FinetuneRT-Colab

# **RT Finetuning Scripts**

*⚠️Clear Notebook Before Use*

This repository contains the training and fine-tuning scripts for the following models and adapters:

- **Llama**
- **Qwen**
- **SmolLM**
- **DeepSeek**
- **Other Adapters**

## Overview

These scripts are designed to help you fine-tune various language models and adapters, making it easy to train or adapt models to new datasets and tasks. Whether you want to improve a model’s performance or specialize it for a specific domain, these scripts will facilitate the process.

## Features

- **Training Scripts**: Easily train models on your own dataset.
- **Fine-Tuning Scripts**: Fine-tune pre-trained models with minimal setup.
- **Support for Multiple Models**: The scripts support a variety of models including Llama, Qwen, SmolLM, and DeepSeek.
- **Adapter Support**: Fine-tune adapters for flexible deployment and specialization.

## Requirements

Before running the scripts, make sure you have the following dependencies:

- Python 3.x
- `transformers` library
- `torch` (CUDA for GPU acceleration)
- Additional dependencies (see `requirements.txt`)

## Installation

Clone the repository and install dependencies:

```bash
git clone https://github.com/your-repo/rt-finetuning-scripts.git
cd rt-finetuning-scripts
pip install -r requirements.txt
```

## Usage

### Fine-Tuning a Model

1. **Choose a model**: Select from Llama, Qwen, SmolLM, or DeepSeek.
2. **Prepare your dataset**: Ensure your dataset is formatted correctly for fine-tuning.
3. **Run the fine-tuning script**: Execute the script for your chosen model.

## Contributing

Contributions are welcome! If you have improvements or bug fixes, feel free to submit a pull request.

# **RT微调脚本集**

⚠️使用前请清空Notebook

本仓库包含适用于以下模型与适配器的训练与微调脚本：
- **Llama**
- **Qwen**
- **SmolLM**
- **DeepSeek**
- **其他适配器**

## 项目概述

本系列脚本旨在协助您对各类语言模型与适配器进行微调，可便捷地将模型训练或适配至全新数据集与任务场景。无论您希望提升模型的综合性能，还是将其针对特定领域进行专业化定制，本脚本均可简化并推进整个流程。

## 核心特性

- **训练脚本**：支持基于自定义数据集快速完成模型训练。
- **微调脚本**：仅需极简配置即可对预训练模型进行微调。
- **多模型支持**：本脚本兼容多款主流模型，涵盖Llama、Qwen、SmolLM与DeepSeek。
- **适配器支持**：支持适配器微调，便于灵活部署与模型专业化定制。

## 环境依赖

运行脚本前，请确保已安装以下依赖：
- Python 3.x
- `transformers` 库
- `torch`（如需GPU加速需安装CUDA版本）
- 其他依赖项（详见`requirements.txt`文件）

## 安装步骤

克隆本仓库并安装依赖：
bash
git clone https://github.com/your-repo/rt-finetuning-scripts.git
cd rt-finetuning-scripts
pip install -r requirements.txt

## 使用方法

### 模型微调流程

1. **选择模型**：从Llama、Qwen、SmolLM或DeepSeek中选取目标模型。
2. **准备数据集**：确保数据集格式符合微调任务要求。
3. **运行微调脚本**：执行对应目标模型的微调脚本。

## 贡献指南

欢迎社区贡献！若您有功能改进或漏洞修复方案，欢迎提交拉取请求（Pull Request）。