abhiraj1 / eval_triple_encoder

SentenceTransformer based on UKPLab/triple-encoders-dailydialog

This is a sentence-transformers model finetuned from UKPLab/triple-encoders-dailydialog . It maps sentences & paragraphs to a 1024-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: UKPLab/triple-encoders-dailydialog
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 1024 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': 1024, '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("abhiraj1/eval_triple_encoder")
# Run inference
sentences = [
    '[B1] [O] [BEFORE] s',
    '[B2] [E] [BEFORE]  ',
    '[AFTER] u',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Training Details

Training Datasets

Unnamed Dataset

• Size: 43,506 training samples
• Columns: sentence_0 , sentence_1 , sentence_2 , and label
• Approximate statistics based on the first 1000 samples:

  	sentence_0	sentence_1	sentence_2	label
  type	string	string	string	float
  details	min: 5 tokens mean: 5.86 tokens max: 6 tokens	min: 5 tokens mean: 5.84 tokens max: 6 tokens	min: 3 tokens mean: 3.81 tokens max: 4 tokens	min: 0.0 mean: 0.2 max: 1.0

• Samples:

  sentence_0	sentence_1	sentence_2	label
  [B1] [O] [BEFORE]	[B2] [E] [BEFORE]	[AFTER] u	0.0
  [B1] [E] [BEFORE] e	[B2] [O] [BEFORE] :	[AFTER] C	0.0
  [B1] [O] [BEFORE] s	[B2] [E] [BEFORE]	[AFTER] u	0.6000000000000001

• Loss: triple_encoders.losses.CosineSimilarityTripleEncoderLoss.CosineSimilarityTripleEncoderLoss

Unnamed Dataset

• Size: 942,069 training samples
• Columns: sentence_0 , sentence_1 , and label
• Approximate statistics based on the first 1000 samples:

  	sentence_0	sentence_1	label
  type	string	string	int
  details	min: 5 tokens mean: 20.26 tokens max: 182 tokens	min: 4 tokens mean: 11.94 tokens max: 31 tokens	0: ~32.40% 1: ~33.70% 2: ~33.90%

• Samples:

  sentence_0	sentence_1	label
  And the reason Lincoln and his goons had shown up? Well, not everybody was full of respect.	Lincoln didn't show up.	0
  a rally car driving down a roadway with people on the side taking pictures	People on the side of road taking picture of a rally car driving down	1
  The dog is wearing a purple cape.	THE ANIMAL IS IN A PAGEANT	2

• Loss: ContrastiveLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size : 16
• per_device_eval_batch_size : 16
• num_train_epochs : 1
• multi_dataset_batch_sampler : round_robin

All Hyperparameters

Training Logs

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.0.1
• Transformers: 4.41.2
• PyTorch: 2.3.0+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",
}

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}
}