Discussions

I would love to see an offering of a compressor model like microsoft/llmlingua-2, that we could use both for prompts and RAG results