Skip to content

workaround for issue "TypeError argument 'tokens': 'NoneType' object cannot be converted to 'PyString" #4103

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Nov 6, 2025
Merged

workaround for issue "TypeError argument 'tokens': 'NoneType' object cannot be converted to 'PyString" #4103

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
a8e417c
workaround for issue TypeError argument 'tokens': 'NoneType' object c…
lvhan028 Nov 5, 2025
df8f2c9
fix
lvhan028 Nov 5, 2025
a60ab3c
new_tokens and prev_tokens should be lists
lvhan028 Nov 6, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions lmdeploy/tokenizer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -290,12 +290,18 @@ def detokenize_incrementally(self,
# This is the first iteration for this sequence
new_tokens = tokenizer.convert_ids_to_tokens(all_input_ids[ids_offset:],
skip_special_tokens=skip_special_tokens)
# `convert_ids_to_tokens` returns None for out-of-range token_id
new_tokens = new_tokens or []
new_tokens = [x for x in new_tokens if x is not None] if None in new_tokens else new_tokens
if prev_tokens is None:
# Please notice that in VLLM, indexes are detokenized one by one
# while in LMDeploy, every turn, the detokenized indexes length
# can be different.
prev_tokens = tokenizer.convert_ids_to_tokens(all_input_ids[:ids_offset],
skip_special_tokens=skip_special_tokens)
# `convert_ids_to_tokens` returns None for out-of-range token_id
prev_tokens = prev_tokens or []
prev_tokens = [x for x in prev_tokens if x is not None] if None in prev_tokens else prev_tokens
read_offset = len(prev_tokens)
if skip_special_tokens and new_tokens and new_tokens[0] in tokenizer.all_special_ids:
read_offset = read_offset + 1 # skip special token
Expand Down