capture_pre_autograd_graph -> export_for_training

docs/optimization_guide/nncf/ptq/code/ptq_torch_fx.py

@@ -22,11 +22,8 @@ def transform_fn(data_item):
 input_fp32 = ... # FP32 model input
 
 with nncf.torch.disable_patching():
-    # Using capture_pre_autograd_graph to export torch.fx.GraphModule
-    # the same way it done for PyTorch 2 Export Post Training Quantization:
-    # https://pytorch.org/tutorials/prototype/pt2e_quant_ptq.html
-    captured_graph = torch._export.capture_pre_autograd_graph(model, args=(input_fp32,))
-    quantized_model = nncf.quantize(captured_graph, calibration_dataset)
+    exported_model = torch.export.export_for_training(model, args=(input_fp32,)).module()
+    quantized_model = nncf.quantize(exported_model, calibration_dataset)
 #! [quantization]
 
 #! [inference]