abetlen / abetlen/llama-cpp-python
Load time in the log is much different from real load model time?
未關閉
- 主要語言
- Python
- 星號
- 10.6k
- 分支
- 1.4k
- PR 合併指標
- PR 指標待擷取
描述
I see that load time in the log is much smaller than real load time.
I calculate real oad model time by setting time before and after calling Llama-cpp class. From here, I saw thật, real load model time is much larger than the load time in the log.
Could you help me, what is wrong here? Thank you.
貢獻指南
評估
這個 Issue 還沒有評估資料。