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

[Refactor] Consolidate SupportsEagle (#36063)

Browse Source 
Signed-off-by: Benjamin Chislett <bchislett@nvidia.com>
This commit is contained in:
Benjamin Chislett
2026-03-13 19:22:40 -04:00
committed by GitHub
parent 54a6db827f
commit 8b346309a5
24 changed files with 229 additions and 235 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
vllm/model_executor/models/mimo_v2_flash.py
View File

@@ -682,13 +682,6 @@ class MiMoV2FlashForCausalLM(nn.Module, SupportsPP, MixtureOfExperts):
self.model.make_empty_intermediate_tensors
)
def set_aux_hidden_state_layers(self, layers: tuple[int, ...]) -> None:
self.model.aux_hidden_state_layers = layers
def get_eagle3_aux_hidden_state_layers(self) -> tuple[int, ...]:
num_layers = len(self.model.layers)
return (2, num_layers // 2, num_layers - 3)
def embed_input_ids(self, input_ids: torch.Tensor) -> torch.Tensor:
return self.model.embed_input_ids(input_ids)