models/labse-ig-ha-yo/README.md

metadata

tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - dense
  - generated_from_trainer
  - dataset_size:76150
  - loss:CachedMultipleNegativesRankingLoss
  - loss:MultipleNegativesRankingLoss
base_model: sentence-transformers/LaBSE
widget:
  - source_sentence: po pelu ni akoko pelu awon
    sentences:
      - About time with them too.
      - 'Us: We''re just like stars!'
      - It can get you to the next day.
  - source_sentence: Ki ló n ṣẹlẹ / Ki lo n shele?
    sentences:
      - Sure this time it's fine.
      - What's going on/happened?
      - (I've got something in my eye!
  - source_sentence: ban ga laihi gare su int mm
    sentences:
      - '"Cities have been paralyzed"'
      - I wouldn't blame them. (NM)
      - Inside, there are no paths.
  - source_sentence: '"A cikin gõnaki da marẽmari."'
    sentences:
      - How Many Days Are In A 2020?
      - 'And they would say: "Our Lord!'
      - —amid gardens and springs,
  - source_sentence: Mo ti ri pe ninu ara mi ."
    sentences:
      - Bring my Soul out of Prison.
      - I've found it within myself'."
      - I looked and couldn't believe it!
pipeline_tag: sentence-similarity
library_name: sentence-transformers

SentenceTransformer based on sentence-transformers/LaBSE

This is a sentence-transformers model finetuned from sentence-transformers/LaBSE. 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: sentence-transformers/LaBSE
• Maximum Sequence Length: 128 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• Supported Modality: Text

Model Sources

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

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': 'BertModel'})
  (1): Pooling({'embedding_dimension': 768, 'pooling_mode': 'cls', 'include_prompt': True})
  (2): Dense({'in_features': 768, 'out_features': 768, 'bias': True, 'activation_function': 'torch.nn.modules.activation.Tanh', 'module_input_name': 'sentence_embedding', 'module_output_name': 'sentence_embedding'})
  (3): Normalize({'module_input_name': 'sentence_embedding', 'module_output_name': 'sentence_embedding'})
)

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("sentence_transformers_model_id")
# Run inference
sentences = [
    'Mo ti ri pe ninu ara mi ."',
    'I\'ve found it within myself\'."',
    'Bring my Soul out of Prison.',
]
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, 0.8338, 0.0731],
#         [0.8338, 1.0000, 0.1770],
#         [0.0731, 0.1770, 1.0000]])

Training Details

Training Dataset

Unnamed Dataset

• Size: 76,150 training samples
• Columns: anchor and positive
• Approximate statistics based on the first 100 samples:

  	anchor	positive
  type	string	string
  modality	text	text
  details	min: 9 tokens mean: 36.91 tokens max: 128 tokens	min: 9 tokens mean: 35.99 tokens max: 128 tokens

• Samples:

  anchor	positive
  Ilé Ẹjọ́ Gíga Jù Lọ Nílẹ̀ Korea á lè lo ìdájọ́ tí Ilé Ẹjọ́ yìí ṣe nínú ọ̀rọ̀ ọ̀kọ̀ọ̀kan àwọn tí ẹ̀rí ọkàn wọn ò jẹ́ kí wọ́n ṣiṣẹ́ ológun.	The Constitutional Court’s decision now opens the door for the Supreme Court of Korea to apply this ruling to specific cases involving conscientious objectors. Hundreds of thousands of people were evacuated, a process that proved to be especially complicated because of government-mandated physical distancing.
  "Wanda Ya sanya muku ƙasa shimfiɗa, kuma Ya shigar muku da hanyõyi a cikinta, kuma Ya saukar da ruwa daga sama."	Who has made earth for you like a bed (spread out); and has opened roads (ways and paths etc.) for you therein; and has sent down water (rain) from the sky.
  Ìwọ ni Èlíjà bí?"+ Ó sì wí pé: "Èmi kọ́."	Are you Elijah?" and he says, "I am not."

• Loss: CachedMultipleNegativesRankingLoss with these parameters:

  {
      "scale": 30.0,
      "similarity_fct": "cos_sim",
      "mini_batch_size": 128,
      "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: 256
• num_train_epochs: 4.0
• learning_rate: 2e-05
• lr_scheduler_type: cosine
• warmup_steps: 0.1
• bf16: True
• dataloader_num_workers: 4
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• per_device_train_batch_size: 256
• num_train_epochs: 4.0
• max_steps: -1
• learning_rate: 2e-05
• lr_scheduler_type: cosine
• 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: True
• fp16: False
• 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: 42
• 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: 4
• 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

Training Logs

Epoch	Step	Training Loss
0.0671	20	0.4747
0.1342	40	0.3165
0.2013	60	0.2757
0.2685	80	0.2269
0.3356	100	0.2092
0.4027	120	0.1850
0.4698	140	0.1616
0.5369	160	0.1554
0.6040	180	0.1499
0.6711	200	0.1600
0.7383	220	0.1278
0.8054	240	0.1107
0.8725	260	0.1230
0.9396	280	0.1177
1.0067	300	0.0924
1.0738	320	0.0679
1.1409	340	0.0665
1.2081	360	0.0771
1.2752	380	0.0646
1.3423	400	0.0757
1.4094	420	0.0728
1.4765	440	0.0767
1.5436	460	0.0732
1.6107	480	0.0615
1.6779	500	0.0639
1.7450	520	0.0576
1.8121	540	0.0686
1.8792	560	0.0585
1.9463	580	0.0655
2.0134	600	0.0615
2.0805	620	0.0430
2.1477	640	0.0376
2.2148	660	0.0377
2.2819	680	0.0384
2.3490	700	0.0393
2.4161	720	0.0365
2.4832	740	0.0421
2.5503	760	0.0367
2.6174	780	0.0433
2.6846	800	0.0360
2.7517	820	0.0363
2.8188	840	0.0370
2.8859	860	0.0295
2.9530	880	0.0327
3.0201	900	0.0321
3.0872	920	0.0318
3.1544	940	0.0249
3.2215	960	0.0248
3.2886	980	0.0236
3.3557	1000	0.0249
3.4228	1020	0.0332
3.4899	1040	0.0298
3.5570	1060	0.0283
3.6242	1080	0.0261
3.6913	1100	0.0346
3.7584	1120	0.0270
3.8255	1140	0.0284
3.8926	1160	0.0321
3.9597	1180	0.0267

Training Time

• Training: 5.0 minutes

Framework Versions

• Python: 3.12.3
• Sentence Transformers: 6.1.0
• Transformers: 5.19.0
• PyTorch: 2.13.0+cu129
• Accelerate: 1.15.0
• Datasets: 5.1.0
• Tokenizers: 0.23.2

Additional Resources

• Training and Finetuning Embedding Models with Sentence Transformers: the end-to-end guide for training or finetuning Sentence Transformer models.
• Introduction to Matryoshka Embedding Models: variable-size embeddings that can be truncated with minimal quality loss.
• Binary and Scalar Embedding Quantization for Significantly Faster & Cheaper Retrieval: post-training compression of embedding vectors.
• Multimodal Embedding & Reranker Models with Sentence Transformers: use text, image, audio, and video models through the same API.
• Training and Finetuning Multimodal Embedding & Reranker Models with Sentence Transformers: train multimodal embedding models, with a Visual Document Retrieval walkthrough.

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",
}

CachedMultipleNegativesRankingLoss

@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

@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},
}