Skip to content

[Optimization] add warmup for _sample_from_probs #7956

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
ckl117 wants to merge 3 commits into
base: develop
Choose a base branch
from
Open

[Optimization] add warmup for _sample_from_probs #7956

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
9ac25e8
add warmup for _sample_from_probs
ckl117 May 29, 2026
fb61230
simplify code
ckl117 May 29, 2026
614775f
warmup must return next_tokens
ckl117 May 31, 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
2 changes: 2 additions & 0 deletions fastdeploy/model_executor/layers/sample/meta_data.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -67,3 +67,5 @@ class SamplingMetadata:
# Add for HPU post-processing
seq_lens_encoder: Optional[paddle.Tensor] = None
seq_lens_decoder: Optional[paddle.Tensor] = None
# Add for sampler to distinguish dummy run and profile run
is_dummy_or_profile_run: bool = False
2 changes: 2 additions & 0 deletions fastdeploy/model_executor/layers/sample/sampler.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -221,6 +221,8 @@ def _sample_from_probs(probs, sampling_metadata, top_p=None, top_k=None, topp_se
if need_top_k_sampling:
probs = dispatch_top_k_renorm_probs(probs, top_k)
next_tokens = _random_sample(probs, topp_seed=topp_seed)
if sampling_metadata.is_dummy_or_profile_run: # warmup top_p != 1.0 path
_, next_tokens = top_k_top_p_sampling(probs, top_p, top_k, top_k_list, topp_seed=topp_seed)
else:
_, next_tokens = top_k_top_p_sampling(
probs,
Expand Down
1 change: 1 addition & 0 deletions fastdeploy/worker/gpu_model_runner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1383,6 +1383,7 @@ def _prepare_inputs(self, cached_token_num=-1, cached_real_bsz=-1, is_dummy_or_p
top_p_normalized_logprobs=self.share_inputs["top_p_normalized_logprobs"],
logits_processors=self.share_inputs["logits_processors"],
share_inputs=self.share_inputs,
is_dummy_or_profile_run=is_dummy_or_profile_run,

This comment was marked as outdated.

Sign in to view

)
return token_num, token_num_event

Expand Down
Loading