Skip to content

(REPLAY) Bug fix with --no-use-tokenizer-from-checkpoint-args #3059

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
jon-barker merged 8 commits into from
Jan 24, 2026
Merged

(REPLAY) Bug fix with --no-use-tokenizer-from-checkpoint-args #3059

Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
dc41b8e
don't pull tokenizer_type from checkpoint when using --no-use-tokeniz…
Jan 22, 2026
8646830
Merge branch 'main' into jbarker/no_tokenizer_from_ckpt_fix
jon-barker Jan 22, 2026
170dab0
Merge branch 'main' into jbarker/no_tokenizer_from_ckpt_fix
jon-barker Jan 22, 2026
e2fe617
Merge branch 'main' into jbarker/no_tokenizer_from_ckpt_fix
jon-barker Jan 23, 2026
cfaa218
Merge branch 'main' into jbarker/no_tokenizer_from_ckpt_fix
jon-barker Jan 23, 2026
87bb1fe
Move set_arg in checkpointing.py
jon-barker Jan 23, 2026
80d561e
Merge branch 'main' into jbarker/no_tokenizer_from_ckpt_fix
jon-barker Jan 23, 2026
8af1a2e
Merge branch 'main' into jbarker/no_tokenizer_from_ckpt_fix
chtruong814 Jan 24, 2026
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
2 changes: 1 addition & 1 deletion megatron/training/checkpointing.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1449,10 +1449,10 @@ def _set_arg(arg_name, old_arg_name=None, force=False):
_set_arg('moe_latent_size', force=True)

# Tokenizer args.
_set_arg('tokenizer_type', force=True)
# Using checkpoint version might not always be safe (e.g., if running on different cluster).
if args.use_tokenizer_model_from_checkpoint_args:
_set_arg('tokenizer_model', force=True)
_set_arg('tokenizer_type', force=True)
_set_arg('tiktoken_pattern', force=True)
_set_arg('padded_vocab_size')

Expand Down
Loading