dipanjanS/bge-base-en-v1.5-fte

SentenceTransformer based on BAAI/bge-base-en-v1.5

This is a sentence-transformers model finetuned from BAAI/bge-base-en-v1.5. 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: BAAI/bge-base-en-v1.5
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 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': 512, 'do_lower_case': True}) 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): 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("dipanjanS/bge-base-en-v1.5-fte")
# Run inference
sentences = [
    'What political roles did Rao hold in Andhra Pradesh?',
    'Balli Durga Prasad Rao (15 June 1956 – 16 September 2020) was an Indian politician. He was elected to the Lok Sabha, lower house of the Parliament of India in the 2019 Indian general election. He was a member of the YSR Congress Party. Rao was also a member of the Andhra Pradesh MLA from 1985 to 1989, 1994 to 1999, and 2009 to 2014.',
    'Ayyavazhi (, "path of the father"), is a religion with one god that started in South India in the middle of the 19th century. The \'zhi\' () in the word, \'Ayyavazhi\', is a retroflex, ri.',
]
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: 1,340 training samples
• Columns: question and context
• Approximate statistics based on the first 1000 samples:

  	question	context
  type	string	string
  details	min: 6 tokens mean: 12.39 tokens max: 24 tokens	min: 9 tokens mean: 83.99 tokens max: 510 tokens

• Samples:

  question	context
  What is Basil commonly known as?	Basil ("Ocimum basilicum") ( or ) is a plant of the Family Lamiaceae. It is also known as Sweet Basil or Tulsi. It is a tender low-growing herb that is grown as a perennial in warm, tropical climates. Basil is originally native to India and other tropical regions of Asia. It has been cultivated there for more than 5,000 years. It is prominently featured in many cuisines throughout the world. Some of them are Italian, Thai, Vietnamese and Laotian cuisines. It grows to between 30–60 cm tall. It has light green, silky leaves 3–5 cm long and 1–3 cm broad. The leaves are opposite each other. The flowers are quite big. They are white in color and arranged as a spike.
  Where is Basil originally native to?	Basil ("Ocimum basilicum") ( or ) is a plant of the Family Lamiaceae. It is also known as Sweet Basil or Tulsi. It is a tender low-growing herb that is grown as a perennial in warm, tropical climates. Basil is originally native to India and other tropical regions of Asia. It has been cultivated there for more than 5,000 years. It is prominently featured in many cuisines throughout the world. Some of them are Italian, Thai, Vietnamese and Laotian cuisines. It grows to between 30–60 cm tall. It has light green, silky leaves 3–5 cm long and 1–3 cm broad. The leaves are opposite each other. The flowers are quite big. They are white in color and arranged as a spike.
  What is the significance of the Roerich Pact?	The Roerich Pact is a treaty on Protection of Artistic and Scientific Institutions and Historic Monuments, signed by the representatives of 21 states in the Oval Office of the White House on 15 April 1935. As of January 1, 1990, the Roerich Pact had been ratified by ten nations: Brazil, Chile, Colombia, Cuba, the Dominican Republic, El Salvador, Guatemala, Mexico, the United States, and Venezuela. It went into effect on 26 August 1935. The Government of India approved the Treaty in 1948, but did not take any further formal action. The Roerich Pact is also known as "Pax Cultura" ("Cultural Peace" or "Peace through Culture"). The most important part of the Roerich Pact is the legal recognition that the protection of culture is always more important than any military necessity.

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Evaluation Dataset

Unnamed Dataset

• Size: 100 evaluation samples
• Columns: question and context
• Approximate statistics based on the first 1000 samples:

  	question	context
  type	string	string
  details	min: 7 tokens mean: 12.36 tokens max: 19 tokens	min: 12 tokens mean: 84.15 tokens max: 235 tokens

• Samples:

  question	context
  What is the demographic composition of Kolathur?	Kolathur () is a town in Salem district in the Indian state of Tamil Nadu. As of the 2001 India census, Kolathur had a population of 10,319. Males make up 53% of the population and females 47%. A total of 9% of the population is under 6 years of age.
  What is notable about India's democracy?	India is a country in Asia. It has an area of . It is at the center of South Asia. India has more than 1.2 billion (1,210,000,000) people, which is the second largest population in the world. It is the seventh largest country in the world by area and the largest country in South Asia. It is also the most populous democracy in the world.
  Who was the Chief Justice of India before Dipak Misra?	Justice Dipak Misra (born 3 October 1953) was the Judge of the Supreme Court and the Chief Justice of India. He took over as the 45th Chief Justice of India (CJI), succeeding the 44th CJI, Justice J. S. Khehar.

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• learning_rate: 3e-06
• max_steps: 332
• warmup_ratio: 0.1
• fp16: True
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• 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: 3e-06
• weight_decay: 0.0
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 3.0
• max_steps: 332
• 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: True
• 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: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	loss
0.2381	20	0.1832	0.0491
0.4762	40	0.1118	0.0246
0.7143	60	0.0991	0.0152
0.9524	80	0.0518	0.0106
1.1905	100	0.0665	0.0073
1.4286	120	0.0539	0.0058
1.6667	140	0.0548	0.0048
1.9048	160	0.0354	0.0041
2.1429	180	0.038	0.0034
2.3810	200	0.0592	0.0030
2.6190	220	0.0203	0.0027
2.8571	240	0.0441	0.0025
3.0952	260	0.023	0.0024
3.3333	280	0.0452	0.0023
3.5714	300	0.0128	0.0022
3.8095	320	0.0495	0.0022

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.0.1
• Transformers: 4.42.4
• PyTorch: 2.3.1+cu121
• Accelerate: 0.32.1
• 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",
}

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}
}