xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-09 03:35:01 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Fix] Remove divisibility requirement between num_kv_heads and tp_size in bailing_moe (#26876)

Browse Source 
Signed-off-by: vito.yy <vito.yy@antgroup.com>
This commit is contained in:
ant-yy 2025-10-15 16:44:04 +08:00 committed by GitHub
parent 5210dc3940
commit 5c3bae1a6a
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
vllm/model_executor/models/bailing_moe.py
View File

@ -86,13 +86,12 @@ class BailingAttention(nn.Module):
tp_size = get_tensor_model_parallel_world_size()
assert self.total_num_heads % tp_size == 0
assert self.total_kv_heads % tp_size == 0
assert self.total_num_heads >= self.total_kv_heads
self.num_heads = self.total_num_heads // tp_size
self.head_dim = config.head_dim or (self.hidden_size // self.total_num_heads)
self.q_size_per_rank = self.head_dim * self.num_heads
self.num_kv_heads = self.total_kv_heads // tp_size
self.num_kv_heads = max(1, self.total_kv_heads // tp_size)
self.kv_size_per_rank = self.num_kv_heads * self.head_dim
self.scale = self.head_dim**-0.5
self.use_qk_norm = getattr(config, "use_qk_norm", False)