resolve comments

Signed-off-by: zhongboz <[email protected]>

Author: zhongbozhu

transformer_engine/common/recipe/current_scaling.cu

@@ -146,8 +146,6 @@ void compute_amax_impl(const NVTETensor input_, const NVTETensor output_, cudaSt
              to_string(output.amax.dtype), ")");
   CheckOutputTensor(output, "output_compute_amax", true);
 
-  // Optionally use config_ for future extension (e.g., for cuda graph/noop tensor)
-  // For now, config_ is unused, but the API is ready for future use.
   float *noop_ptr = nullptr;
   if (config_ != nullptr) {
     const QuantizationConfig *config_cpp = reinterpret_cast<const QuantizationConfig *>(config_);

transformer_engine/pytorch/csrc/quantizer.cpp

@@ -518,7 +518,6 @@ void Float8CurrentScalingQuantizer::quantize_impl(const TensorWrapper& input, Te
 
   // Compute amax
   if (compute_amax) {
-    // NVTE_SCOPED_GIL_RELEASE({ nvte_compute_amax(input.data(), out.data(), stream); });
     NVTE_SCOPED_GIL_RELEASE(
         { nvte_compute_amax_with_config(input.data(), out.data(), quant_config, stream); });
   }