Skip to content

[None][fix] Move chunked mode transfer generation logits to handle_response #10469

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
yibinl-nvidia wants to merge 2 commits into
base: main
Choose a base branch
from
Open

[None][fix] Move chunked mode transfer generation logits to handle_response #10469

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
33ade0a
move transfer remaining logits logic to handle_response
yibinl-nvidia Jan 6, 2026
f44e555
fix test
yibinl-nvidia Jan 7, 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
5 changes: 0 additions & 5 deletions tensorrt_llm/_torch/pyexecutor/handle_logits.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -83,8 +83,3 @@ def __call__(
logits_view = logits[logits_begin:logits_end].reshape(
1, beam_width, -1)
llm_req.py_result.append_generation_logits(logits_view)

# Finalize any remaining logits transfers for all requests in chunked mode
for llm_req in chain(context_requests, generation_requests):
if llm_req.py_use_chunked_generation_logits and llm_req.py_return_generation_logits:
llm_req.py_result.transfer_remaining_device_logits()
6 changes: 6 additions & 0 deletions tensorrt_llm/_torch/pyexecutor/py_executor.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2637,6 +2637,12 @@ def _handle_responses(self):
if request.return_perf_metrics and request.py_decoding_iter >= 1:
request.update_perf_metrics(self.iter_counter)

if request.is_finished:
# Finalize any remaining logits transfers for the finished request in chunked mode
if request.py_use_chunked_generation_logits and request.py_return_generation_logits:
with torch.inference_mode():
request.py_result.transfer_remaining_device_logits()

request_done = False
if request.py_decoding_iter == 1 or request.is_finished or \
request.py_decoding_iter % self.stream_interval == 0:
Expand Down