codersan/FaLaBSE-v7

SentenceTransformer based on sentence-transformers/LaBSE

This is a sentence-transformers model finetuned from sentence-transformers/LaBSE. 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/LaBSE
• Maximum Sequence Length: 256 tokens
• Output Dimensionality: 768 dimensions
• 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': 256, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
  (2): Dense({'in_features': 768, 'out_features': 768, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh'})
  (3): Normalize()
)

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("codersan/FaLaBSE-v7")
# Run inference
sentences = [
    'مرزهای صفحه چیست؟برخی از انواع چیست؟',
    'مرزهای صفحه چیست؟',
    'اتانول چند ایزومر دارد؟',
]
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 Dataset

Unnamed Dataset

• Size: 142,964 training samples
• Columns: anchor and positive
• Approximate statistics based on the first 1000 samples:

  	anchor	positive
  type	string	string
  details	min: 6 tokens mean: 15.36 tokens max: 82 tokens	min: 6 tokens mean: 14.69 tokens max: 50 tokens

• Samples:

  anchor	positive
  گاو یونجه می خورد	گاو در حال چریدن است
  ماشینی به شکلی خطرناک از روی دختری می‌پرد.	دختر با بی‌احتیاطی روی ماشین می‌پرد.
  چگونه می توانم کارتهای هدیه iTunes رایگان را در هند دریافت کنم؟	چگونه می توانم کارتهای هدیه iTunes رایگان دریافت کنم؟

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 32
• learning_rate: 3e-05
• weight_decay: 0.15
• num_train_epochs: 4
• warmup_ratio: 0.15
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: no
• prediction_loss_only: True
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 8
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 1
• eval_accumulation_steps: None
• torch_empty_cache_steps: None
• learning_rate: 3e-05
• weight_decay: 0.15
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 4
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.15
• 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: None
• hub_always_push: False
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• include_inputs_for_metrics: False
• include_for_metrics: []
• 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
• use_liger_kernel: False
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: None
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss
0.0448	100	0.1819
0.0895	200	0.0985
0.1343	300	0.0879
0.1791	400	0.0601
0.2238	500	0.0644
0.2686	600	0.0586
0.3133	700	0.0731
0.3581	800	0.0636
0.4029	900	0.0622
0.4476	1000	0.0504
0.4924	1100	0.0603
0.5372	1200	0.0613
0.5819	1300	0.0546
0.6267	1400	0.0525
0.6714	1500	0.0606
0.7162	1600	0.0523
0.7610	1700	0.0581
0.8057	1800	0.0534
0.8505	1900	0.0531
0.8953	2000	0.0526
0.9400	2100	0.0498
0.9848	2200	0.0462
1.0295	2300	0.0555
1.0743	2400	0.0553
1.1191	2500	0.0505
1.1638	2600	0.0441
1.2086	2700	0.0365
1.2534	2800	0.0348
1.2981	2900	0.0406
1.3429	3000	0.0403
1.3876	3100	0.0409
1.4324	3200	0.0324
1.4772	3300	0.0285
1.5219	3400	0.0362
1.5667	3500	0.026
1.6115	3600	0.0271
1.6562	3700	0.0285
1.7010	3800	0.028
1.7457	3900	0.032
1.7905	4000	0.0324
1.8353	4100	0.0236
1.8800	4200	0.0267
1.9248	4300	0.0343
1.9696	4400	0.0234
2.0143	4500	0.0281
2.0591	4600	0.0272
2.1038	4700	0.0295
2.1486	4800	0.0251
2.1934	4900	0.0235
2.2381	5000	0.0219
2.2829	5100	0.0237
2.3277	5200	0.0283
2.3724	5300	0.0262
2.4172	5400	0.0218
2.4620	5500	0.0174
2.5067	5600	0.024
2.5515	5700	0.0185
2.5962	5800	0.019
2.6410	5900	0.0208
2.6858	6000	0.0188
2.7305	6100	0.0213
2.7753	6200	0.0251
2.8201	6300	0.0193
2.8648	6400	0.0175
2.9096	6500	0.0234
2.9543	6600	0.0172
2.9991	6700	0.0171
3.0439	6800	0.0215
3.0886	6900	0.0206
3.1334	7000	0.019
3.1782	7100	0.0166
3.2229	7200	0.0154
3.2677	7300	0.0178
3.3124	7400	0.0203
3.3572	7500	0.0174
3.4020	7600	0.0159
3.4467	7700	0.0149
3.4915	7800	0.0184
3.5363	7900	0.017
3.5810	8000	0.0133
3.6258	8100	0.0146
3.6705	8200	0.0148
3.7153	8300	0.0131
3.7601	8400	0.0184
3.8048	8500	0.0143
3.8496	8600	0.0137
3.8944	8700	0.0156
3.9391	8800	0.0171
3.9839	8900	0.0119

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.3.1
• Transformers: 4.47.0
• PyTorch: 2.5.1+cu121
• Accelerate: 1.2.1
• Datasets: 3.3.0
• Tokenizers: 0.21.0

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",
}

MultipleNegativesRankingLoss

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply},
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}