ayo/ipex-llm
1
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

Fix fastchat top_k (#10560)

Browse source 
* fix -1 top_k

* fix

* done
This commit is contained in:
Guancheng Fu 2024-03-27 16:01:58 +08:00 committed by GitHub
parent fc8c7904f0
commit 04baac5a2e
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 3 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
python/llm/src/ipex_llm/serving/fastchat/ipex_llm_worker.py
View file

@ -100,7 +100,9 @@ class BigDLLLMWorker(BaseModelWorker):
temperature = float(params.get("temperature", 1.0)) temperature = float(params.get("temperature", 1.0))
repetition_penalty = float(params.get("repetition_penalty", 1.0)) repetition_penalty = float(params.get("repetition_penalty", 1.0))
top_p = float(params.get("top_p", 1.0)) top_p = float(params.get("top_p", 1.0))
top_k = int(params.get("top_k", 0)) # 0 means disable top_k = int(params.get("top_k", 1))
if top_k == -1:
top_k = 1
max_new_tokens = int(params.get("max_new_tokens", 256)) max_new_tokens = int(params.get("max_new_tokens", 256))
echo = bool(params.get("echo", True)) echo = bool(params.get("echo", True))
stop_str = params.get("stop", None) stop_str = params.get("stop", None)