apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-06 09:07:10 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[FIX 2.1]fix bad_words when sending requests consecutively (#3199)

Browse Source 
* fix bad_words

* fix log

* fix log
This commit is contained in:
Sunny-bot1
2025-08-06 15:47:27 +08:00
committed by GitHub
parent bc0b92bba4
commit f672a34f95
5 changed files with 45 additions and 35 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
fastdeploy/engine/sampling_params.py
View File

@@ -218,20 +218,22 @@ class SamplingParams:
prompt_token_ids = tokenizer.encode(text=prompt, add_special_tokens=False)["input_ids"]
if len(prompt_token_ids) != 1:
logger.warning(
f"Skip bad_words: {prompt}."
f"Bad words should be a single token."
f"Got tokens: {prompt_token_ids}."
)
if not add_prefix_space:
logger.warning(
f"Skip bad_words: <{prompt}>."
f"Bad words should be a single token."
f"Got tokens: {prompt_token_ids}."
)
continue
if prompt_token_ids[0] > tokenizer.vocab_size:
logger.warning(
f"Skip bad_words: {prompt}."
f"All token id values should be satisfying:"
f" 0 <= token_id < {tokenizer.vocab_size}."
f"Got token: {prompt_token_ids}."
)
if not add_prefix_space:
logger.warning(
f"Skip bad_words: <{prompt}>."
f"All token id values should be satisfying:"
f" 0 <= token_id < {tokenizer.vocab_size}."
f"Got token: {prompt_token_ids}."
)
continue
if prompt_token_ids not in self._bad_words_token_ids: