From the model card
Fields below are copied from the tags and counters on the HuggingFace repository TencentBAC/Conan-embedding-v1 at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.
- Publisher (HF namespace)
- TencentBAC
- Library
- Sentence Transformers
- Framework tags
- PyTorch
- Weight formats
- safetensors
- License tag
cc-by-nc-4.0— read the license file in the repo before relying on it- Language tags
- Chinese (zh)
- Papers cited
- arXiv:2408.15710
- Downloads (HF counter at last fetch)
- 339,412
- Likes (HF counter at last fetch)
- 167
- Model card
- https://huggingface.co/TencentBAC/Conan-embedding-v1
Use cases
- Chinese semantic search in enterprise knowledge management systems
- Chinese document similarity for duplicate detection
- Chinese RAG systems requiring high-quality native embeddings
- Embedding Chinese user queries for intent clustering
Pros
- Chinese-specialized training gives better semantic accuracy than multilingual models on Chinese text
- Tencent scale backing ensures large and diverse training data
- Compatible with standard embedding inference pipelines
- Targets a real quality gap vs generic multilingual embedders for Chinese
Cons
- No published MTEB Chinese subset scores to compare vs BGE-M3 or GTE-Qwen2
- English and multilingual embedding quality not documented
- Tencent license terms may restrict use outside China or in competing products
- Model card lacks training data and evaluation methodology details
Tags
sentence-transformerspytorchsafetensorsbertmtebzharxiv:2408.15710license:cc-by-nc-4.0model-indexregion:us