Fix(LLMLingua-2): fix the chunk max seq

microsoft / LLMLingua

To speed up LLMs' inference and enhance LLM's perceive of key information, compress the prompt and KV-Cache, which achieves up to 20x compression with minimal performance loss.

https://llmlingua.com/

MIT License

4.48k stars 251 forks source link

Fix(LLMLingua-2): fix the chunk max seq #122

Closed iofu728 closed 6 months ago

iofu728 commented 6 months ago

fix the chunk max seq.