.5306 (+0.0299) | 0.4474 (-0.0080) |
490
- | 1.3499 | 12200 | 0.3721 | - | - | - | - | - |
491
- | 1.3720 | 12400 | 0.3888 | - | - | - | - | - |
492
- | 1.3941 | 12600 | 0.3838 | - | - | - | - | - |
493
- | 1.4162 | 12800 | 0.4023 | - | - | - | - | - |
494
- | 1.4384 | 13000 | 0.3831 | 0.7470 (+0.1558) | 0.5280 (-0.0124) | 0.3479 (+0.0229) | 0.5049 (+0.0043) | 0.4603 (+0.0049) |
495
- | 1.4605 | 13200 | 0.4027 | - | - | - | - | - |
496
- | 1.4826 | 13400 | 0.3977 | - | - | - | - | - |
497
- | 1.5048 | 13600 | 0.384 | - | - | - | - | - |
498
- | 1.5269 | 13800 | 0.3773 | - | - | - | - | - |
499
- | 1.5490 | 14000 | 0.3854 | 0.7447 (+0.1534) | 0.5363 (-0.0041) | 0.3637 (+0.0386) | 0.5521 (+0.0515) | 0.4840 (+0.0287) |
500
- | 1.5711 | 14200 | 0.3632 | - | - | - | - | - |
501
- | 1.5933 | 14400 | 0.3902 | - | - | - | - | - |
502
- | 1.6154 | 14600 | 0.3862 | - | - | - | - | - |
503
- | 1.6375 | 14800 | 0.3702 | - | - | - | - | - |
504
- | 1.6597 | 15000 | 0.374 | 0.7508 (+0.1595) | 0.4950 (-0.0455) | 0.3909 (+0.0658) | 0.5166 (+0.0159) | 0.4675 (+0.0121) |
505
- | 1.6818 | 15200 | 0.3814 | - | - | - | - | - |
506
- | 1.7039 | 15400 | 0.3713 | - | - | - | - | - |
507
- | 1.7260 | 15600 | 0.3697 | - | - | - | - | - |
508
- | 1.7482 | 15800 | 0.3538 | - | - | - | - | - |
509
- | 1.7703 | 16000 | 0.3797 | 0.7491 (+0.1579) | 0.5119 (-0.0285) | 0.3635 (+0.0384) | 0.4958 (-0.0049) | 0.4570 (+0.0017) |
510
- | 1.7924 | 16200 | 0.374 | - | - | - | - | - |
511
- | 1.8146 | 16400 | 0.3874 | - | - | - | - | - |
512
- | 1.8367 | 16600 | 0.3568 | - | - | - | - | - |
513
- | 1.8588 | 16800 | 0.3653 | - | - | - | - | - |
514
- | 1.8809 | 17000 | 0.3706 | 0.7552 (+0.1640) | 0.5025 (-0.0379) | 0.3662 (+0.0412) | 0.5036 (+0.0029) | 0.4574 (+0.0020) |
515
- | 1.9031 | 17200 | 0.3636 | - | - | - | - | - |
516
- | 1.9252 | 17400 | 0.3606 | - | - | - | - | - |
517
- | 1.9473 | 17600 | 0.3599 | - | - | - | - | - |
518
- | 1.9695 | 17800 | 0.3535 | - | - | - | - | - |
519
- | **1.9916** | **18000** | **0.3553** | **0.7579 (+0.1667)** | **0.5106 (-0.0298)** | **0.3632 (+0.0381)** | **0.5182 (+0.0176)** | **0.4640 (+0.0086)** |
520
-
521
- * The bold row denotes the saved checkpoint.
522
 
523
- ### Framework Versions
524
- - Python: 3.11.12
525
- - Sentence Transformers: 4.0.2
526
- - Transformers: 4.51.2
527
- - PyTorch: 2.6.0+cu126
528
- - Accelerate: 1.6.0
529
- - Datasets: 3.5.0
530
- - Tokenizers: 0.21.1
 
 
 
 
 
531
 
532
- ## Citation
533
-
534
- ### BibTeX
535
-
536
- #### Sentence Transformers
537
- ```bibtex
538
- @inproceedings{reimers-2019-sentence-bert,
539
- title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
540
- author = "Reimers, Nils and Gurevych, Iryna",
541
- booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
542
- month = "11",
543
- year = "2019",
544
- publisher = "Association for Computational Linguistics",
545
- url = "https://arxiv.org/abs/1908.10084",
546
- }
547
  ```
548
 
549
  <!--
550
- ## Glossary
 
 
551
 
552
- *Clearly define terms in order to be accessible across audiences.*
553
  -->
554
 
555
  <!--
556
- ## Model Card Authors
 
 
 
 
557
 
558
- *Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
559
  -->
560
 
561
  <!--
562
- ## Model Card Contact
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
563
 
564
- *Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
565
- -->
 
1
  ---
2
  language:
3
  - en
4
+ - es
5
  license: apache-2.0
6
  tags:
7
  - sentence-transformers
 
99
  - type: ndcg@10
100
  value: 0.464
101
  name: Ndcg@10
102
+ datasets:
103
+ - sentence-transformers/gooaq
104
  ---
105
+ [<img src="https://cdn-avatars.huggingface.co/v1/production/uploads/67b2f4e49edebc815a3a4739/R1g957j1aBbx8lhZbWmxw.jpeg" width="200"/>](https://huggingface.co/fjmgAI)
106
 
107
+ ## Fine-Tuned Model
108
 
109
+ **`fjmgAI/rerank1-210M-EuroBERT`**
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
110
 
111
+ ## Base Model
112
+ **`EuroBERT/EuroBERT-210m`**
 
 
 
 
 
 
 
 
 
 
 
113
 
114
+ ## Fine-Tuning Method
115
+ This is a [Cross Encoder](https://www.sbert.net/docs/cross_encoder/usage/usage.html) model finetuned from [EuroBERT/EuroBERT-210m](https://huggingface.co/EuroBERT/EuroBERT-210m) using the [sentence-transformers](https://www.SBERT.net) library. It computes scores for pairs of texts, which can be used for text reranking and semantic search.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
116
 
117
+ ## Dataset
118
+ **[`sentence-transformers/gooaq`](https://huggingface.co/datasets/sentence-transformers/gooaq)**
119
 
120
+ ### Description
121
+ This dataset is a collection of question-answer pairs, collected from Google.
122
 
123
+ ## Fine-Tuning Details
124
+ - The model was trained using 578,402 training samples from sentence-transformer.
125
 
126
  #### Cross Encoder Reranking
127
 
 
192
  *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
193
  -->
194
 
195
+ ## Usage
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
196
 
197
+ ### Direct Usage (Sentence Transformers)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
198
 
199
+ First install the Sentence Transformers library:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
200
 
201
+ ```bash
202
+ pip install -U sentence-transformers
203
+ ```
204
 
205
+ Then you can load this model and run inference.
206
+ ```python
207
+ from sentence_transformers import CrossEncoder
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
208
 
209
+ # Download from the 🤗 Hub
210
+ model = CrossEncoder("fjmgAI/rerank1-210M-EuroBERT")
211
+ # Get scores for pairs of texts
212
+ pairs = [
213
+ ['what are the risks with taking statins?', "['Muscle pain and damage. One of the most common complaints of people taking statins is muscle pain. ... ', 'Liver damage. Occasionally, statin use could cause an increase in the level of enzymes that signal liver inflammation. ... ', 'Increased blood sugar or type 2 diabetes. ... ', 'Neurological side effects.']"],
214
+ ['what are the risks with taking statins?', 'Doctors discovered that statins can help lower blood pressure, as well as lower cholesterol. Statins are often prescribed to people with high cholesterol. Too much cholesterol in your blood increases your risk of heart attacks and strokes.'],
215
+ ['what are the risks with taking statins?', 'Lipitor and Crestor are both effective statins that lower levels of “bad” cholesterol and increase levels of “good” cholesterol. While Crestor is the more potent statin, both medications are effective and have slightly different side effects and drug interactions.'],
216
+ ['what are the risks with taking statins?', "About simvastatin Simvastatin belongs to a group of medicines called statins. It's used to lower cholesterol if you've been diagnosed with high blood cholesterol. It's also taken to prevent heart disease, including heart attacks and strokes."],
217
+ ['what are the risks with taking statins?', 'Zetia works to lower cholesterol in a new way different from the statins: it inhibits the absorption of cholesterol in the small intestine, whereas the statins work by blocking cholesterol production in the liver.'],
218
+ ]
219
+ scores = model.predict(pairs)
220
+ print(scores.shape)
221
+ # (5,)
222
 
223
+ # Or rank different texts based on similarity to a single text
224
+ ranks = model.rank(
225
+ 'what are the risks with taking statins?',
226
+ [
227
+ "['Muscle pain and damage. One of the most common complaints of people taking statins is muscle pain. ... ', 'Liver damage. Occasionally, statin use could cause an increase in the level of enzymes that signal liver inflammation. ... ', 'Increased blood sugar or type 2 diabetes. ... ', 'Neurological side effects.']",
228
+ 'Doctors discovered that statins can help lower blood pressure, as well as lower cholesterol. Statins are often prescribed to people with high cholesterol. Too much cholesterol in your blood increases your risk of heart attacks and strokes.',
229
+ 'Lipitor and Crestor are both effective statins that lower levels of “bad” cholesterol and increase levels of “good” cholesterol. While Crestor is the more potent statin, both medications are effective and have slightly different side effects and drug interactions.',
230
+ "About simvastatin Simvastatin belongs to a group of medicines called statins. It's used to lower cholesterol if you've been diagnosed with high blood cholesterol. It's also taken to prevent heart disease, including heart attacks and strokes.",
231
+ 'Zetia works to lower cholesterol in a new way different from the statins: it inhibits the absorption of cholesterol in the small intestine, whereas the statins work by blocking cholesterol production in the liver.',
232
+ ]
233
+ )
234
+ # [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...]
 
 
 
235
  ```
236
 
237
  <!--
238
+ ### Direct Usage (Transformers)
239
+
240
+ <details><summary>Click to see the direct usage in Transformers</summary>
241
 
242
+ </details>
243
  -->
244
 
245
  <!--
246
+ ### Downstream Usage (Sentence Transformers)
247
+
248
+ You can finetune this model on your own dataset.
249
+
250
+ <details><summary>Click to expand</summary>
251
 
252
+ </details>
253
  -->
254
 
255
  <!--
256
+ ### Out-of-Scope Use
257
+
258
+ *List how the model may foreseeably be misused and address what users ought not to do with the model.*
259
+ -->
260
+
261
+ ### Framework Versions
262
+ - Python: 3.11.12
263
+ - Sentence Transformers: 4.0.2
264
+ - Transformers: 4.51.2
265
+ - PyTorch: 2.6.0+cu126
266
+ - Accelerate: 1.6.0
267
+ - Datasets: 3.5.0
268
+ - Tokenizers: 0.21.1
269
+
270
+ ## Purpose
271
+ This tuned reranker model is optimized for **Spanish and English applications**, prioritizing **accurate reordering of results** by leveraging semantic similarity through refined embedding comparisons, ideal for enhancing **question-answering** and **document retrieval** tasks.
272
+
273
+ - **Developed by:** fjmgAI
274
+ - **License:** apache-2.0
275
 
276
+ [<img src="https://sbert.net/_static/logo.png" width="200"/>](https://github.com/UKPLab/sentence-transformers)