[Bug] llama2-70B NTK，输入长度只能支持到8K？
Open
- Dominant language
- Python
- Stars
- 8.1k
- Forks
- 748
- Avg merge
- 6d 2h
- Merged PRs (30d)
- 54
Description
### Checklist
- [x] 1. I have searched related issues but cannot get the expected help.
- [x] 2. The bug has not been fixed in the latest version.
### Describe the bug
我测试发现，session_len支持配置小于8192长度，配置超过8192，生成会是乱码。请问目前最大支持8192吗？
### Reproduction
CUDA_VISIBLE_DEVICES=0 python3 -m lmdeploy.serve.openai.api_server ./workspace 0.0.0.0 4070 --instance_num 32 --tp 2
### Error traceback
_No response_
Contributor guide
Assessment
This issue has not been assessed yet.