---
title: HTK
app_file: app1.py
sdk: gradio
sdk_version: 4.36.1
---
# MRC-RetroReader

## Introduction

MRC-RetroReader is a machine reading comprehension (MRC) model designed for reading comprehension tasks. The model leverages advanced neural network architectures to provide high accuracy in understanding and responding to textual queries.

## Table of Contents

- [Introduction](#introduction)
- [Table of Contents](#table-of-contents)
- [Installation](#installation)
- [Usage](#usage)
- [Features](#features)
- [Dependencies](#dependencies)
- [Configuration](#configuration)
- [Documentation](#documentation)
- [Examples](#examples)
- [Troubleshooting](#troubleshooting)
- [Contributors](#contributors)
- [License](#license)

## Installation

1. Clone the repository:
    ```
    git clone https://github.com/phanhoang1803/MRC-RetroReader.git
    cd MRC-RetroReader
    ```
2. Install the required dependencies:
    ```
    pip install -r requirements.txt
    ```

## Usage

- For notebooks: to running automatically, turn off wandb, warning if necessary:
```
wandb off
import warnings
warnings.filterwarnings('ignore')
``` 
- To train the model using the SQuAD v2 dataset:
```
python train_squad_v2.py --config path-to-yaml-file --module intensive --batch_size batch_size
```

## Features

- High accuracy MRC model
- Easy to train on custom datasets
- Configurable parameters for model tuning

## Dependencies

- Python 3.x
- PyTorch
- Transformers
- Tokenizers

For a full list of dependencies, see `requirements.txt`.

## Configuration

Configuration files can be found in the `configs` directory. Adjust the parameters in these files to customize the model training and evaluation.

## Documentation

For detailed documentation, refer to the `documentation` directory. This includes:
- Model architecture
- Training procedures
- Evaluation metrics

## Examples

Example training and evaluation scripts are provided in the repository. To train on the SQuAD v2 dataset:


## Troubleshooting

For common issues and their solutions, refer to the `troubleshooting guide`.

## Contributors

- phanhoang1803

## License

This project is licensed under the MIT License. See the `LICENSE` file for details.