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

Add padding to encoder_seq_lens #610

Merged
merged 4 commits into from
Dec 12, 2024
Merged

Add padding to encoder_seq_lens #610

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
db8f480
Add padding to encoder_seq_lens
kdamaszk Dec 10, 2024
e4142e8
format.sh
kdamaszk Dec 10, 2024
6d428f0
Merge branch 'habana_main' into dev/kdamaszke/fix-padding-on-encoder-…
kdamaszk Dec 11, 2024
c957f3b
Use real_batch_size
kdamaszk Dec 12, 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
9 changes: 9 additions & 0 deletions vllm/worker/hpu_enc_dec_model_runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -332,6 +332,15 @@ def _prepare_encoder_model_input_tensors(
attn_metadata.cross_block_groups = block_groups
attn_metadata.cross_block_usage = block_usage

# add padding to align with language model shapes
real_batch_size = len(seq_group_metadata_list)
batch_size_padded = self.bucketing_ctx.get_padded_batch_size(
real_batch_size, is_prompt)
batch_size_padding = batch_size_padded - real_batch_size
if batch_size_padding > 0:
encoder_seq_lens.extend(encoder_seq_lens[0]
for _ in range(batch_size_padding))

encoder_seq_lens_tensor = self._list_to_int32_tensor(encoder_seq_lens)
attn_metadata.encoder_seq_lens = encoder_seq_lens
attn_metadata.encoder_seq_lens_tensor = encoder_seq_lens_tensor
Expand Down
Loading