GPTQ official (#853)

## What does this PR do?
Implements official version of GPTQ with decoder level sequential
calibration flow.
Ref:
[https://github.com/IST-DASLab/FP-Quant/tree/master](https://github.com/IST-DASLab/FP-Quant/tree/master)

**Type of change:** New feature <!-- Use one of the following: Bug fix,
new feature, new example, new tests, documentation. -->

**Overview:** 
1. Deletes gptq_lite configuration. gptq() is a more generic
implementation and will resolve to gptq_lite when we set
use_sequential:False
2. Introduce GPTQHelper class to handle hessian collection
patching/unpatching, blockwise weight update, hessian initialization,
hessian_inverse computation etc

## Usage
<!-- You can potentially add a usage example below. -->

```python
# Add a code snippet demonstrating how to use this
python hf_ptq.py     --pyt_ckpt_path Qwen/Qwen3-8B     --qformat nvfp4_gptq     --kv_cache_qformat none     --dataset cnn_dailymail     --batch_size 32     --calib_seq 512     --calib_size 512    --export_path exported_model
```

## Testing
<!-- Mention how have you tested your change if applicable. -->
Measure perplexity and activation MSE on the following models

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

## Additional Information
<!-- E.g. related issue. -->


<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **New Features**
  * Added a new NVFP4 GPTQ quantization option for CLI/configuration.

* **Improvements**
* Replaced GPTQ-lite with a full GPTQ calibration pipeline for more
accurate, Hessian-aware blockwise updates and token-based sampling.
* Added sequential layer-by-layer calibration support and automatic
promotion of NVFP4 static quantizers.
  * Improved logging and timing for calibration runs.

* **Tests**
* Expanded GPTQ tests, including export/roundtrip validation for
quantized weights.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

---------

Signed-off-by: Fridah-nv <201670829+Fridah-nv@users.noreply.github.com>
Signed-off-by: Suguna Velury <178320438+sugunav14@users.noreply.github.com>
Signed-off-by: realAsma <akuriparambi@nvidia.com>
Co-authored-by: Fridah-nv <201670829+Fridah-nv@users.noreply.github.com>
Co-authored-by: realAsma <86726418+realAsma@users.noreply.github.com>

Author: 3 people

.pre-commit-config.yaml

@@ -75,6 +75,7 @@ repos:
         # Instead, we should manually add the license header to those files *after* the original header.
         exclude: >
           (?x)^(
+              modelopt/torch/quantization/utils/calib_utils.py|
               modelopt/onnx/quantization/operators.py|
               modelopt/onnx/quantization/ort_patching.py|
               modelopt/torch/_deploy/utils/onnx_utils.py|

modelopt/torch/quantization/config.py

@@ -1503,24 +1503,20 @@ class SVDQuantConfig(QuantizeAlgorithmConfig):
     )
 
 
-class GPTQLiteConfig(QuantizeAlgorithmConfig):
-    """The config for GPTQ lite.
+class GPTQCalibConfig(QuantizeAlgorithmConfig):
+    """The config for GPTQ quantization.
 
-    GPTQ lite is a variant of GPTQ that does not exactly follow the official GPTQ implementation.
-
-    GPTQ lite does not perform sequential quantization of layers. This means that the updated
-    activations are not used to process the next layer.
+    GPTQ minimizes the layer-wise quantization error by using second-order (Hessian) information
+    to perform blockwise weight updates that compensate for rounding loss. Layers are quantized
+    sequentially so that each layer's Hessian is computed from activations that already reflect
+    the quantization of preceding layers.
 
     The default values are taken from the official GPTQ implementation:
     https://github.com/IST-DASLab/FP-Quant/blob/d2e3092f968262c4de5fb050e1aef568a280dadd/src/quantization/gptq.py#L35
-
-    Note: This feature is currently experimental and may not translate to improved accuracy as expected.
-
-
     """
 
-    method: Literal["gptq_lite"] = ModeloptField("gptq_lite")
-    percdamp: float | None = ModeloptField(
+    method: Literal["gptq"] = ModeloptField("gptq")
+    perc_damp: float | None = ModeloptField(
         default=0.01,
         gt=0.0,
         le=1.0,
@@ -1533,12 +1529,6 @@ class GPTQLiteConfig(QuantizeAlgorithmConfig):
         description="""The block size for GPTQ weight update, which must be a multiple of the
         group_size used in the quantization.""",
     )
-    hessian_state_path: str | None = ModeloptField(
-        default=None,
-        title="Path to the Hessian state file.",
-        description="""The path to the Hessian state file. If hessian path exists, we load from
-         hessian file instead of recomputing them.""",
-    )
 
 
 QuantizeQuantCfgType = list[QuantizerCfgEntry]

modelopt/torch/quantization/mode.py

@@ -37,7 +37,7 @@
     AWQFullCalibConfig,
     AWQLiteCalibConfig,
     CompressConfig,
-    GPTQLiteConfig,
+    GPTQCalibConfig,
     LocalHessianCalibConfig,
     MaxCalibConfig,
     MseCalibConfig,
@@ -59,7 +59,7 @@
 )
 from .model_calib import (
     awq,
-    gptq_lite,
+    gptq,
     local_hessian_calibrate,
     max_calibrate,
     mse_calibrate,
@@ -240,8 +240,8 @@ def wrapped_calib_func(
         if sequential:
             if forward_loop is None:
                 raise ValueError("forward_loop is required for calibration but got None.")
-            assert method in ["max"], (
-                f"Sequential calibration currently only supports max calibration, got {method}"
+            assert method in ["max", "gptq"], (
+                f"Sequential calibration currently only supports max and gptq calibration, got {method}"
             )
             # Wrap with sequential processing
             sequential_calibrate(
@@ -493,12 +493,12 @@ def restore(self) -> RestoreEntrypoint:
 
 
 @CalibrateModeRegistry.register_mode
-class GPTQLiteModeDescriptor(BaseCalibrateModeDescriptor):
+class GPTQModeDescriptor(BaseCalibrateModeDescriptor):
     """Mode for GPTQ calibration algorithm."""
 
     @property
     def config_class(self) -> type[QuantizeAlgorithmConfig]:
         """Specifies the config class for the mode."""
-        return GPTQLiteConfig
+        return GPTQCalibConfig
 
-    _calib_func = gptq_lite
+    _calib_func = gptq