A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[Bugifx] Fix qwen-next packed_modules_mapping (#24656)
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
Jee Jee Li committed
Sep 11, 2025 at 22:26 UTC
817beef7f3860cb52047c59578ea6eee5426ce67
1 file changed
+1
-1
vllm/model_executor/models/qwen3_next.py
+1
-1
@@ -1053,7 +1053,7 @@ class Qwen3NextForCausalLM(nn.Module, HasInnerState, SupportsLoRA, SupportsPP,
1053
"k_proj",
1054
"v_proj",
1055
],
1056
- "gate_up_proj": ["up_proj", "down_proj"],
1056
+ "gate_up_proj": ["gate_proj", "up_proj"],
1057
"in_proj": ["in_proj_qkvz", "in_proj_ba"],
1058
}
1059