A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[Model] [gpt-oss] fix gpt-oss pp support (#23815)
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
Jiangyun Zhu committed
Aug 28, 2025 at 20:36 UTC
bfab219648fdd6d398c09cd022117b0e663c9e36
1 file changed
+2
-3
vllm/model_executor/models/gpt_oss.py
+2
-3
@@ -668,9 +668,8 @@ class GptOssForCausalLM(nn.Module, SupportsPP):
668
positions: torch.Tensor,
669
intermediate_tensors: Optional[IntermediateTensors] = None,
670
inputs_embeds: Optional[torch.Tensor] = None) -> torch.Tensor:
671
- assert intermediate_tensors is None
672
- assert inputs_embeds is None
673
- return self.model(input_ids, positions)
671
+ return self.model(input_ids, positions, intermediate_tensors,
672
+ inputs_embeds)
673
674
def compute_logits(self, hidden_states: torch.Tensor,
675
sampling_metadata: SamplingMetadata) -> torch.Tensor: