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-v1")
# Run inference
sentences = [
    'dating and other speed networking activities',
    ' pressure, pushbutton, snap, tumbler switches)',
    ' dializy, chemioterapia, insulinoterapia, radioterapia',
]
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: 6,413 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: 17.31 tokens max: 128 tokens	0: ~0.10% 1: ~0.30% 2: ~0.40% 3: ~0.20% 4: ~0.40% 5: ~0.10% 6: ~0.30% 8: ~0.30% 9: ~0.20% 10: ~0.10% 11: ~0.40% 12: ~0.30% 13: ~0.10% 14: ~0.10% 15: ~0.20% 16: ~0.10% 17: ~0.30% 18: ~0.20% 19: ~0.20% 20: ~0.40% 21: ~0.30% 22: ~0.10% 23: ~0.30% 24: ~0.20% 25: ~0.20% 26: ~0.10% 27: ~0.30% 28: ~0.30% 29: ~0.20% 31: ~0.10% 34: ~0.20% 36: ~0.10% 37: ~0.30% 39: ~0.20% 40: ~0.60% 41: ~0.10% 42: ~0.30% 43: ~0.20% 44: ~0.60% 45: ~0.50% 46: ~0.20% 47: ~0.10% 48: ~0.10% 49: ~0.20% 50: ~0.20% 51: ~0.20% 52: ~0.20% 53: ~0.60% 54: ~0.10% 55: ~0.20% 57: ~0.20% 58: ~0.10% 59: ~0.10% 60: ~0.20% 62: ~0.10% 63: ~0.20% 64: ~0.80% 65: ~0.60% 66: ~0.70% 67: ~0.10% 68: ~0.20% 69: ~0.30% 70: ~0.70% 72: ~0.20% 73: ~0.90% 74: ~0.40% 75: ~0.10% 76: ~0.40% 77: ~0.10% 78: ~0.30% 79: ~0.20% 81: ~0.10% 82: ~0.60% 83: ~0.20% 85: ~0.20% 87: ~0.30% 88: ~0.20% 89: ~0.10% 90: ~0.50% 95: ~0.20% 96: ~0.10% 97: ~0.40% 98: ~0.30% 99: ~0.70% 100: ~0.60% 102: ~1.00% 103: ~0.30% 104: ~0.10% 106: ~0.20% 107: ~0.10% 108: ~0.20% 109: ~0.20% 110: ~0.30% 112: ~0.10% 115: ~0.10% 116: ~0.30% 120: ~0.40% 122: ~0.20% 123: ~0.20% 124: ~0.10% 125: ~0.30% 126: ~0.50% 127: ~0.40% 128: ~0.70% 130: ~0.10% 132: ~0.10% 135: ~0.20% 136: ~0.10% 140: ~0.10% 141: ~0.10% 143: ~0.10% 145: ~0.10% 148: ~0.30% 149: ~0.20% 150: ~0.10% 151: ~0.40% 152: ~0.40% 153: ~0.20% 154: ~0.50% 158: ~0.20% 159: ~0.10% 161: ~0.10% 163: ~0.10% 164: ~0.10% 167: ~0.10% 168: ~0.20% 169: ~0.10% 171: ~0.10% 172: ~0.10% 173: ~0.10% 179: ~0.10% 181: ~0.40% 182: ~0.50% 183: ~0.20% 184: ~0.10% 185: ~0.30% 186: ~0.20% 188: ~0.10% 189: ~0.40% 190: ~0.20% 191: ~0.20% 192: ~0.60% 193: ~0.20% 194: ~0.30% 195: ~0.40% 196: ~0.10% 198: ~0.10% 199: ~0.40% 200: ~0.20% 201: ~0.20% 202: ~0.30% 206: ~0.10% 209: ~0.10% 210: ~0.10% 211: ~0.20% 212: ~0.10% 213: ~0.10% 221: ~0.20% 222: ~0.10% 224: ~0.20% 227: ~0.10% 228: ~0.10% 229: ~0.40% 231: ~0.30% 233: ~0.10% 235: ~0.10% 236: ~0.40% 237: ~0.30% 238: ~0.10% 241: ~0.20% 242: ~0.30% 243: ~0.60% 244: ~0.30% 245: ~0.10% 246: ~0.20% 247: ~0.20% 248: ~0.10% 249: ~0.10% 250: ~0.20% 254: ~0.30% 255: ~0.10% 258: ~0.10% 259: ~0.10% 260: ~0.50% 261: ~0.10% 262: ~0.20% 264: ~0.20% 265: ~0.20% 270: ~0.20% 272: ~0.10% 273: ~0.10% 274: ~0.20% 276: ~0.10% 277: ~0.30% 279: ~0.10% 280: ~0.10% 283: ~0.10% 284: ~0.10% 285: ~0.40% 286: ~0.20% 287: ~0.20% 288: ~0.10% 289: ~0.40% 291: ~0.10% 292: ~0.40% 293: ~0.40% 294: ~0.10% 295: ~0.30% 296: ~0.30% 297: ~0.20% 298: ~0.20% 300: ~0.20% 302: ~0.20% 303: ~0.30% 304: ~0.20% 308: ~0.30% 310: ~0.30% 311: ~0.50% 312: ~0.20% 313: ~0.20% 314: ~0.30% 315: ~0.10% 316: ~0.20% 317: ~0.10% 319: ~0.20% 322: ~0.10% 323: ~0.10% 324: ~0.30% 325: ~0.30% 328: ~0.20% 329: ~0.30% 330: ~0.10% 332: ~0.20% 333: ~0.30% 335: ~0.20% 336: ~0.60% 337: ~0.40% 338: ~0.10% 339: ~0.10% 340: ~0.10% 341: ~0.10% 342: ~0.10% 344: ~0.20% 346: ~0.10% 347: ~0.30% 348: ~0.10% 349: ~0.30% 350: ~0.20% 351: ~0.10% 352: ~0.40% 353: ~0.30% 354: ~0.20% 356: ~0.20% 357: ~0.40% 358: ~0.40% 359: ~0.40% 360: ~0.20% 361: ~0.40% 362: ~0.20% 363: ~0.10% 366: ~0.10% 367: ~0.10% 368: ~0.70% 369: ~0.20% 370: ~0.30% 372: ~0.30% 373: ~0.20% 374: ~0.40% 375: ~0.10% 376: ~0.10% 377: ~0.10% 379: ~0.20% 381: ~0.30% 383: ~0.40% 384: ~0.20% 385: ~0.20% 386: ~0.20% 387: ~0.20% 389: ~0.10% 390: ~0.30% 391: ~0.20% 392: ~0.20% 393: ~0.20% 395: ~0.10% 397: ~0.40% 398: ~0.20% 399: ~0.30% 400: ~0.40% 402: ~0.10% 407: ~0.10% 408: ~0.20% 409: ~0.30% 411: ~0.20% 412: ~0.20% 414: ~0.20% 415: ~0.20% 416: ~0.10% 417: ~0.30% 418: ~0.10% 420: ~0.20% 422: ~0.50% 423: ~0.10% 425: ~0.40% 426: ~0.10% 427: ~0.10% 428: ~0.40% 429: ~0.20% 430: ~0.10% 431: ~0.10% 432: ~0.10% 433: ~0.20% 434: ~0.30% 435: ~0.20% 436: ~0.40% 437: ~0.10% 438: ~0.40% 440: ~0.80% 441: ~0.20% 442: ~0.50% 443: ~0.20% 444: ~0.30% 445: ~0.30% 446: ~0.10% 447: ~0.20% 450: ~0.30% 451: ~0.20% 452: ~0.20% 453: ~0.10% 454: ~0.20% 455: ~0.30% 456: ~0.10% 457: ~0.10% 458: ~0.20% 459: ~0.20% 460: ~0.10% 461: ~0.10% 462: ~0.10% 463: ~0.40% 464: ~0.30% 467: ~0.10% 469: ~0.10% 470: ~0.10% 472: ~0.10% 475: ~0.50% 476: ~0.30% 478: ~0.10% 479: ~0.20% 480: ~0.10% 482: ~0.30% 483: ~0.50% 484: ~0.30% 485: ~0.40% 486: ~0.20% 487: ~0.20% 489: ~0.10% 490: ~0.20% 491: ~0.10% 492: ~0.40% 493: ~0.40% 495: ~0.10% 497: ~0.10% 498: ~0.10% 499: ~0.30% 501: ~0.20% 502: ~0.20% 503: ~0.10% 504: ~0.30% 505: ~0.10% 506: ~0.10% 507: ~0.10% 508: ~0.20% 509: ~0.10% 510: ~0.10% 511: ~0.10% 512: ~0.50% 514: ~0.20% 517: ~0.40% 518: ~0.10% 519: ~0.60% 520: ~0.90% 521: ~0.60% 522: ~0.10% 523: ~0.10% 524: ~0.10% 525: ~0.10% 526: ~0.10% 527: ~0.10% 528: ~0.30% 529: ~0.40% 530: ~0.60% 531: ~0.20% 532: ~0.10% 533: ~0.30% 535: ~0.50% 537: ~0.20% 540: ~0.10% 541: ~0.10% 542: ~0.20% 543: ~0.30% 544: ~0.20% 545: ~0.50% 546: ~0.30% 547: ~0.10% 548: ~0.50% 549: ~0.10% 550: ~0.30% 551: ~0.30% 552: ~0.10% 554: ~0.60% 555: ~0.20% 556: ~0.10% 557: ~0.10% 560: ~0.20% 561: ~0.10% 563: ~0.20% 564: ~0.20% 565: ~0.20% 567: ~0.10% 568: ~0.20% 570: ~0.20% 571: ~0.10% 572: ~0.20% 573: ~0.20% 576: ~0.20% 579: ~0.10%

Samples:

sentence_0	label
`retail sale of wooden, cork and wickerwork goods`	`202`
`e`	`298`
`produkcję maszyn do obróbki miękkiej gumy lub tworzyw sztucznych oraz wytwarzania wyrobów z tych materiałów: wytłaczarek, maszyn do formowania, maszyn do produkcji lub bieżnikowania opon pneumatycznych oraz pozostałych maszyn do produkcji wyrobów z gumy lub tworzyw sztucznych`	`79`

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
batch_sampler: batch_sampler
multi_dataset_batch_sampler: round_robin

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}
}

annazdr
/

nace-pl-v1