Skip to content

Fix: Mamba2 generation mismatch between input_ids and inputs_embeds #32694

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
ArthurZucker merged 2 commits into from
Aug 19, 2024
Merged

Fix: Mamba2 generation mismatch between input_ids and inputs_embeds #32694

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
44821d2
fix cache when using input embeddings
vasqu Aug 14, 2024
9147953
simplify check, we can always add input ids seq len since its 0 in fi…
vasqu Aug 14, 2024
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
4 changes: 2 additions & 2 deletions src/transformers/models/mamba2/modeling_mamba2.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -964,8 +964,8 @@ def prepare_inputs_for_generation(
attention_mask: Optional[torch.Tensor] = None,
**kwargs,
):
if input_ids.shape[1] == 0:
past_len = inputs_embeds.shape[1]
if inputs_embeds is not None:
past_len = inputs_embeds.shape[1] + input_ids.shape[1]
Copy link
Collaborator

@ArthurZucker ArthurZucker Aug 16, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this is no longer the past length, but the current length 😉 Otherwise as this is used to create the cache positions, this is valid 😉

Copy link
Contributor Author

@vasqu vasqu Aug 16, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Fair enough :p

else:
past_len = input_ids.shape[1]
if use_cache:
Expand Down