fix

Signed-off-by: Pawel Gadzinski <pgadzinski@nvidia.com>

Author: pggPL

transformer_engine/debug/pytorch/debug_quantization.py

@@ -36,7 +36,7 @@
 }
 
 API_CALL_MODIFY = "modify_tensor()"
-STANDARD_QUANTIZE = "Quantize"  # Generalized: works with FP8, NVFP4, etc.
+STANDARD_QUANTIZE = "Quantize"
 HIGH_PRECISION = "High Precision"
 
 

transformer_engine/pytorch/tensor/mxfp8_tensor.py

@@ -70,7 +70,7 @@ def update_quantized(
 
     def quantize_impl(self, tensor: torch.Tensor) -> QuantizedTensor:
         """Quantize tensor implementation"""
-        return tex.quantize(tensor, self, None)
+        return tex.quantize(tensor, self)
 
     def is_quantizable(self, inp: torch.Tensor) -> bool:
         """Returns whether or not given inp can be quantized"""