xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-11 07:20:11 +08:00
Code Issues Packages Projects Releases Wiki Activity

sync with #29788

Browse Source 
Signed-off-by: Xingyu Liu <charlotteliu12x@gmail.com>
This commit is contained in:
Xingyu Liu 2025-12-23 16:02:40 -08:00
parent a55be1ffc5
commit e1b6bfa824
1 changed files with 6 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
vllm/transformers_utils/model_arch_config_convertor.py
View File

@ -74,12 +74,12 @@ class ModelArchConfigConvertorBase:
# For ChatGLM: # For ChatGLM:
"multi_query_group_num", "multi_query_group_num",
] ]
for attr in attributes: # For non-grouped-query attention models, the number of KV heads is
num_kv_heads = getattr(self.hf_text_config, attr, None) # equal to the number of attention heads.
if num_kv_heads is not None: default_factory = lambda: self.hf_text_config.num_attention_heads
return num_kv_heads return getattr_iter(
self.hf_text_config, attributes, default_factory=default_factory
return self.hf_text_config.num_attention_heads )
def get_num_experts(self) -> int: def get_num_experts(self) -> int:
"""Returns the number of experts in the model.""" """Returns the number of experts in the model."""