Merge branch 'v0.15.1-dev-fth' into 'v0.15.1-dev'

修复channel-wise 用triton报错的bug

See merge request dcutoolkit/deeplearing/vllm!443

vllm/model_executor/layers/quantization/compressed_tensors/schemes/compressed_tensors_w8a8_fp8.py

@@ -161,9 +161,8 @@ class CompressedTensorsW8A8Fp8(CompressedTensorsScheme):
             )
             if envs.VLLM_W8A8_BACKEND == 3:
                 weight = weight.t().contiguous()
-            # triton不用转置，torch需要
-            # else:
-            #     weight = weight.t()
+            else:
+                weight = weight.t()
         elif self.strategy == QuantizationStrategy.BLOCK:
             assert self.is_static_input_scheme is False
             weight, weight_scale = process_fp8_weight_block_strategy(