Skip to content

Fix issue with async_scheduling when dealing with chunked input #359

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

Open
wants to merge 6 commits into
base: v0.10.2_next
Choose a base branch
from
Open

Fix issue with async_scheduling when dealing with chunked input #359

Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
09c183d
Fix issue with async_scheduling when dealing with chunked input
tianmu-li Oct 8, 2025
5e339b1
Dummy commit
tianmu-li Oct 8, 2025
68a272e
Clarify invalid_req_indices
tianmu-li Oct 8, 2025
d2c88fe
Merge branch 'v0.10.2_next' into async_scheduling_chunk_fix
michalkuligowski Oct 9, 2025
828401e
Merge branch 'v0.10.2_next' into async_scheduling_chunk_fix
michalkuligowski Oct 13, 2025
6530113
Merge branch 'v0.10.2_next' into async_scheduling_chunk_fix
michalkuligowski Oct 13, 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
11 changes: 6 additions & 5 deletions vllm_gaudi/v1/worker/hpu_model_runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2848,15 +2848,16 @@ def execute_model(
# If logits_indices is smaller than req_id,
# add the last token position
if logits_indices.shape[0] < len(req_id):
if structured_output:
if structured_output or self.use_async_scheduling:
logits_append = torch.tensor([torch.sum(prompt_len) - 1],
device=token_ids.device,
dtype=torch.int32)
logits_indices = torch.cat([logits_indices, logits_append])
elif self.use_async_scheduling:
# Discard partial prefill logits for async scheduling
if self.use_async_scheduling:
# Discard partial prefill logit for async scheduling
# Depends on 1 decode token/batch
invalid_req_indices.append(num_decodes + idx)
prefill_start_idx = num_decodes
invalid_req_indices.append(prefill_start_idx + idx)
htorch.core.mark_step()
non_flattened_hidden_states, aux_hidden_states, \
sample_hidden_states, logits_device = \
Expand Down Expand Up @@ -3148,7 +3149,7 @@ def execute_model(
return AsyncHPUModelRunnerOutput(
model_runner_output=model_runner_output,
sampled_token_ids=sampled_token_ids,
invalid_req_indices=[],
invalid_req_indices=invalid_req_indices,
async_output_copy_stream=self.async_output_copy_stream,
)
model_runner_output = ModelRunnerOutput(
Expand Down