[Bugfix] fix bug when tp=1 (#3193)
### What this PR does / why we need it? Addresses a bug in DenseOptimRowParallelOp that occurs when tensor parallelism is not used ### Does this PR introduce _any_ user-facing change? No ### How was this patch tested? - vLLM version: v0.10.2 - vLLM main: https://github.com/vllm-project/vllm/commit/52d0cb845866869d587fc013a7c59e60a86ebcf2
This commit is contained in:
1 parent
b72e3327a6
commit
d8a9cb8458
1 file changed
+3
-1
@@ -390,7 +390,9 @@ class SequenceRowParallelOp(CustomRowParallelOp):
|
||||
bias_ = None if (self.tp_rank > 0 or self.skip_bias_add) else self.bias
|
||||
|
||||
if self.tp_size == 1 or not self.reduce_results:
|
||||
output = self.quant_method.apply(self, input_parallel, bias=bias_)
|
||||
output = self.quant_method.apply(self.layer,
|
||||
input_parallel,
|
||||
bias=bias_)
|
||||
else:
|
||||
output_parallel = self.quant_method.apply(self.layer,
|
||||
input_parallel,
|
||||
|
||||
Reference in new issue
Block a user