search

HKUDS / LightRAG

"LightRAG: Simple and Fast Retrieval-Augmented Generation"
https://arxiv.org/abs/2410.05779
MIT License
9.64k stars 1.2k forks source link

Customized rprompt on any language #165

Closed Roman12322 closed 2 weeks ago

Roman12322 commented 3 weeks ago

Would you consider adding a custom prompt from user for rag requests, like you just adding it to your LightRAG like below:

rag = LightRAG(
    working_dir=WORKING_DIR,
    llm_model_func=llm_model_func,
    rag_prompt = "SMTH_USER_WANT_HERE_TO_BE"
    embedding_func=EmbeddingFunc(
        embedding_dim=384,
        max_token_size=5000,
        func=lambda texts: hf_embedding(
            texts,
            tokenizer=AutoTokenizer.from_pretrained("sentence-transformers/all-MiniLM-L6-v2"),
            embed_model=AutoModel.from_pretrained("sentence-transformers/all-MiniLM-L6-v2")
        )
    ),
)
Roman12322 commented 3 weeks ago

I would love to see it, cuz otherwise i have no other option but bringing LightRAG repo into my project =)

antoniocirclemind commented 3 weeks ago

Hi, this is supported in this repo here. Just use the domain to customize the prompt