Fix cutlass_fp8_supported flag set on HPU (#550)

HabanaAI · Nov 26, 2024 · 38c2d10 · 38c2d10
2 parents 0f513bd + 7133502
commit 38c2d10
Showing 1 changed file with 1 addition and 0 deletions.
diff --git a/vllm/model_executor/layers/quantization/fp8.py b/vllm/model_executor/layers/quantization/fp8.py
@@ -117,6 +117,7 @@ class Fp8LinearMethod(LinearMethodBase):
 
     def __init__(self, quant_config: Fp8Config):
         self.quant_config = quant_config
+        self.cutlass_fp8_supported = False
         if current_platform.is_cuda_alike():
             self.cutlass_fp8_supported = cutlass_fp8_supported()