[OMNIML-2244] Create the nvfp4 quant exporter (NVIDIA#636)

## What does this PR do?

**Type of change:**
New feature

**Overview:** 
- Implemented the NVFP4QuantExporter
- Deprecated fp4qdq_to_2dq
- Updated tests

## Usage

```python
python torch_quant_to_onnx.py --quantize_mode=nvfp4 \
	--onnx_save_path=vit_base_patch16_224.nvfp4.onnx \
	--calibration_data_size 64 \
	--batch_size 128
```

## Testing
<!-- Mention how have you tested your change if applicable. -->

```
python evaluate.py --onnx_path=vit_base_patch16_224.nvfp4.onnx \
	--model_name=vit_base_patch16_224 \
	--results_path=./results.txt \
	--batch_size 128
```

Results:
```
The top1 accuracy of the model is 84.39%
The top5 accuracy of the model is 97.312%
Inference latency of the model is 7.22412 ms
```

## Before your PR is "*Ready for review*"
<!-- If you haven't finished some of the above items you can still open
`Draft` PR. -->

- **Make sure you read and follow [Contributor
guidelines](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CONTRIBUTING.md)**
and your commits are signed.
- **Is this change backward compatible?**: No
- Deprecated fp4qdq_to_2dq
- **Did you write any new necessary tests?**: No
- **Did you add or update any necessary documentation?**: No
- **Did you update
[Changelog](https://github.com/NVIDIA/TensorRT-Model-Optimizer/blob/main/CHANGELOG.rst)?**:
No <!--- Only for new features, API changes, critical bug fixes or bw
breaking changes. -->

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

Author: ajrasane

examples/diffusers/quantization/onnx_utils/export.py

@@ -47,7 +47,7 @@
 from diffusers.models.unets import UNet2DConditionModel
 from torch.onnx import export as onnx_export
 
-from modelopt.onnx.quantization.qdq_utils import fp4qdq_to_2dq
+from modelopt.onnx.export import NVFP4QuantExporter
 from modelopt.torch.quantization.export_onnx import configure_linear_module_onnx_quantizers
 from modelopt.torch.utils import torch_to
 
@@ -547,6 +547,6 @@ def modelopt_export_sd(backbone, onnx_dir, model_name, precision):
         else:
             flux_convert_rope_weight_type(onnx_model)
     if precision == "fp4":
-        onnx_model = fp4qdq_to_2dq(onnx_model)
+        onnx_model = NVFP4QuantExporter.process_model(onnx_model)
     save_onnx(onnx_model, q_output)
     shutil.rmtree(tmp_subfolder, ignore_errors=True)

examples/onnx_ptq/llm_export.py

@@ -30,15 +30,14 @@
 from transformers import AutoConfig, AutoTokenizer
 
 import modelopt
-from modelopt.onnx.export import INT4QuantExporter
+from modelopt.onnx.export import INT4QuantExporter, NVFP4QuantExporter
 from modelopt.onnx.llm_export_utils.export_utils import (
     ModelLoader,
     WrapperModelForCausalLM,
     llm_to_onnx,
 )
 from modelopt.onnx.llm_export_utils.quantization_utils import quantize
 from modelopt.onnx.llm_export_utils.surgeon_utils import fold_fp8_qdq_to_dq
-from modelopt.onnx.quantization.qdq_utils import fp4qdq_to_2dq
 from modelopt.torch.export import export_hf_checkpoint
 from modelopt.torch.quantization.utils import is_quantized_linear
 
@@ -275,7 +274,7 @@ def time_operation(operation_name):
 
     if dtype == "nvfp4":
         with time_operation("quantizing weights to nvfp4"):
-            onnx_model = fp4qdq_to_2dq(onnx_model, verbose=True)
+            onnx_model = NVFP4QuantExporter.process_model(onnx_model)
 
     elif dtype == "int4_awq":
         with time_operation("quantizing weights to int4"):