internlm

/

internlm-chat-20b-4bit

like 24

Reproduce:

1. Download this repo's weight;
2. python3 -m lmdeploy.serve.turbomind.deploy internlm-chat-7b /path/to/internlm-chat-7b;
3. python3 -m lmdeploy.turbomind.chat ./workspace;
4. type in a very long input, reach max 2056 tokens and it warns exeed max input length?

Issue:
find turbomind/turbomind.py, to 101th row, self.session_len is always 2048?