Anakeen/paraphrase-multilingual-mpnet-base-v2_df_meta

SentenceTransformer based on sentence-transformers/paraphrase-multilingual-mpnet-base-v2

This is a sentence-transformers model finetuned from sentence-transformers/paraphrase-multilingual-mpnet-base-v2 on the cla and def datasets. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: sentence-transformers/paraphrase-multilingual-mpnet-base-v2
• Maximum Sequence Length: 128 tokens
• Output Dimensionality: 768 tokens
• Similarity Function: Cosine Similarity
• Training Datasets:
  • cla
  • def

Model Sources

• Documentation: Sentence Transformers Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 128, 'do_lower_case': False}) with Transformer model: XLMRobertaModel 
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
)

Usage

Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("Anakeen/paraphrase-multilingual-mpnet-base-v2_df_meta")
# Run inference
sentences = [
    'Article 32 - Governing Law (BRMA 71B)\nThis Contract shall be governed by and construed in accordance with the laws of the State of California.',
    'governing law',
    'service of suit',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Training Details

Training Datasets

cla

• Dataset: cla
• Size: 4,832 training samples
• Columns: sentence, refined_name, and label
• Approximate statistics based on the first 1000 samples:

  	sentence	refined_name	label
  type	string	string	int
  details	min: 14 tokens mean: 109.03 tokens max: 128 tokens	min: 3 tokens mean: 7.38 tokens max: 26 tokens	0: ~1.50% 1: ~0.10% 2: ~0.20% 3: ~0.30% 4: ~0.30% 5: ~0.70% 6: ~0.20% 7: ~0.10% 8: ~0.10% 9: ~0.10% 10: ~0.10% 11: ~0.10% 12: ~0.20% 13: ~0.70% 14: ~0.20% 15: ~0.20% 16: ~0.80% 17: ~0.40% 18: ~0.50% 19: ~0.50% 20: ~8.50% 21: ~0.40% 22: ~0.70% 23: ~0.30% 24: ~0.30% 25: ~0.90% 26: ~0.10% 27: ~0.10% 28: ~0.50% 29: ~0.30% 30: ~0.50% 31: ~0.20% 32: ~0.40% 33: ~0.10% 34: ~0.30% 35: ~0.10% 36: ~0.10% 37: ~0.50% 38: ~0.70% 39: ~0.10% 40: ~0.20% 41: ~0.20% 42: ~0.70% 43: ~0.10% 44: ~0.60% 45: ~0.30% 46: ~0.80% 47: ~4.10% 48: ~0.30% 49: ~0.10% 50: ~0.40% 51: ~0.50% 52: ~0.50% 53: ~1.60% 54: ~0.10% 55: ~0.30% 56: ~0.20% 57: ~0.60% 58: ~0.10% 59: ~0.10% 60: ~0.20% 61: ~0.30% 62: ~0.30% 63: ~0.50% 64: ~0.20% 65: ~0.30% 66: ~0.20% 67: ~0.10% 68: ~0.30% 69: ~0.20% 70: ~0.20% 71: ~0.10% 72: ~0.10% 73: ~3.30% 74: ~0.70% 75: ~0.60% 76: ~0.20% 77: ~0.40% 78: ~0.30% 79: ~3.70% 80: ~0.50% 81: ~0.40% 82: ~0.10% 83: ~0.20% 84: ~0.50% 85: ~1.80% 86: ~0.30% 87: ~2.70% 88: ~0.30% 89: ~3.00% 90: ~0.30% 91: ~1.60% 92: ~0.10% 93: ~0.40% 94: ~0.40% 95: ~0.10% 96: ~0.60% 97: ~0.40% 98: ~0.10% 99: ~0.20% 100: ~0.10% 101: ~0.30% 102: ~0.30% 103: ~0.80% 104: ~0.10% 105: ~0.20% 106: ~0.50% 107: ~3.90% 108: ~0.10% 109: ~0.10% 110: ~0.20% 111: ~0.40% 112: ~0.20% 113: ~0.20% 114: ~0.10% 115: ~0.20% 116: ~0.10% 117: ~0.20% 118: ~0.40% 119: ~0.10% 120: ~0.10% 121: ~0.10% 122: ~0.10% 123: ~0.10% 124: ~0.20% 125: ~0.10% 126: ~0.10% 127: ~0.10% 128: ~0.30% 129: ~0.30% 130: ~1.20% 131: ~0.10% 132: ~0.20% 133: ~0.10% 134: ~0.20% 135: ~2.00% 136: ~0.30% 137: ~0.50% 138: ~0.40% 139: ~0.70% 140: ~0.10% 141: ~0.30% 142: ~1.00% 143: ~0.30% 144: ~0.60% 145: ~0.10% 146: ~0.30% 147: ~0.10% 148: ~0.10% 149: ~0.20% 150: ~0.80% 151: ~0.10% 152: ~0.30% 153: ~0.10% 154: ~0.20% 155: ~0.10% 156: ~0.10% 157: ~0.10% 158: ~0.10% 159: ~0.30% 160: ~0.20% 161: ~3.80% 162: ~0.10% 163: ~0.10% 164: ~0.10% 165: ~0.20% 166: ~0.20% 167: ~0.10% 168: ~0.20% 169: ~0.20% 170: ~0.20% 171: ~0.50% 172: ~0.10% 173: ~0.10% 174: ~0.30% 175: ~0.90% 176: ~0.80% 177: ~0.50% 178: ~0.40% 179: ~0.30% 180: ~0.30% 181: ~0.20% 182: ~0.10% 183: ~0.10% 184: ~0.10% 185: ~0.10% 186: ~0.10% 187: ~1.00% 188: ~0.10% 189: ~0.20% 190: ~0.40% 191: ~0.10% 192: ~0.20% 193: ~0.10% 194: ~0.10% 195: ~0.50% 196: ~0.10% 197: ~0.50% 198: ~0.10% 199: ~0.20% 200: ~0.20% 201: ~0.10% 202: ~0.10% 203: ~0.20% 204: ~0.10% 205: ~0.10% 206: ~0.10% 207: ~0.10% 208: ~1.10% 209: ~0.10% 210: ~0.20% 211: ~0.10% 212: ~0.10% 213: ~0.10% 214: ~0.10% 215: ~0.10% 216: ~0.30% 217: ~0.30% 218: ~0.10% 219: ~0.10% 220: ~0.10% 221: ~0.10% 222: ~0.10% 223: ~0.10% 224: ~0.10% 225: ~0.20% 226: ~0.10% 227: ~0.10% 228: ~0.10%

• Samples:

  sentence	refined_name	label
  1. This Contract does not cover any loss or liability accruing to the Reassured; directly or indirectly and whether as Insurer or Reinsurer; from any Pool of Insurers or Reinsurers formed for the purpose of covering Atomic or Nuclear Energy risks. 2. Without in any way restricting the operation of paragraph (1) of this Clause; this Contract does not cover any loss or liability accruing to the Reassured; directly or indirectly and whether as Insurer or Reinsurer; from any insurance against Physical Damage (including business interruption or consequential loss arising out of such Physical Damage) to: I. Nuclear reactor power plants including all auxiliary property on the site; or II. Any other nuclear reactor installation; including laboratories handling radioactive materials in connection with reactor installations; and “critical facilities” as such; or III. Installations for fabricating complete fuel elements or for processing substantial quantities of “special nuclear material;” and for reprocessing; salvaging; chemically separating; storing or disposing of “spent” nuclear fuel or waste materials; or IV. Installations other than those listed in paragraph (2) III above using substantial quantities of radioactive isotopes or other products of nuclear fission. 3. Without in any way restricting the operations of paragraphs (1) and (2) hereof; this Contract does not cover any loss or liability by radioactive contamination accruing to the Reassured; directly or indirectly; and whether as Insurer or Reinsurer; from any insurance on property which is on the same site as a nuclear reactor power plant or other nuclear installation and which normally would be insured therewith except that this paragraph (3) shall not operate (a) where the Reassured does not have knowledge of such nuclear reactor power plant or nuclear installation; or (b) where said insurance contains a provision excluding coverage for damage to property caused by or resulting from radioactive contamination; however caused. However on and after 1st January 1960 this sub-paragraph (b) shall only apply provided the said radioactive contamination exclusion provision has been approved by the Governmental Authority having jurisdiction thereof. 4. Without in any way restricting the operations of paragraphs (1); (2) and (3) hereof; this Contract does not cover any loss or liability by radioactive contamination accruing to the Reassured; directly or indirectly; and whether as Insurer or Reinsurer; when such radioactive contamination is a named hazard specifically insured against. 5. It is understood and agreed that this Clause shall not extend to risks using radioactive isotopes in any form where the nuclear exposure is not considered by the Reassured to be the primary hazard. 6. The term “special nuclear material” shall have the meaning given it in the Atomic Energy Act of 1954 or by any law amendatory thereof. 7. The Reassured to be sole judge of what constitutes: (a) substantial quantities; and (b) the extent of installation; plant or site. Note. Without in any way restricting the operation of paragraph (1) hereof; it is understood and agreed that (a) all policies issued by the Reassured on or before 31st December 1957 shall be free from the application of the other provisions of this Clause until expiry date or 31st December 1960 whichever first occurs whereupon all the provisions of this Clause shall apply; (b) with respect to any risk located in Canada policies issued by the Reassured on or before 31st December 1958 shall be free from the application of the other provisions of this Clause until expiry date or 31st December 1960 whichever first occurs whereupon all the provisions of this Clause shall apply.	nuclear incident exclusion physical damage reinsurance u.s.a.	0
  Downgrading clause ~ ABR1001 (Amended) Reinsurer with an S&P Rating Unless otherwise agreed by the Reinsured; the Reinsurer shall at all times during the Period of this Contract maintain an Insurer Financial Strength (IFS) rating from Standard & Poor's Rating Group of 55 Water Street; New York; NY 10041; USA ("S&P") equal to or greater than a rating of A minus as applied by S&P to that Reinsurer.	termination and downgrading	1
  Dispute Resolution ~ ABR1004 Where any dispute or difference between the parties arising out of or in connection with this Contract; including formation and validity and whether arising during or after the period of this Contract; has not been settled through negotiation; both parties agree to try in good faith to settle such dispute by non- binding mediation; before resorting to arbitration in the manner set out below.	dispute resolution	2

• Loss: BatchAllTripletLoss

def

• Dataset: def
• Size: 6,333 training samples
• Columns: sentence, refined_name, and label
• Approximate statistics based on the first 1000 samples:

  	sentence	refined_name	label
  type	string	string	int
  details	min: 3 tokens mean: 78.14 tokens max: 128 tokens	min: 3 tokens mean: 5.58 tokens max: 16 tokens	0: ~0.10% 1: ~1.40% 2: ~0.40% 3: ~3.60% 4: ~0.50% 5: ~0.50% 6: ~0.50% 7: ~0.60% 8: ~2.10% 9: ~0.30% 10: ~0.10% 11: ~0.10% 12: ~0.10% 13: ~0.10% 14: ~0.10% 15: ~0.20% 16: ~0.10% 17: ~0.10% 18: ~0.10% 19: ~1.70% 20: ~0.40% 21: ~0.70% 22: ~0.10% 23: ~0.20% 24: ~0.40% 25: ~0.30% 26: ~0.40% 27: ~0.10% 28: ~0.10% 29: ~1.30% 30: ~0.10% 31: ~0.20% 32: ~3.80% 33: ~1.90% 34: ~0.80% 35: ~0.10% 36: ~2.40% 37: ~0.10% 38: ~0.10% 39: ~1.60% 40: ~0.10% 41: ~0.10% 42: ~0.10% 43: ~0.30% 44: ~0.10% 45: ~0.10% 46: ~0.20% 47: ~0.10% 48: ~0.30% 49: ~0.10% 50: ~0.10% 51: ~0.10% 52: ~0.10% 53: ~0.10% 54: ~5.60% 55: ~0.20% 56: ~0.10% 57: ~0.10% 58: ~0.30% 59: ~0.10% 60: ~0.40% 61: ~0.50% 62: ~1.30% 63: ~1.40% 64: ~0.50% 65: ~0.10% 66: ~0.80% 67: ~0.10% 68: ~0.60% 69: ~1.10% 70: ~0.20% 71: ~0.20% 72: ~0.10% 73: ~0.20% 74: ~1.30% 75: ~0.20% 76: ~0.10% 77: ~0.10% 78: ~0.50% 79: ~0.30% 80: ~0.40% 81: ~0.20% 82: ~0.40% 83: ~0.50% 84: ~1.70% 85: ~0.50% 86: ~0.10% 87: ~0.20% 88: ~0.90% 89: ~0.60% 90: ~0.10% 91: ~0.50% 92: ~0.10% 93: ~0.20% 94: ~0.10% 95: ~0.20% 96: ~0.10% 97: ~0.10% 98: ~0.20% 99: ~0.10% 100: ~0.10% 101: ~1.10% 102: ~0.20% 103: ~0.10% 104: ~0.50% 105: ~0.10% 106: ~0.10% 107: ~0.10% 108: ~0.70% 109: ~0.50% 110: ~0.20% 111: ~0.10% 112: ~0.20% 113: ~0.20% 114: ~0.10% 115: ~0.20% 116: ~0.20% 117: ~0.30% 118: ~0.20% 119: ~0.20% 120: ~0.50% 121: ~0.20% 122: ~0.10% 123: ~0.10% 124: ~0.30% 125: ~0.10% 126: ~0.20% 127: ~0.20% 128: ~0.30% 129: ~0.40% 130: ~0.30% 131: ~0.10% 132: ~0.30% 133: ~0.10% 134: ~0.20% 135: ~0.20% 136: ~0.20% 137: ~0.10% 138: ~0.40% 139: ~0.10% 140: ~0.10% 141: ~0.20% 142: ~0.50% 143: ~0.70% 144: ~0.10% 145: ~0.10% 146: ~0.20% 147: ~0.10% 148: ~0.10% 149: ~0.20% 150: ~0.20% 151: ~0.40% 152: ~0.10% 153: ~0.20% 154: ~0.10% 155: ~0.20% 156: ~0.10% 157: ~0.10% 158: ~0.10% 159: ~0.10% 160: ~0.20% 161: ~0.10% 162: ~0.10% 163: ~0.20% 164: ~0.10% 165: ~0.10% 166: ~0.20% 167: ~0.40% 168: ~0.20% 169: ~0.20% 170: ~0.10% 171: ~0.10% 172: ~0.20% 173: ~0.10% 174: ~0.10% 175: ~0.10% 176: ~0.10% 177: ~0.10% 178: ~0.20% 179: ~0.20% 180: ~0.30% 181: ~0.20% 182: ~0.10% 183: ~1.10% 184: ~0.10% 185: ~0.30% 186: ~0.10% 187: ~0.10% 188: ~0.10% 189: ~0.10% 190: ~0.10% 191: ~0.10% 192: ~0.10% 193: ~0.10% 194: ~0.20% 195: ~0.10% 196: ~0.10% 197: ~0.10% 198: ~0.10% 199: ~0.10% 200: ~0.10% 201: ~0.30% 202: ~0.10% 203: ~1.00% 204: ~0.20% 205: ~0.10% 206: ~0.10% 207: ~0.10% 208: ~0.10% 209: ~0.10% 210: ~0.10% 211: ~0.10% 212: ~0.10% 213: ~0.10% 214: ~0.10% 215: ~0.10% 216: ~0.10% 217: ~0.10% 218: ~0.10% 219: ~0.10% 220: ~0.10% 221: ~0.10% 222: ~0.10% 223: ~0.10% 224: ~0.10% 225: ~0.10% 226: ~0.10% 227: ~0.10% 228: ~0.10% 229: ~0.10% 230: ~0.10% 231: ~0.10% 232: ~0.10% 233: ~0.10% 234: ~0.20% 235: ~0.10% 236: ~0.10% 237: ~0.10% 238: ~0.10% 239: ~0.10% 240: ~0.30% 241: ~0.20% 242: ~0.10% 243: ~0.90% 244: ~0.60% 245: ~0.10% 246: ~0.70% 247: ~0.10% 248: ~0.40% 249: ~0.20% 250: ~0.10% 251: ~0.10% 252: ~0.10% 253: ~0.10% 254: ~0.20% 255: ~0.10% 256: ~0.20% 257: ~0.10% 258: ~0.10% 259: ~0.20% 260: ~0.10% 261: ~0.10% 262: ~0.20% 263: ~0.10% 264: ~0.20% 265: ~0.20% 266: ~0.10% 267: ~0.10% 268: ~0.10% 269: ~0.10% 270: ~0.10% 271: ~0.10% 272: ~0.10% 273: ~0.20% 274: ~0.10% 275: ~0.10% 276: ~0.10% 277: ~0.10% 278: ~0.10% 279: ~0.10% 280: ~0.10% 281: ~0.10% 282: ~0.10% 283: ~0.10% 284: ~0.10% 285: ~0.10% 286: ~0.10% 287: ~0.20% 288: ~0.10% 289: ~0.10% 290: ~0.20% 291: ~0.10% 292: ~0.20% 293: ~0.20% 294: ~0.10% 295: ~0.10% 296: ~0.10% 297: ~0.10% 298: ~0.10% 299: ~0.10% 300: ~0.10% 301: ~0.10% 302: ~0.10% 303: ~0.10% 304: ~0.10% 305: ~0.10% 306: ~0.10% 307: ~0.20% 308: ~0.20% 309: ~0.10% 310: ~0.10% 311: ~0.10% 312: ~0.20% 313: ~0.10% 314: ~0.20% 315: ~0.10% 316: ~0.10% 317: ~0.10% 318: ~0.10% 319: ~0.10% 320: ~0.10% 321: ~0.10% 322: ~0.10% 323: ~0.20% 324: ~0.10% 325: ~0.10% 326: ~0.10% 327: ~0.10% 328: ~0.10% 329: ~0.10% 330: ~0.20% 331: ~0.10% 332: ~0.30% 333: ~0.10% 334: ~0.10% 335: ~0.10% 336: ~0.10% 337: ~0.10% 338: ~0.10% 339: ~0.10% 340: ~0.10% 341: ~0.10% 342: ~0.10% 343: ~0.10% 344: ~0.10% 345: ~0.10% 346: ~0.10% 347: ~0.10% 348: ~0.10% 349: ~0.10% 350: ~0.10% 351: ~0.10% 352: ~0.10% 353: ~0.10% 354: ~0.10% 355: ~0.10% 356: ~0.10% 357: ~0.10% 358: ~0.20% 359: ~0.10% 360: ~0.10% 361: ~0.10% 362: ~0.10% 363: ~0.10% 364: ~0.10% 365: ~0.10% 366: ~0.20% 367: ~0.10% 368: ~0.20% 369: ~0.20% 370: ~0.20% 371: ~0.20% 372: ~0.20% 373: ~0.10% 374: ~0.10% 375: ~0.10% 376: ~0.10% 377: ~0.10% 378: ~0.10% 379: ~0.20% 380: ~0.10% 381: ~0.10% 382: ~0.10% 383: ~0.20% 384: ~0.10% 385: ~0.10% 386: ~0.30% 387: ~0.10% 388: ~0.10% 389: ~0.20% 390: ~0.10% 391: ~0.10% 392: ~0.10%

• Samples:

  sentence	refined_name	label
  “North American CAT Perils” means certain Named Storms and Earthquake; each as defined below; in respect of that portion of losses which occur in the United States and Canada and their possessions and territories; excluding the Territory of Guam; the Territory of American Samoa; the Commonwealth of the Northern Mariana Islands; Wake Island; Johnston Atoll; Palmyra Atoll; and the State of Hawaiiterritory of Guam.	north american cat perils	0
  For the purposes of this Paragraph A.; “Named Storm” means any windstorm or windstorm system that has been named by a Reporting Agency at any time in its lifecycle and ensuing losses therefrom.	named storm	1
  For the purposes of this Paragraph A.; “Earthquake” means earthquake shake and ensuing losses therefrom.	earthquake	2

• Loss: BatchAllTripletLoss

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• num_train_epochs: 6
• warmup_ratio: 0.1

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: no
• prediction_loss_only: True
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 1
• eval_accumulation_steps: None
• learning_rate: 5e-05
• weight_decay: 0.0
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 6
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.1
• warmup_steps: 0
• log_level: passive
• log_level_replica: warning
• log_on_each_node: True
• logging_nan_inf_filter: True
• save_safetensors: True
• save_on_each_node: False
• save_only_model: False
• restore_callback_states_from_checkpoint: False
• no_cuda: False
• use_cpu: False
• use_mps_device: False
• seed: 42
• data_seed: None
• jit_mode_eval: False
• use_ipex: False
• bf16: False
• fp16: False
• fp16_opt_level: O1
• half_precision_backend: auto
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: None
• local_rank: 0
• ddp_backend: None
• tpu_num_cores: None
• tpu_metrics_debug: False
• debug: []
• dataloader_drop_last: False
• dataloader_num_workers: 0
• dataloader_prefetch_factor: None
• past_index: -1
• disable_tqdm: False
• remove_unused_columns: True
• label_names: None
• load_best_model_at_end: False
• ignore_data_skip: False
• fsdp: []
• fsdp_min_num_params: 0
• fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
• fsdp_transformer_layer_cls_to_wrap: None
• accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
• deepspeed: None
• label_smoothing_factor: 0.0
• optim: adamw_torch
• optim_args: None
• adafactor: False
• group_by_length: False
• length_column_name: length
• ddp_find_unused_parameters: None
• ddp_bucket_cap_mb: None
• ddp_broadcast_buffers: False
• dataloader_pin_memory: True
• dataloader_persistent_workers: False
• skip_memory_metrics: True
• use_legacy_prediction_loop: False
• push_to_hub: False
• resume_from_checkpoint: None
• hub_model_id: None
• hub_strategy: every_save
• hub_private_repo: False
• hub_always_push: False
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• include_inputs_for_metrics: False
• eval_do_concat_batches: True
• fp16_backend: auto
• push_to_hub_model_id: None
• push_to_hub_organization: None
• mp_parameters:
• auto_find_batch_size: False
• full_determinism: False
• torchdynamo: None
• ray_scope: last
• ddp_timeout: 1800
• torch_compile: False
• torch_compile_backend: None
• torch_compile_mode: None
• dispatch_batches: None
• split_batches: None
• include_tokens_per_second: False
• include_num_input_tokens_seen: False
• neftune_noise_alpha: None
• optim_target_modules: None
• batch_eval_metrics: False
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss
0.7163	500	1.2076
1.4327	1000	1.3144
2.1490	1500	1.1513
2.8653	2000	0.8245
3.5817	2500	0.6458
4.2980	3000	0.4437
5.0143	3500	0.2403
5.7307	4000	0.1507

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.0.1
• Transformers: 4.41.2
• PyTorch: 2.3.0+cu121
• Accelerate: 0.31.0
• Datasets: 2.20.0
• Tokenizers: 0.19.1

Citation

BibTeX

Sentence Transformers

@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",
}

BatchAllTripletLoss

@misc{hermans2017defense,
    title={In Defense of the Triplet Loss for Person Re-Identification}, 
    author={Alexander Hermans and Lucas Beyer and Bastian Leibe},
    year={2017},
    eprint={1703.07737},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}