Chranos/enginex-mlu370-vllm
1
0
Fork 0
forked from EngineX-Cambricon/enginex-mlu370-vllm
Code Pull Requests Activity

add deepseekv3 and llama4

Browse Source
This commit is contained in:
Chranos
2026-02-11 15:09:59 +08:00
parent 44ffd2094a
commit eaeb5169e0
1 changed files with 4 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
vllm-v0.6.2/vllm/model_executor/model_loader/utils.py
View File

@@ -21,13 +21,10 @@ def set_default_torch_dtype(dtype: torch.dtype):
def get_model_architecture(
model_config: ModelConfig) -> Tuple[Type[nn.Module], str]:
architectures = getattr(model_config.hf_config, "architectures", None) or []
logger.warning("[DEBUG-ARCH] get_model_architecture: "
"type(hf_config)=%s, architectures=%s, "
"id(hf_config)=%s, has_text_config=%s",
type(model_config.hf_config).__name__,
getattr(model_config.hf_config, "architectures", "MISSING"),
id(model_config.hf_config),
hasattr(model_config.hf_config, "text_config"))
print(f"[DEBUG-ARCH] get_model_architecture: "
f"type(hf_config)={type(model_config.hf_config).__name__}, "
f"architectures={getattr(model_config.hf_config, 'architectures', 'MISSING')}, "
f"id(hf_config)={id(model_config.hf_config)}")
# Special handling for quantized Mixtral.
# FIXME(woosuk): This is a temporary hack.
mixtral_supported = [