biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Model] Cleanup: Remove redundant manual definition of make_empty_intermediate_tensors in GLM-4-MoE (#31869)

Browse Source 
Signed-off-by: maang <maang_h@163.com>
This commit is contained in:
maang
2026-01-07 16:18:28 +08:00
committed by GitHub
parent 59fe6f298e
commit 1f33e38e81
1 changed files with 0 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
vllm/model_executor/models/glm4_moe.py
View File

@@ -478,20 +478,6 @@ class Glm4MoeModel(nn.Module):
hidden_states, _ = self.norm(hidden_states, residual)
return hidden_states
def make_empty_intermediate_tensors(
self, batch_size: int, dtype: torch.dtype, device: torch.device
) -> IntermediateTensors:
return IntermediateTensors(
{
"hidden_states": torch.zeros(
(batch_size, self.config.hidden_size), dtype=dtype, device=device
),
"residual": torch.zeros(
(batch_size, self.config.hidden_size), dtype=dtype, device=device
),
}
)
def get_expert_mapping(self) -> list[tuple[str, str, int, str]]:
# Params for weights, fp8 weight scales, fp8 activation scales
# (param_name, weight_name, expert_id, shard_id)