LexSemBridge: Fine-Grained Dense Representation Enhancement through Token-Aware Embedding Augmentation

This model implements LexSemBridge, a unified framework that enhances dense query representations through fine-grained, input-aware vector modulation. LexSemBridge constructs latent enhancement vectors from input tokens using statistical, learned, and contextual paradigms, integrating them with dense embeddings via element-wise interaction. It naturally extends to both text and vision modalities with an appropriate tokenization, aiming to improve performance on fine-grained retrieval tasks where precise keyword alignment and span-level localization are crucial.

The model is based on the paper LexSemBridge: Fine-Grained Dense Representation Enhancement through Token-Aware Embedding Augmentation.

For the official code and further details, please refer to the GitHub repository.

Model Details

Model Description

• Model Type: Sentence Transformer
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 1024 tokens
• Similarity Function: Cosine Similarity

Model Sources

• Paper: LexSemBridge: Fine-Grained Dense Representation Enhancement through Token-Aware Embedding Augmentation
• Code/GitHub Repository: https://github.com/Jasaxion/LexSemBridge/
• Documentation: Sentence Transformers Documentation
• Repository (Sentence Transformers Library): Sentence Transformers on GitHub
• Hugging Face (Sentence Transformers Models): 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': 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("Jasaxion/LexSemBridge_CLR_snowflake") # Example: LexSemBridge-CLR-snowflake model
# Run inference
sentences = [
    'The weather is lovely today.',
    "It's so sunny outside!",
    'He drove to the stadium.',
]
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

Framework Versions

• Python: 3.10.14
• Sentence Transformers: 3.1.0.dev0
• Transformers: 4.44.2
• PyTorch: 2.4.1+cu121
• Accelerate: 0.34.2
• Datasets: 2.21.0
• Tokenizers: 0.19.1

Citation

BibTeX

@article{zhan2025lexsembridge,
  title={LexSemBridge: Fine-Grained Dense Representation Enhancement through Token-Aware Embedding Augmentation},
  author={Zhan, Shaoxiong and Lin, Hai and Tan, Hongming and Cai, Xiaodong and Zheng, Hai-Tao and Su, Xin and Shan, Zifei and Liu, Ruitong and Kim, Hong-Gee},
  journal={arXiv preprint arXiv:2508.17858},
  year={2025}
}