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

Fix AttributeError in Qwen3.5 GDN layers with quantized models (#37448)

Browse Source 
Signed-off-by: Jim Smith <jim@joshua8.ai>
Signed-off-by: mgoin <mgoin64@gmail.com>
Signed-off-by: Michael Goin <mgoin64@gmail.com>
Co-authored-by: mgoin <mgoin64@gmail.com>
Co-authored-by: Xin Yang <105740670+xyang16@users.noreply.github.com>
This commit is contained in:
Jim Smith
2026-03-19 19:21:14 -04:00
committed by GitHub
parent 98ff042917
commit 4120a05ff1
2 changed files with 4 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/model_executor/models/qwen3_5.py
View File

@@ -182,8 +182,8 @@ class Qwen3_5GatedDeltaNet(Qwen3NextGatedDeltaNet):
# ============================================================
mixed_qkvz, ba = torch.ops.vllm.gdn_in_proj(
hidden_states,
self.in_proj_qkvz.weight.shape[0],
self.in_proj_ba.weight.shape[0],
sum(self.in_proj_qkvz.output_sizes) // self.tp_size,
sum(self.in_proj_ba.output_sizes) // self.tp_size,
self.prefix,
)
qkv_size = (self.key_dim * 2 + self.value_dim) // self.tp_size

4
vllm/model_executor/models/qwen3_next.py
View File

@@ -660,8 +660,8 @@ class Qwen3NextGatedDeltaNet(nn.Module, MambaBase):
# ============================================================
projected_states_qkvz, projected_states_ba = torch.ops.vllm.gdn_in_proj(
hidden_states,
self.in_proj_qkvz.weight.shape[0],
self.in_proj_ba.weight.shape[0],
sum(self.in_proj_qkvz.output_sizes) // self.tp_size,
sum(self.in_proj_ba.output_sizes) // self.tp_size,
self.prefix,
)
query, key, value, z, b, a = self.fix_query_key_value_ordering(