Datasets:
File size: 41,271 Bytes
6aeb886 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 | ---
tags:
- sentence-transformers
- sentence-similarity
- feature-extraction
- dense
- generated_from_trainer
- dataset_size:59612
- loss:CachedMultipleNegativesRankingLoss
- loss:MultipleNegativesRankingLoss
base_model: intfloat/multilingual-e5-base
widget:
- source_sentence: 'query: Erasmus students at Liverpool John Moores University
Could the Commission provide details of how many Erasmus students studied at Liverpool
John Moores University between 2004 and 2014?'
sentences:
- 'passage: The Jean Monnet Action offers chairs and modules to support scholars
in developing European Union studies.
However, no scholars from Liverpool John Moores University have benefitted from
Jean Monnet Action''s support between 2004 and 2014, nor before.'
- 'passage: r Hogano
n behalf of the Commission(16.2.2018)
The Commission is aware of the widespread use in Mexico of the term ''manchego''
associated to a type of cheese whose c
haracteristics greatly differ from the EU geographical indication ''Queso Manchego'',
which has never been protected in Mexico.
''Queso Manchego'' is one of the geographical indications submitted by the EU
to Mexico for protection under the modernised F
ree Trade Agreement. Negotiations on geographical indications are on-going and
the Commisson will make all efforts to obtain the most satisfactory result possible
on every term submitted to Mexico.'
- 'passage: Statistics for Erasmus mobility broken down at the level of detail requested
by the Honourable Member are available only since the beginning of the Lifelong
Learning Programme in 2007. Since the academic year 2007/2008 until the academic
year 2012/2013, the last year for which data are currently available, the four
higher education institutions received the following numbers of Erasmus students
:
Manchester Metropolitan University: 1.631
Salford University: 1.299
University of Liverpool: 823
Liverpool John Moores University: 447
The full breakdown of the data by academic year and institution of destination
is attached.'
- source_sentence: 'query: Seat of the European Parliament
What are the Council’s intentions as regards the seat of the European Parliament?
Though Parliament extends over three cities, only one has genuinely been regarded
as its seat from the outset: Strasbourg.
Transport links to and from Strasbourg are improving all the time - the high-speed
rail link and the new airlines there are major factors in this - and ensuring
security on the Strasbourg site is much more straightforward than in Brussels.
Can the Council make a clear statement in favour of upgrading Strasbourg?'
sentences:
- 'passage: Article 341 of the TFEU provides that the seat of the institutions of
the Union shall be determined by common accord of the Governments of the Member
States. In accordance with paragraph (a) of the sole article of the sixth protocol
annexed to the Treaties, ''the European Parliament shall have its seat in Strasbourg
where the 12 periods of monthly plenary sessions, including the budget session,
shall be held.''
With regard to the questions posed by the Honourable Member, the Council has not
discussed these issues and is not in a position to give an answer.'
- 'passage: The Commission has referred the questions on how the European Food Safety
Authority (EFSA) strives to ensure that its workforce is representative of European
citizens to the EFSA, as the Authority is an independent EU agency with responsibility
for its staff matters. The response will be communicated directly to the Honourable
Member as soon as possible.'
- 'passage: The matter of the location of the seats of the institutions is addressed
by Protocol No 6 to the Treaty.
It is not for the Council to comment on the seat of the European Parliament.'
- source_sentence: 'query: What more can the European Fisheries Control Agency do
to have a workforce that is representative of EU citizens?
Previously, the Commission has stated that ‘the Commission strives to be representative
of the increasingly diverse European citizens it serves’ (response to Written
Questions [REF], [REF] and [REF]).
How does the European Fisheries Control Agency think it’s currently doing on having
a workforce that is representative of EU citizens?
What more can the European Fisheries Control Agency do to be more representative
of EU citizens?'
sentences:
- 'passage: The Commission has asked the European Fisheries Control Agency (EFCA)
to provide a response to the question raised by the Honourable Member. The Agency''s
reply will be sent by the Commission to the Honourable Member as soon as possible.'
- 'passage: As a follow-up to its recent response to written questions [REF] and
[REF][1] on how the European Food Safety Authority (EFSA) strives to ensure that
its workforce is representative of European citizens, the Commission has provided
to the Honourable Member on behalf of the EFSA, an extensive list of activities
EFSA is undertaking to ensure that its workforce is representative of the European
Union citizens.
To respond to the present question, the Commission has asked EFSA to provide 10
examples of its efforts to be representative of EU citizens in practice. The response
will be communicated directly to the Honourable Member as soon as possible.
[1] http://www.europarl.europa.eu/plenary/en/parliamentary-questions.html'
- 'passage: The President of the European Council has undertaken to reply to parliamentary
questions in so far as they concern his own political activities. As discussions
held in the context of the EU-Ukraine Summit of 27 April 2015 fall outside the
scope of this commitment, the President of the European Council is not in a position
to answer the question put by the Honourable Members.'
- source_sentence: 'query: Update on the case involving two Italian marines
Can the Commission report on how discussions are progressing with the Indian authorities
on the case involving two Italian marines - a sequence of events which started
in 2012 but seems no closer to a conclusion that would definitively return the
two navy riflemen to Italy?'
sentences:
- 'passage: The EU has been following the case of the two Italian marines since
its beginning, in close contact with both the Italian and Indian authorities.
The High Representative Catherine Ashton has consistently raised this issue in
her encounters with representatives of the Indian government in recent months.
On all these occasions the EU has encouraged India to find a rapid and satisfactory
outcome to this dispute as soon as possible, based on the UN Convention on the
Law of the Sea and international law. The EU has also stressed several times that
the delays in the trial against the two marines are totally unacceptable.
The EU welcomes the decision of India''s Supreme Court to let one of the two marines
go back to Italy to receive medical treatment after he suffered a stroke at the
end of August.'
- 'passage: The Commission is aware of mounting price pressure in the dairy sector
since the beginning of the year, following a period of historically high prices,
and as a consequence of oversupply.
The Commission does not rate Member States on the application of the “Milk Package”
For details about how Member States and operators make use of the “Milk Package”
the Honourable Member is referred to the Commission Report of 13.6.2014[1].
The Commission can confirm that Spain has satisfactorily implemented the Rural
Development measures on ''Modernisation of agricultural holdings'' and ''Increasing
the added value of basic agricultural and forestry production''.
[1] COM(2014)354 http://ec.europa.eu/agriculture/milk/milk-package/com-2014-354_en.pdf'
- 'passage: The EU has been following the case of the two Italian marines in close
contact with the Italian government and has repeatedly called for a swift solution
to this longstanding case in accordance with international law and the UN Convention
on the Law of the Sea (UNCLOS). On 9th April 2015 the Indian Supreme Court decided
to extend Mr. Latorre''s medical leave in Italy until 15th July 2015.
The EU continues to support efforts towards a mutually acceptable solution that
would allow the return of the two marines.'
- source_sentence: 'query: Atlantic Trench
In its answer to question [REF] from this Member, the Commission noted that, under
Article 35 of the Euratom Treaty, the Commission can verify Spain’s monitoring
of marine radioactivity, but that, currently, ‘it is not included in the Commission
verification mission planning for 2018-2020’.
1. Is there any possibility of including it in said planning or in the next one?
2. What steps are required to ensure this monitoring is included in the verifications
conducted by the Commission?'
sentences:
- 'passage: Implementation of Article 16 of the Tobacco Products Directive[1] is
an ongoing process. Last year the Commission published a feasibility study[2]
which was followed by a targeted stakeholder consultation. The feasibility study
contained an initial mapping of available technologies for security features.
The preparatory work for drafting the necessary implementing acts continues and
will include a public consultation, stakeholder workshops and an implementation
study.
[1] Directive 2014/40/EU of 3 April 2014, OJ L 127, 29.4.2014, p. 1
[2] http://ec.europa.eu/health/tobacco/docs/2015_tpd_tracking_tracing_frep_en.pdf'
- 'passage: The Commission has not received any evidence that new studies on the
state of radioactive waste in the Atlantic trenches are necessary.
Monitoring of marine radioactivity by Spain can be covered by verification mission
by the Commission under Article 35 of the Euratom Treaty. Currently, it is not
included in the Commission verification mission planning for 2018-2020.'
- 'passage: Euratom Article 35 verifications are carried out according to a three-year
rolling programme, which is updated by the Commission services every six months.
This programme takes into account general principles such as: adequate coverage
of all Member States; inclusion of installations across the complete spectrum
of the nuclear fuel cycle as well as nation-wide monitoring systems. Verification
targets are selected according to their radiological significance or public interest.
The Commission services have, in June 2018, identified marine monitoring on the
coast of Galicia as a verification target in 2019. The Commission is following
the situation and will review the verification planning in December 2018, where
verification targets for 2019 will be confirmed.'
pipeline_tag: sentence-similarity
library_name: sentence-transformers
---
# SentenceTransformer based on intfloat/multilingual-e5-base
This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [intfloat/multilingual-e5-base](https://huggingface.co/intfloat/multilingual-e5-base). It maps inputs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, classification, clustering, and more.
## Model Details
### Model Description
- **Model Type:** Sentence Transformer
- **Base model:** [intfloat/multilingual-e5-base](https://huggingface.co/intfloat/multilingual-e5-base) <!-- at revision d128750597153bb5987e10b1c3493a34e5a4502a -->
- **Maximum Sequence Length:** 512 tokens
- **Output Dimensionality:** 768 dimensions
- **Similarity Function:** Cosine Similarity
- **Supported Modality:** Text
<!-- - **Training Dataset:** Unknown -->
<!-- - **Language:** Unknown -->
<!-- - **License:** Unknown -->
### Model Sources
- **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
- **Repository:** [Sentence Transformers on GitHub](https://github.com/huggingface/sentence-transformers)
- **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)
### Full Model Architecture
```
SentenceTransformer(
(0): Transformer({'transformer_task': 'feature-extraction', 'modality_config': {'text': {'method': 'forward', 'method_output_name': 'last_hidden_state'}}, 'module_output_name': 'token_embeddings', 'architecture': 'XLMRobertaModel'})
(1): Pooling({'embedding_dimension': 768, 'pooling_mode': 'mean', 'include_prompt': True})
(2): Normalize({'module_input_name': 'sentence_embedding', 'module_output_name': 'sentence_embedding'})
)
```
## Usage
### Direct Usage (Sentence Transformers)
First install the Sentence Transformers library:
```bash
pip install -U sentence-transformers
```
Then you can load this model and run inference.
```python
from sentence_transformers import SentenceTransformer
# Download from the 🤗 Hub
model = SentenceTransformer("sentence_transformers_model_id")
# Run inference
sentences = [
'query: Atlantic Trench\nIn its answer to question [REF] from this Member, the Commission noted that, under Article 35 of the Euratom Treaty, the Commission can verify Spain’s monitoring of marine radioactivity, but that, currently, ‘it is not included in the Commission verification mission planning for 2018-2020’.\n1. Is there any possibility of including it in said planning or in the next one?\n2. What steps are required to ensure this monitoring is included in the verifications conducted by the Commission?',
'passage: Euratom Article 35 verifications are carried out according to a three-year rolling programme, which is updated by the Commission services every six months. This programme takes into account general principles such as: adequate coverage of all Member States; inclusion of installations across the complete spectrum of the nuclear fuel cycle as well as nation-wide monitoring systems. Verification targets are selected according to their radiological significance or public interest.\nThe Commission services have, in June 2018, identified marine monitoring on the coast of Galicia as a verification target in 2019. The Commission is following the situation and will review the verification planning in December 2018, where verification targets for 2019 will be confirmed.',
'passage: The Commission has not received any evidence that new studies on the state of radioactive waste in the Atlantic trenches are necessary.\nMonitoring of marine radioactivity by Spain can be covered by verification mission by the Commission under Article 35 of the Euratom Treaty. Currently, it is not included in the Commission verification mission planning for 2018-2020.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]
# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.7552, 0.6100],
# [0.7552, 1.0000, 0.5385],
# [0.6100, 0.5385, 1.0000]])
```
<!--
### Direct Usage (Transformers)
<details><summary>Click to see the direct usage in Transformers</summary>
</details>
-->
<!--
### Downstream Usage (Sentence Transformers)
You can finetune this model on your own dataset.
<details><summary>Click to expand</summary>
</details>
-->
<!--
### Out-of-Scope Use
*List how the model may foreseeably be misused and address what users ought not to do with the model.*
-->
<!--
## Bias, Risks and Limitations
*What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
-->
<!--
### Recommendations
*What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
-->
## Training Details
### Training Dataset
#### Unnamed Dataset
* Size: 59,612 training samples
* Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
* Approximate statistics based on the first 100 samples:
| | anchor | positive | negative |
|:---------|:------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------|
| type | string | string | string |
| modality | text | text | text |
| details | <ul><li>min: 29 tokens</li><li>mean: 235.8 tokens</li><li>max: 483 tokens</li></ul> | <ul><li>min: 37 tokens</li><li>mean: 327.66 tokens</li><li>max: 512 tokens</li></ul> | <ul><li>min: 94 tokens</li><li>mean: 392.99 tokens</li><li>max: 512 tokens</li></ul> |
* Samples:
| anchor | positive | negative |
|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| <code>query: EU support for the Vál valley narrow-gauge railway<br>After a delay of several months, Prime Minister Viktor Orbán inaugurated the Vál valley narrow-gauge railway, co-financed by the EU, on 30 April 2016. In October 2013, by Order 1693/2013 (X.2.), the Hungarian Government had declared the narrow-gauge railway project a priority investment, and set aside EU development funding from the Central Transdanubia Operational Programme (KDOP-2.1.1.) for it. In the summer of 2015, a foundation belonging to the Mayor of Felcsút, Lőrinc Mészáros, the ‘Felcsúti Utánpótlás Nevelésért Alapítvány’, which is worth billions and which runs the football academy, secured non-refundable EU funding of HUF 600 million from the European Regional Development Fund. However, this sum was not enough to pay for the construction of the line, with its three stops, so the government contributed a further HUF 257 million. The narrow-gauge line, which is 5.7 km long, has been built in the village where the Prime Mi...</code> | <code>passage: The Commission asked the Hungarian authorities to provide additional information on the project mentioned by the Honourable Member. On the basis of the information received, there is no evidence to prove the existence of irregularities in this project. It is necessary to underline that cohesion policy is implemented in the framework of the so-called shared management, whereby Member States are responsible for the selection of operations under an operational programme.<br>According to the information available to the Commission, data provided in the application and in the grant contract with the beneficiary of the project regarding amount expected of passengers do not correspond with those presented by the Honourable Member. The mandatory results to be achieved (including the passenger numbers) were agreed between the programme's managing authority and the beneficiary and are included in the grant contract. They will therefore be verified before carrying out the final payment to t...</code> | <code>passage: The state of development of the two core network corridors and the priorities for their further development are detailed in the Coordinators' work plans[1].<br>The two Spanish branches of the Mediterranean corridor are both integral parts of the Corridor and not variants; they have the same priority. Projects along the core network corridors receive EU co-funding once selected following a competitive tendering procedure open to all Member States. They are selected notably on the basis of their relevance for the corridor development and their maturity. Under the "use-it-or-lose-it" principle, EU funds, allocated to a specific project by a Member State, may be re-attributed to other projects if the funded project is severely delayed or not implemented.<br>The Commission is well aware of the technical and financial challenges arising from the different gauge used by the Spanish railway network. The Commission welcomes the policy of the Spanish authorities to implement the European gaug...</code> |
| <code>query: Effectiveness of fines for breaking commercial laws<br>The EU is based on the rule of law. Imposing dissuasive sanctions for not complying with European law is justified. Yet it seems that sanctions are being imposed on large enterprises more and more often, which is worrying in terms of the rule of law.<br>In the case of Google, the regularity and severity of the fines imposed by the EU has raised doubts about their effectiveness in changing the behaviour of large enterprises. Nevertheless, the reports recently submitted by Google to the Commission have not resulted in more fines or sanctions, with the Commissioner concerned insisting that progress had been made. The severity of the sanctions imposed on Google may have persuaded other large enterprises not to break commercial laws. That being said, an exploratory survey has been initiated on commercial practices at Amazon.<br>I would therefore like to ask the Commission the following questions:<br>(1) Will the sanctions we have seen thus f...</code> | <code>passage: 1. The purpose of the fines imposed on undertakings for infringements of Article 101 or 102 of the Treaty on the Functioning of the European Union (TFEU) is two-fold. While fines are set in each individual case at a level that has a deterrent effect to sanction the undertaking concerned (specific deterrence), they are also aimed at deterring other undertakings from engaging in, or continuing, behaviour that is contrary to Articles 101 and 102 TFEU (general deterrence). The impact of the fine imposed in a specific case therefore is of broader relevance for the market as such.<br>As regards the fines imposed on Google during the last years, the cases concerned addressed different types of conduct. While the prohibition decision in Google Shopping of June 2017 dealt with Google's abuse of dominance in the market for general search which comprised Google treating more favourably its own comparison shopping service in its search results, the prohibition decision in Google Android of J...</code> | <code>passage: Under EU competition law, the Commission investigates and sanctions possible abuses of dominant position, including by companies in the digital sector. In its two decisions of 27 June 2017 and 18 July 2018, the Commission found that Google engaged in various abuses of dominant position and imposed fines of EUR 2.42 billion and EUR 4.34 billion[1].<br>The Commission also assesses if mergers by large IT players could strengthen the market position of the merging companies and harm consumers and small businesses. The Commission has for instance carefully investigated if Facebook’s purchase of Whatsapp could lead to competition concerns in consumer communications apps or strengthen Facebook's position in online advertising.<br>In the future, the Commission will continue to enforce competition law in the digital sector based on an analysis of the evidence in each case. Competition enforcement is complementary to data protection rules.<br>The Commission also seeks to set rules to protect bus...</code> |
| <code>query: The journalist Can Dündar and Turkish support for ISIS<br>The Cumhuriyet newspaper, a prominent supporter of Turkey’s opposition party, recently published evidence proving that the Turkish Government has supplied arms to Islamic terrorist groups in Syria. Photos show vast amounts of weapons crossing the Turkish border hidden in humanitarian aid convoys, escorted by the Turkish army and police. The government has reacted by threatening the journalist behind the exposé, Can Dündar, for which it has been quite rightly condemned by the international community. I have long accused Turkey of financing and supporting international Islamic terrorism, and the secret services of numerous countries have proved me right.<br>1. What further evidence does the Commission need for it to accept that Turkey is supporting terrorist barbarities?<br>2. Does the Commission not believe that the time has come to place Turkey on the list of ‘rogue’ states and to subject it to appropriate sanctions?<br>3. Does it no...</code> | <code>passage: The High Representative does not have evidence to support press allegations that Turkey has been assisting ISIL. The Turkish authorities have repeatedly dismissed such allegations. Turkey is a key partner in the coalition against ISIL, as recognised in the Foreign Affairs Council Conclusions of 9 February 2015. It is also providing outstanding humanitarian support to more than 2.5 million Syrian and Iraqi refugees fleeing violence across the border.<br>At the same time, the situation in Syria and Iraq to which the Honourable Member refers demands closer EU/Turkey foreign policy dialogue and cooperation. The Enlargement Strategy 2014-2015[1] highlights that Turkey has made clear its readiness to play an active role in the coalition against ISIL/Da'esh. The efforts to tackle ISIL/Da'esh should be further strengthened. Therefore we are expanding and intensifying our political dialogue with Turkey on foreign and security policy. The EU continues to encourage Turkey to develop its for...</code> | <code>passage: The Commission is following with concern the situation in regards to the freedom of expression and fundamental rights in Turkey. The Commission continues to closely monitor the situation of journalists in Turkey, including Can Dündar and Erdem Gül from the opposition daily newspaper, Cumhuriyet.<br>The EU is aware and has taken note of the request from the Director-General of Reporters Without Borders for the creation of a UN mechanism for protecting journalists.<br>The Commission has repeatedly called on Turkey to ensure that existing legislation complies with European standards and is implemented in a manner which ensures proportionality and equality before the law, in line with Article 10 of the European Convention for Human Rights[1] and the case-law of the European Court of Human Rights. Turkey, as any country negotiating EU membership, needs to ensure the respect of human rights, including freedom of expression. These are non-negotiable benchmarks for Turkish accession and a c...</code> |
* Loss: [<code>CachedMultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cachedmultiplenegativesrankingloss) with these parameters:
```json
{
"scale": 20.0,
"similarity_fct": "cos_sim",
"mini_batch_size": 8,
"mini_batch_num_tokens": null,
"gather_across_devices": false,
"directions": [
"query_to_doc"
],
"partition_mode": "joint",
"hardness_mode": null,
"hardness_strength": 0.0
}
```
### Training Hyperparameters
#### Non-Default Hyperparameters
- `per_device_train_batch_size`: 32
- `num_train_epochs`: 1
- `learning_rate`: 2e-05
- `warmup_steps`: 0.1
- `fp16`: True
- `seed`: 15
- `dataloader_num_workers`: 2
- `batch_sampler`: no_duplicates
#### All Hyperparameters
<details><summary>Click to expand</summary>
- `per_device_train_batch_size`: 32
- `num_train_epochs`: 1
- `max_steps`: -1
- `learning_rate`: 2e-05
- `lr_scheduler_type`: linear
- `lr_scheduler_kwargs`: None
- `warmup_steps`: 0.1
- `optim`: adamw_torch_fused
- `optim_args`: None
- `weight_decay`: 0.0
- `adam_beta1`: 0.9
- `adam_beta2`: 0.999
- `adam_epsilon`: 1e-08
- `optim_target_modules`: None
- `gradient_accumulation_steps`: 1
- `average_tokens_across_devices`: True
- `max_grad_norm`: 1.0
- `label_smoothing_factor`: 0.0
- `bf16`: False
- `fp16`: True
- `bf16_full_eval`: False
- `fp16_full_eval`: False
- `tf32`: None
- `gradient_checkpointing`: False
- `gradient_checkpointing_kwargs`: None
- `torch_compile`: False
- `torch_compile_backend`: None
- `torch_compile_mode`: None
- `use_liger_kernel`: False
- `liger_kernel_config`: None
- `use_cache`: False
- `neftune_noise_alpha`: None
- `torch_empty_cache_steps`: None
- `auto_find_batch_size`: False
- `log_on_each_node`: True
- `logging_nan_inf_filter`: True
- `include_num_input_tokens_seen`: no
- `log_level`: passive
- `log_level_replica`: warning
- `disable_tqdm`: False
- `project`: huggingface
- `trackio_space_id`: None
- `trackio_bucket_id`: None
- `trackio_static_space_id`: None
- `per_device_eval_batch_size`: 8
- `prediction_loss_only`: True
- `eval_on_start`: False
- `eval_do_concat_batches`: True
- `eval_use_gather_object`: False
- `eval_accumulation_steps`: None
- `include_for_metrics`: []
- `batch_eval_metrics`: False
- `save_only_model`: False
- `save_on_each_node`: False
- `enable_jit_checkpoint`: False
- `push_to_hub`: False
- `hub_private_repo`: None
- `hub_model_id`: None
- `hub_strategy`: every_save
- `hub_always_push`: False
- `hub_revision`: None
- `load_best_model_at_end`: False
- `ignore_data_skip`: False
- `restore_callback_states_from_checkpoint`: False
- `full_determinism`: False
- `seed`: 15
- `data_seed`: None
- `use_cpu`: False
- `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
- `parallelism_config`: None
- `dataloader_drop_last`: False
- `dataloader_num_workers`: 2
- `dataloader_pin_memory`: True
- `dataloader_persistent_workers`: False
- `dataloader_prefetch_factor`: None
- `dataloader_multiprocessing_context`: None
- `dataloader_in_order`: True
- `remove_unused_columns`: True
- `label_names`: None
- `train_sampling_strategy`: random
- `length_column_name`: length
- `ddp_find_unused_parameters`: None
- `ddp_bucket_cap_mb`: None
- `ddp_broadcast_buffers`: False
- `ddp_static_graph`: None
- `ddp_backend`: None
- `ddp_timeout`: 1800
- `fsdp`: None
- `fsdp_config`: None
- `deepspeed`: None
- `debug`: []
- `skip_memory_metrics`: True
- `do_predict`: False
- `resume_from_checkpoint`: None
- `local_rank`: -1
- `prompts`: None
- `batch_sampler`: no_duplicates
- `multi_dataset_batch_sampler`: proportional
- `router_mapping`: {}
- `learning_rate_mapping`: {}
- `warmup_ratio`: None
</details>
### Training Logs
| Epoch | Step | Training Loss |
|:------:|:----:|:-------------:|
| 0.0268 | 50 | 2.2394 |
| 0.0537 | 100 | 0.6713 |
| 0.0805 | 150 | 0.4274 |
| 0.1074 | 200 | 0.3913 |
| 0.1342 | 250 | 0.3676 |
| 0.1610 | 300 | 0.3400 |
| 0.1879 | 350 | 0.3518 |
| 0.2147 | 400 | 0.3311 |
| 0.2415 | 450 | 0.3545 |
| 0.2684 | 500 | 0.3267 |
| 0.2952 | 550 | 0.3253 |
| 0.3221 | 600 | 0.3495 |
| 0.3489 | 650 | 0.3104 |
| 0.3757 | 700 | 0.3170 |
| 0.4026 | 750 | 0.3122 |
| 0.4294 | 800 | 0.3115 |
| 0.4563 | 850 | 0.2795 |
| 0.4831 | 900 | 0.2809 |
| 0.5099 | 950 | 0.2867 |
| 0.5368 | 1000 | 0.3012 |
| 0.5636 | 1050 | 0.2657 |
| 0.5904 | 1100 | 0.2778 |
| 0.6173 | 1150 | 0.2811 |
| 0.6441 | 1200 | 0.2919 |
| 0.6710 | 1250 | 0.2686 |
| 0.6978 | 1300 | 0.2649 |
| 0.7246 | 1350 | 0.2599 |
| 0.7515 | 1400 | 0.2616 |
| 0.7783 | 1450 | 0.2710 |
| 0.8052 | 1500 | 0.2566 |
| 0.8320 | 1550 | 0.2414 |
| 0.8588 | 1600 | 0.2438 |
| 0.8857 | 1650 | 0.2673 |
| 0.9125 | 1700 | 0.2625 |
| 0.9393 | 1750 | 0.2457 |
| 0.9662 | 1800 | 0.2556 |
| 0.9930 | 1850 | 0.2477 |
### Training Time
- **Training**: 16.0 minutes
### Framework Versions
- Python: 3.14.4
- Sentence Transformers: 6.1.0
- Transformers: 5.20.2
- PyTorch: 2.14.1+cu130
- Accelerate: 1.15.0
- Datasets: 5.1.0
- Tokenizers: 0.23.3
## Additional Resources
- [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.
- [Introduction to Matryoshka Embedding Models](https://huggingface.co/blog/matryoshka): variable-size embeddings that can be truncated with minimal quality loss.
- [Binary and Scalar Embedding Quantization for Significantly Faster & Cheaper Retrieval](https://huggingface.co/blog/embedding-quantization): post-training compression of embedding vectors.
- [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.
- [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.
## Citation
### BibTeX
#### Sentence Transformers
```bibtex
@inproceedings{reimers-2019-sentence-bert,
title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
author = "Reimers, Nils and Gurevych, Iryna",
booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
month = "11",
year = "2019",
publisher = "Association for Computational Linguistics",
url = "https://arxiv.org/abs/1908.10084",
}
```
#### CachedMultipleNegativesRankingLoss
```bibtex
@misc{gao2021scaling,
title={Scaling Deep Contrastive Learning Batch Size under Memory Limited Setup},
author={Luyu Gao and Yunyi Zhang and Jiawei Han and Jamie Callan},
year={2021},
eprint={2101.06983},
archivePrefix={arXiv},
primaryClass={cs.LG}
}
```
#### MultipleNegativesRankingLoss
```bibtex
@misc{oord2019representationlearningcontrastivepredictive,
title={Representation Learning with Contrastive Predictive Coding},
author={Aaron van den Oord and Yazhe Li and Oriol Vinyals},
year={2019},
eprint={1807.03748},
archivePrefix={arXiv},
primaryClass={cs.LG},
url={https://arxiv.org/abs/1807.03748},
}
```
<!--
## Glossary
*Clearly define terms in order to be accessible across audiences.*
-->
<!--
## Model Card Authors
*Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
-->
<!--
## Model Card Contact
*Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
--> |