mirror of
https://github.com/vllm-project/vllm.git
synced 2025-10-20 14:53:52 +08:00
[Bugfix][Hardware][CPU] Fix CPU model input for decode (#9044)
This commit is contained in:
@ -133,7 +133,7 @@ class ModelInputForCPUBuilder(ModelRunnerInputBuilderBase[ModelInputForCPU]):
|
||||
(input_tokens, input_positions,
|
||||
attn_metadata) = self._prepare_decode(
|
||||
self.seq_group_metadata_list)
|
||||
seq_lens = []
|
||||
seq_lens = None
|
||||
|
||||
return self.model_input_cls(
|
||||
input_tokens=input_tokens,
|
||||
|
Reference in New Issue
Block a user