srikarvar/multilingual-e5-small-cogcache-contrastive

SentenceTransformer based on intfloat/multilingual-e5-small

This is a sentence-transformers model finetuned from intfloat/multilingual-e5-small. 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: intfloat/multilingual-e5-small
• Maximum Sequence Length: 512 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': 512, '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})
  (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("srikarvar/multilingual-e5-small-cogcache-contrastive")
# Run inference
sentences = [
    'What is the capital of Italy?',
    "Italy's capital city",
    'I need help with my homework',
]
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]

Evaluation

Metrics

Binary Classification

• Dataset: pair-class-dev
• Evaluated with BinaryClassificationEvaluator

Metric	Value
cosine_accuracy	1.0
cosine_accuracy_threshold	0.8237
cosine_f1	1.0
cosine_f1_threshold	0.8237
cosine_precision	1.0
cosine_recall	1.0
cosine_ap	1.0
dot_accuracy	1.0
dot_accuracy_threshold	0.8237
dot_f1	1.0
dot_f1_threshold	0.8237
dot_precision	1.0
dot_recall	1.0
dot_ap	1.0
manhattan_accuracy	0.973
manhattan_accuracy_threshold	7.9234
manhattan_f1	0.9796
manhattan_f1_threshold	9.903
manhattan_precision	0.96
manhattan_recall	1.0
manhattan_ap	0.9983
euclidean_accuracy	1.0
euclidean_accuracy_threshold	0.5938
euclidean_f1	1.0
euclidean_f1_threshold	0.5938
euclidean_precision	1.0
euclidean_recall	1.0
euclidean_ap	1.0
max_accuracy	1.0
max_accuracy_threshold	7.9234
max_f1	1.0
max_f1_threshold	9.903
max_precision	1.0
max_recall	1.0
max_ap	1.0

Binary Classification

• Dataset: pair-class-test
• Evaluated with BinaryClassificationEvaluator

Metric	Value
cosine_accuracy	1.0
cosine_accuracy_threshold	0.8053
cosine_f1	1.0
cosine_f1_threshold	0.8053
cosine_precision	1.0
cosine_recall	1.0
cosine_ap	1.0
dot_accuracy	1.0
dot_accuracy_threshold	0.8053
dot_f1	1.0
dot_f1_threshold	0.8053
dot_precision	1.0
dot_recall	1.0
dot_ap	1.0
manhattan_accuracy	1.0
manhattan_accuracy_threshold	9.7795
manhattan_f1	1.0
manhattan_f1_threshold	9.7795
manhattan_precision	1.0
manhattan_recall	1.0
manhattan_ap	1.0
euclidean_accuracy	1.0
euclidean_accuracy_threshold	0.6236
euclidean_f1	1.0
euclidean_f1_threshold	0.6236
euclidean_precision	1.0
euclidean_recall	1.0
euclidean_ap	1.0
max_accuracy	1.0
max_accuracy_threshold	9.7795
max_f1	1.0
max_f1_threshold	9.7795
max_precision	1.0
max_recall	1.0
max_ap	1.0

Training Details

Training Dataset

Unnamed Dataset

• Size: 333 training samples
• Columns: sentence1, label, and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	label	sentence2
  type	string	int	string
  details	min: 6 tokens mean: 10.25 tokens max: 20 tokens	0: ~51.65% 1: ~48.35%	min: 4 tokens mean: 9.42 tokens max: 22 tokens

• Samples:

  sentence1	label	sentence2
  How to improve my credit score?	1	Improving my credit score tips
  How does photosynthesis work?	0	What are the steps of photosynthesis?
  What is the population of Germany?	0	How many people live in Berlin?

• Loss: ContrastiveLoss with these parameters:

  {
      "distance_metric": "SiameseDistanceMetric.COSINE_DISTANCE",
      "margin": 0.5,
      "size_average": true
  }
  

Evaluation Dataset

Unnamed Dataset

• Size: 37 evaluation samples
• Columns: sentence1, label, and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	label	sentence2
  type	string	int	string
  details	min: 7 tokens mean: 10.0 tokens max: 13 tokens	0: ~35.14% 1: ~64.86%	min: 6 tokens mean: 8.68 tokens max: 12 tokens

• Samples:

  sentence1	label	sentence2
  What is the price of Bitcoin?	1	Bitcoin's current value
  Who discovered gravity?	1	Who found out about gravity?
  What is the most spoken language in the world?	1	Language spoken by the most people

• Loss: ContrastiveLoss with these parameters:

  {
      "distance_metric": "SiameseDistanceMetric.COSINE_DISTANCE",
      "margin": 0.5,
      "size_average": true
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: epoch
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• gradient_accumulation_steps: 2
• learning_rate: 3e-05
• weight_decay: 0.01
• num_train_epochs: 5
• lr_scheduler_type: reduce_lr_on_plateau
• warmup_ratio: 0.1
• load_best_model_at_end: True
• optim: adamw_torch_fused
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: epoch
• 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: 2
• eval_accumulation_steps: None
• learning_rate: 3e-05
• weight_decay: 0.01
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 5
• max_steps: -1
• lr_scheduler_type: reduce_lr_on_plateau
• 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: 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: True
• 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_fused
• 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: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	loss	pair-class-dev_max_ap	pair-class-test_max_ap
0	0	-	-	0.8544	-
0.9524	10	0.0318	0.0106	0.9935	-
1.9048	20	0.0126	-	-	-
2.0	21	-	0.0043	1.0	-
2.8571	30	0.008	-	-	-
2.9524	31	-	0.004	1.0	-
3.8095	40	0.0056	-	-	-
4.0	42	-	0.0040	1.0	-
4.7619	50	0.0039	0.0045	1.0	1.0

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.0.1
• Transformers: 4.41.2
• PyTorch: 2.1.2+cu121
• Accelerate: 0.32.1
• Datasets: 2.19.1
• 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",
}

ContrastiveLoss

@inproceedings{hadsell2006dimensionality,
    author={Hadsell, R. and Chopra, S. and LeCun, Y.},
    booktitle={2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06)}, 
    title={Dimensionality Reduction by Learning an Invariant Mapping}, 
    year={2006},
    volume={2},
    number={},
    pages={1735-1742},
    doi={10.1109/CVPR.2006.100}
}