apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 08:37:06 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[Feature] support seed parameter (#3161)

Browse Source 
* support seed

* fix

* add SamplingMetadata seed test

* The next_tokens values are inconsistent!

* add air and rejection seed test

* fix

* add SamplingParams seed test

* fix seed=0

* Default to defualt

* fix

* fix args_utils

* fix review

* fix review

* fix

* fix

* add xpu,gcu,iluvatar support seed

* fix
This commit is contained in:
lizexu123
2025-08-06 15:20:47 +08:00
committed by GitHub
parent 20839abccf
commit afff4d37ea
15 changed files with 386 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
fastdeploy/model_executor/layers/sample/meta_data.py
View File

@@ -43,6 +43,7 @@ class SamplingMetadata:
top_p: paddle.Tensor
top_k: Optional[paddle.Tensor] = None
min_p: Optional[paddle.Tensor] = None
seed: Optional[paddle.Tensor] = None
max_num_logprobs: Optional[int] = None
enable_early_stop: Optional[int] = False
stop_flags: Optional[paddle.Tensor] = None

4
fastdeploy/model_executor/layers/sample/sampler.py
View File

@@ -282,7 +282,9 @@ class Sampler(nn.Layer):
probs = min_p_sampling(probs, sampling_metadata.min_p)
_, next_tokens = top_k_top_p_sampling(probs, sampling_metadata.top_p, sampling_metadata.top_k)
_, next_tokens = top_k_top_p_sampling(
probs, sampling_metadata.top_p, sampling_metadata.top_k, seed=sampling_metadata.seed[0, 0]
)
logprobs_tensors = (
None if num_logprobs is None else self.gather_logprobs(raw_logprobs, num_logprobs, token_ids=next_tokens)