Add example of PyTorch model tensor dump with Neural Insights (#1305)

* Add example of PyTorch model tensor dump with Neural Insights

Signed-off-by: bmyrcha <bartosz.myrcha@intel.com>

* Move instruction to correct directory

Signed-off-by: bmyrcha <bartosz.myrcha@intel.com>

---------

Signed-off-by: bmyrcha <bartosz.myrcha@intel.com>

Author: bmyrcha

neural_insights/README.md

@@ -114,6 +114,9 @@ When the quantization is started, the workload should appear on the Neural Insig
 
 > Note that above example uses dummy data which is used to describe usage of Neural Insights. For diagnosis purposes you should use real dataset specific for your use case.
 
+## Tensor dump examples
+- [Step by step example how to dump weights data for PyTorch model with Neural Insights](docs/source/pytorch_nlp_cli_mode.md)
+
 ## Step by Step Diagnosis Example
 Refer to [Step by Step Diagnosis Example with TensorFlow](https://github.com/intel/neural-compressor/tree/master/neural_insights/docs/source/tf_accuracy_debug.md) and [Step by Step Diagnosis Example with ONNXRT](https://github.com/intel/neural-compressor/tree/master/neural_insights/docs/source/onnx_accuracy_debug.md) to get started with some basic quantization accuracy diagnostic skills.
 

neural_insights/docs/source/pytorch_nlp_cli_mode.md

@@ -0,0 +1,63 @@
+# Step by step example how to dump weights data for PyTorch model with Neural Insights
+1. [Introduction](#introduction)
+2. [Preparation](#preparation)
+3. [Running the quantization](#running-the-quantization)
+
+# Introduction
+In this instruction weight data will be dumped using Neural Insights. PyTorch GPT-J-6B model will be used as an example.
+
+# Preparation
+## Source
+First you need to install Intel® Neural Compressor.
+```shell
+# Install Neural Compressor
+git clone https://github.com/intel/neural-compressor.git
+cd neural-compressor 
+pip install -r requirements.txt 
+python setup.py install
+
+# Install Neural Insights
+pip install -r neural_insights/requirements.txt
+python setup.py install neural_insights
+```
+
+## Requirements
+```shell
+cd /examples/pytorch/nlp/huggingface_models/language-modeling/quantization/ptq_static/fx
+pip install -r requirements.txt
+```
+
+# Running the quantization
+Before applying quantization, modify some code in `run_clm.py` file to enable Neural Insights:
+1. Set the argument `diagnosis` to be `True` in `PostTrainingQuantConfig` so that Neural Insights will dump weights of quantizable Ops in this model.
+
+```python
+conf = PostTrainingQuantConfig(
+    accuracy_criterion=accuracy_criterion,
+    diagnosis=True,
+)
+```
+2. Quantize the model with following command:
+```shell
+python run_clm.py \
+  --model_name_or_path EleutherAI/gpt-j-6B \
+  --dataset_name wikitext\
+  --dataset_config_name wikitext-2-raw-v1 \
+  --do_train \
+  --do_eval \
+  --tune \
+  --output_dir saved_results
+```
+
+Results would be dumped into `nc_workspace` directory in similar structure:
+```
+├── history.snapshot
+├── input_model.pt
+├── inspect_saved
+│   ├── fp32
+│   │   └── inspect_result.pkl
+│   └── quan
+│       └── inspect_result.pkl
+├── model_summary.txt
+└── weights_table.csv
+```