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 15, 22:15 UTC
[Bugfix] Fix deepseek-v2 error: "missing 1 required positional argument: 'residual'" (#13802)
Lucas Wilkinson committed
Feb 24, 2025 at 23:33 UTC
4a8cfc75516f3df2ead816c0270b63470eb8e4ee
1 file changed
+1
-1
vllm/model_executor/models/deepseek_v2.py
+1
-1
@@ -614,7 +614,7 @@ class DeepseekV2Model(nn.Module):
614
residual = intermediate_tensors["residual"]
615
616
for layer in self.layers[self.start_layer:self.end_layer]:
617
- hidden_states, residual = layer(positions, hidden_states)
617
+ hidden_states, residual = layer(positions, hidden_states, residual)
618
619
if not get_pp_group().is_last_rank:
620
return IntermediateTensors({