143 repos across 8 sub-areas
Dense vector embeddings for semantic text search and retrieval, with a focus on multilingual and efficient embedding models. The cluster centers on transformer-based sentence embeddings (particularly BGE/BAAI models), inference infrastructure for embedding services, and standardized formats for model distribution. Repositories here support building retrieval-augmented systems, semantic similarity matching, and end-to-end embedding pipelines compatible with common inference frameworks.
Sentence Embeddings and Semantic Similarity
32 repos
Libraries and models for generating dense vector representations of text and measuring semantic similarity between sentences and documents. The cluster centers on sentence-transformers implementations across multiple languages and domains (Thai, Bengali, biomedical), alongside retrieval and dense passage ranking systems like Contriever. These tools power semantic search, information retrieval, and similarity matching tasks using transformer-based embeddings distributed via safetensors and PyTorch formats.
Text Embedding Models and Inference
31 repos
Dense vector embeddings for semantic search and text similarity, built on transformer architectures like sentence-transformers and XLM-RoBERTa. The cluster includes highly optimized embedding models (BGE, E5 families) designed for efficient retrieval and ranking tasks, along with inference infrastructure (text-embeddings-inference) to serve them at scale. Repositories here span multilingual variants, unsupervised training approaches, and production-ready endpoints compatible with standard APIs.
Text Embedding Models & Inference
31 repos
Production-ready text embedding models and their serving infrastructure, focused on semantic search and retrieval tasks. The cluster centers on standardized embedding model formats (safetensors), evaluation benchmarks (MTEB), and inference endpoints compatible with common deployment patterns. Key models like BGE, GTE, and Nomic Embed represent different size/performance tradeoffs, while the supporting infrastructure enables efficient inference at scale.
Cluster 638404
19 repos
Cluster 638410
13 repos
Cluster 638407
10 repos
Cluster 638403
5 repos
Dense and Sparse Text Embeddings
2 repos
Libraries and models for generating semantic text embeddings using both dense (neural) and sparse (lexical) retrieval methods. The cluster centers on sentence-transformers and related approaches for encoding text into fixed-dimensional representations, with significant emphasis on SPLADE (Sparse Lexical and Dense Embeddings) variants that combine sparse keyword-based and dense semantic matching. These embeddings power semantic search, retrieval-augmented generation, and recommendation systems, with tooling for model deployment via endpoints-compatible interfaces and efficient inference.