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 17, 09:15 UTC
Remove redundant code in ColumnParallelLinear (#146)
Woosuk Kwon committed
Jun 10, 2023 at 21:25 UTC
da5ddcd544ac5ce6bc4f522af9cbdc315f94620e
1 file changed
+1
-1
cacheflow/model_executor/parallel_utils/tensor_parallel/layers.py
+1
-1
@@ -305,7 +305,7 @@ class ColumnParallelLinear(torch.nn.Module):
305
"""
306
bias = self.bias if not self.skip_bias_add else None
307
308
- input_parallel = copy_to_tensor_model_parallel_region(input_)
308
+ input_parallel = input_
309
# Matrix multiply.
310
output_parallel = F.linear(input_parallel, self.weight, bias)
311
if self.gather_output: