Skip to content
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

Whisper: move to tensor cpu before converting to np array at decode time #31954

Merged
merged 2 commits into from
Jul 14, 2024
Merged

Whisper: move to tensor cpu before converting to np array at decode time #31954

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
7a678ae
move to cpu before converting to np
gante Jul 14, 2024
b0d61fe
tensorflow
gante Jul 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
7 changes: 5 additions & 2 deletions src/transformers/models/whisper/tokenization_whisper.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -872,8 +872,11 @@ def _strip_prompt(self, token_ids: List[int], prompt_token_id: int, decoder_star
@staticmethod
def _convert_to_list(token_ids):
# convert type to ndarray if necessary
if "torch" in str(type(token_ids)) or "tensorflow" in str(type(token_ids)) and hasattr(token_ids, "numpy"):
token_ids = token_ids.numpy()
if hasattr(token_ids, "numpy"):
if "torch" in str(type(token_ids)):
token_ids = token_ids.cpu().numpy()
Copy link
Collaborator

Choose a reason for hiding this comment

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

Following from this - will token_ids ever have a grad? In which case, this will also fail on the cpu call

Copy link
Member Author

Choose a reason for hiding this comment

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

token_ids, the output of generate, will not have gradients :) generate is decorated with @no_grad

elif "tensorflow" in str(type(token_ids)):
token_ids = token_ids.numpy()
# now the token ids are either a numpy array, or a list of lists
if isinstance(token_ids, np.ndarray):
token_ids = token_ids.tolist()
Expand Down
7 changes: 5 additions & 2 deletions src/transformers/models/whisper/tokenization_whisper_fast.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -605,8 +605,11 @@ def _strip_prompt(self, token_ids: List[int], prompt_token_id: int, decoder_star
# Copied from transformers.models.whisper.tokenization_whisper.WhisperTokenizer._convert_to_list
def _convert_to_list(token_ids):
# convert type to ndarray if necessary
if "torch" in str(type(token_ids)) or "tensorflow" in str(type(token_ids)) and hasattr(token_ids, "numpy"):
token_ids = token_ids.numpy()
if hasattr(token_ids, "numpy"):
if "torch" in str(type(token_ids)):
token_ids = token_ids.cpu().numpy()
elif "tensorflow" in str(type(token_ids)):
token_ids = token_ids.numpy()
# now the token ids are either a numpy array, or a list of lists
if isinstance(token_ids, np.ndarray):
token_ids = token_ids.tolist()
Expand Down
Loading