annazdr/nace-pl-v3

SentenceTransformer based on sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2

This is a sentence-transformers model finetuned from sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2. It maps sentences & paragraphs to a 384-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-MiniLM-L12-v2
• Maximum Sequence Length: 128 tokens
• Output Dimensionality: 384 tokens
• Similarity Function: Cosine Similarity

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: BertModel 
  (1): Pooling({'word_embedding_dimension': 384, '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("annazdr/nace-pl-v3")
# Run inference
sentences = [
    ' this class also includes: laboratory doping control activities',
    'manufacture of toner cartridges',
    'telecommunication carrier equipment',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

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

Training Details

Training Dataset

Unnamed Dataset

• Size: 12,826 training samples
• Columns: sentence_0 and label
• Approximate statistics based on the first 1000 samples:

  	sentence_0	label
  type	string	int
  details	min: 2 tokens mean: 18.09 tokens max: 128 tokens	1: ~0.10% 2: ~0.30% 3: ~0.10% 4: ~0.40% 7: ~0.20% 8: ~0.10% 9: ~0.20% 10: ~0.30% 11: ~0.30% 13: ~0.30% 14: ~0.10% 18: ~0.20% 19: ~0.20% 20: ~0.10% 21: ~0.20% 22: ~0.10% 23: ~0.20% 24: ~0.30% 25: ~0.30% 26: ~0.10% 28: ~0.10% 29: ~0.10% 31: ~0.10% 32: ~0.10% 33: ~0.10% 34: ~0.20% 35: ~0.10% 37: ~0.20% 39: ~0.20% 40: ~0.40% 41: ~0.40% 42: ~0.10% 44: ~0.60% 45: ~0.10% 46: ~0.20% 48: ~0.20% 49: ~0.10% 51: ~0.10% 52: ~0.30% 54: ~0.20% 55: ~0.40% 56: ~0.10% 57: ~0.40% 58: ~0.20% 59: ~0.30% 60: ~0.20% 61: ~0.30% 62: ~0.20% 63: ~0.40% 64: ~0.60% 65: ~0.50% 66: ~1.00% 67: ~0.10% 69: ~0.50% 70: ~0.20% 71: ~0.10% 72: ~0.40% 73: ~0.60% 74: ~0.20% 75: ~0.10% 76: ~0.30% 77: ~0.10% 78: ~0.10% 80: ~0.10% 81: ~0.10% 82: ~0.80% 83: ~0.50% 84: ~0.10% 85: ~0.10% 86: ~0.30% 87: ~0.10% 88: ~0.30% 89: ~0.30% 90: ~0.60% 91: ~0.30% 92: ~0.10% 93: ~0.30% 94: ~0.20% 95: ~0.20% 96: ~0.20% 97: ~0.80% 98: ~0.20% 99: ~0.70% 100: ~0.70% 101: ~0.50% 102: ~0.60% 103: ~0.20% 104: ~0.10% 105: ~0.10% 106: ~0.20% 108: ~0.10% 109: ~0.40% 110: ~0.20% 111: ~0.30% 114: ~0.10% 115: ~0.10% 118: ~0.20% 120: ~0.10% 122: ~0.20% 123: ~0.10% 125: ~0.20% 126: ~0.30% 127: ~0.60% 128: ~0.80% 129: ~0.10% 130: ~0.10% 133: ~0.20% 135: ~0.20% 136: ~0.30% 137: ~0.20% 139: ~0.40% 140: ~0.20% 143: ~0.30% 144: ~0.10% 150: ~0.10% 151: ~0.50% 152: ~0.10% 153: ~0.10% 154: ~0.60% 155: ~0.20% 156: ~0.10% 157: ~0.10% 158: ~0.30% 161: ~0.20% 162: ~0.30% 164: ~0.10% 165: ~0.20% 168: ~0.10% 170: ~0.20% 172: ~0.10% 174: ~0.20% 175: ~0.10% 177: ~0.10% 178: ~0.10% 181: ~0.20% 182: ~0.30% 183: ~0.20% 184: ~0.40% 185: ~0.30% 186: ~0.20% 187: ~0.40% 188: ~0.10% 189: ~0.10% 190: ~0.20% 191: ~0.60% 192: ~0.10% 193: ~0.20% 197: ~0.10% 199: ~0.10% 200: ~0.30% 201: ~0.20% 202: ~0.10% 205: ~0.20% 208: ~0.10% 209: ~0.10% 210: ~0.10% 211: ~0.10% 212: ~0.10% 213: ~0.20% 215: ~0.20% 216: ~0.30% 217: ~0.20% 218: ~0.20% 219: ~0.10% 220: ~0.20% 222: ~0.10% 223: ~0.10% 225: ~0.10% 226: ~0.10% 227: ~0.20% 228: ~0.10% 229: ~0.10% 230: ~0.30% 231: ~0.20% 232: ~0.10% 233: ~0.10% 235: ~0.20% 236: ~0.20% 237: ~0.20% 241: ~0.40% 242: ~0.10% 243: ~0.30% 244: ~0.10% 245: ~0.40% 246: ~0.10% 247: ~0.30% 249: ~0.40% 253: ~0.20% 254: ~0.10% 255: ~0.10% 256: ~0.20% 257: ~0.30% 258: ~0.10% 259: ~0.10% 260: ~0.40% 264: ~0.10% 265: ~0.10% 266: ~0.10% 268: ~0.10% 270: ~0.10% 271: ~0.30% 274: ~0.10% 275: ~0.10% 276: ~0.20% 277: ~0.20% 278: ~0.30% 280: ~0.10% 281: ~0.10% 282: ~0.10% 283: ~0.10% 285: ~0.50% 286: ~0.20% 288: ~0.10% 289: ~0.20% 290: ~0.20% 291: ~0.20% 292: ~0.20% 293: ~0.20% 294: ~0.20% 296: ~0.20% 297: ~0.20% 298: ~0.10% 300: ~0.10% 301: ~0.10% 302: ~0.20% 303: ~0.30% 304: ~0.20% 306: ~0.30% 307: ~0.10% 308: ~0.20% 310: ~0.10% 311: ~0.30% 312: ~0.30% 313: ~0.20% 314: ~0.30% 315: ~0.40% 317: ~0.10% 318: ~0.10% 319: ~0.10% 320: ~0.10% 322: ~0.10% 324: ~0.60% 325: ~0.10% 327: ~0.20% 328: ~0.10% 329: ~0.10% 330: ~0.10% 331: ~0.10% 332: ~0.20% 333: ~0.10% 334: ~0.60% 335: ~0.30% 336: ~0.30% 337: ~0.30% 339: ~0.10% 341: ~0.30% 342: ~0.10% 343: ~0.10% 344: ~0.20% 345: ~0.10% 346: ~0.20% 347: ~0.20% 348: ~0.30% 349: ~0.40% 351: ~0.40% 352: ~0.10% 353: ~0.10% 354: ~0.30% 356: ~0.30% 357: ~0.10% 358: ~0.10% 359: ~0.10% 360: ~0.40% 361: ~0.30% 364: ~0.20% 366: ~0.50% 367: ~0.30% 368: ~0.40% 369: ~0.30% 370: ~0.10% 372: ~0.10% 373: ~0.30% 374: ~0.10% 375: ~0.50% 376: ~0.10% 377: ~0.40% 378: ~0.10% 379: ~0.40% 380: ~0.30% 381: ~0.30% 382: ~0.10% 383: ~0.20% 384: ~0.20% 385: ~0.40% 387: ~0.20% 390: ~0.30% 391: ~0.10% 392: ~0.50% 393: ~0.20% 394: ~0.10% 395: ~0.20% 397: ~0.10% 398: ~0.10% 399: ~0.30% 400: ~0.20% 401: ~0.10% 405: ~0.20% 407: ~0.30% 408: ~0.10% 411: ~0.20% 412: ~0.10% 413: ~0.10% 414: ~0.40% 415: ~0.30% 416: ~0.10% 418: ~0.40% 420: ~0.30% 421: ~0.40% 422: ~0.30% 423: ~0.40% 425: ~0.50% 426: ~0.10% 427: ~0.20% 428: ~0.20% 429: ~0.10% 430: ~0.20% 432: ~0.10% 433: ~0.10% 434: ~0.30% 435: ~0.20% 436: ~0.50% 437: ~0.30% 438: ~0.40% 440: ~0.40% 442: ~0.10% 444: ~0.40% 445: ~0.10% 446: ~0.20% 447: ~0.10% 448: ~0.10% 449: ~0.20% 450: ~0.20% 451: ~0.10% 452: ~0.20% 453: ~0.30% 454: ~0.20% 455: ~0.20% 456: ~0.20% 457: ~0.30% 458: ~0.10% 459: ~0.30% 460: ~0.10% 463: ~0.40% 464: ~0.20% 465: ~0.20% 468: ~0.30% 469: ~0.10% 470: ~0.10% 475: ~0.50% 476: ~0.30% 478: ~0.10% 479: ~0.20% 480: ~0.10% 482: ~0.10% 483: ~0.40% 484: ~0.60% 485: ~0.70% 487: ~0.30% 488: ~0.20% 489: ~0.10% 491: ~0.60% 492: ~0.80% 493: ~0.50% 498: ~0.20% 499: ~0.20% 500: ~0.20% 501: ~0.20% 502: ~0.10% 503: ~0.20% 504: ~0.10% 505: ~0.20% 506: ~0.10% 508: ~0.30% 511: ~0.20% 512: ~0.30% 514: ~0.10% 516: ~0.20% 517: ~0.40% 519: ~0.90% 520: ~0.80% 521: ~1.00% 522: ~0.10% 523: ~0.20% 524: ~0.20% 525: ~0.30% 528: ~0.10% 529: ~0.10% 531: ~0.10% 532: ~0.10% 533: ~0.20% 535: ~0.10% 536: ~0.10% 539: ~0.10% 540: ~0.20% 541: ~0.10% 543: ~0.30% 544: ~0.20% 545: ~0.40% 546: ~0.50% 547: ~0.20% 548: ~0.30% 549: ~0.10% 550: ~0.20% 551: ~0.70% 552: ~0.30% 554: ~0.30% 555: ~0.30% 556: ~0.10% 558: ~0.10% 559: ~0.40% 561: ~0.10% 562: ~0.10% 564: ~0.10% 565: ~0.10% 566: ~0.20% 567: ~0.20% 568: ~0.20% 570: ~0.20% 572: ~0.10% 573: ~0.10% 574: ~0.10% 578: ~0.30%

• Samples:

  sentence_0	label
  )	306
  transport of freight over sea and coastal waters	278
  	319

• Loss: BatchAllTripletLoss

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 256
• per_device_eval_batch_size: 256
• num_train_epochs: 4
• multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: no
• prediction_loss_only: True
• per_device_train_batch_size: 256
• per_device_eval_batch_size: 256
• 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
• num_train_epochs: 4
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.0
• 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
• eval_on_start: False
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: round_robin

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.0.1
• Transformers: 4.42.4
• PyTorch: 2.4.0+cu121
• Accelerate: 0.32.1
• Datasets: 2.21.0
• 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}
}