Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Library for industrial alignment.
| Date | Stars |
|---|---|
| 2026-07-31 | 405 |
| 2026-08-04 | 406 |
| 2026-08-06 | 406 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# 🚀 Turbo-Alignment
> Library for industrial alignment.
## Table of Contents
- [What is Turbo-Alignment?](#-what-is-turbo-alignment)
- [Key Features](#-key-features)
- [Supported Methods](#-supported-methods)
- [Implemented metrics](#-implemented-metrics)
- [How to Use](#-how-to-use)
- [Installation](#-installation)
- [Development](#-development)
- [Library Roadmap](#-library-roadmap)
- [FAQ](#-faq)
- [License](#-license)
<a name="-what-is-turbo-alignment"></a>
## 🌟 What is Turbo-Alignment?
Turbo-Alignment is a library designed to streamline the fine-tuning and alignment of large language models, leveraging advanced techniques to enhance efficiency and scalability.
<a name="-key-features"></a>
## ✨ Key Features
- **📊 Comprehensive Metrics and Logging**: Includes a wide range of metrics such as self-bleu, KL divergence, diversity, etc. all supported out of the box.
- **🛠️ Streamlined Method Deployment**: Simplifies the process of deploying new methods, allowing for quick development and integration of new datasets and trainers into your pipelines.
- **📚 Ready-to-Use Examples**: Convenient examples with configurations and instructions for basic tasks.
- **⚡ Fast Inference**: Optimized for quick inference using vLLM.
- **🔄 End-to-End Pipelines**: From data preprocessing to model alignment.
<a name="-supported-methods"></a>
## 🛠️ Supported Methods
Turbo-Alignment supports a wide range of methods for model training and alignment, including:
- **🎯** Supervised Fine-Tuning
- **🏆** Reward Modeling
- **👍** Offline Preference Optimization
- **🤗** Online Preference Optimization ⌛️ in progress..
<a name="-implemented-metrics"></a>
## 🧮 Implemented Metrics
- **🎯** Accuracy
- **🔠** Distinctness
- **🌈** Diversity
- **🔵** Self-BLEU
- **➗** KL-divergence
- **🏆** Reward
- **📏** Length
- **🌀** Perplexity
<a name="-how-to-use"></a>
## 🤖 How to Use
Turbo-Alignment offers an intuitive interface for training and aligning large language models. Refer to the detailed examples and configuration files in the documentation to get started quickly with your specific use case. User-friendly guid available [here](docs/GUIDE.md).
The most crucial aspect is to prepare the dataset in the required format, after which the pipeline will handle everything automatically.
Examples of datasets are available [here](docs/dataset_example.md).
## Table of use-cases
- [Training](#-train)
- [Supervised Fine-Tuning](#-sft-train)
- [Preference Tuning](#-preftune-train)
- [Reward Modeling](#-rm-train)
- [Offline Preference Optimization](#-dpo-train)
- [Inference](#-inference)
- [Supervised Fine-Tuning](#-sft-inference)
- [Sampling](#-sampling)
- [Random](#-random-sampling)
- [RM](#-rm-sampling)
- [RSO](#-RSO-sampling)
- [Common](#-common)
- [Preprocess](#-preprocess-common)
- [Merge adapters to base](#-merge-adapters-to-base-common)
<a name="-train"></a>
# Train
<a name="-sft-train"></a>
## Supervised Fine-Tuning
- **📚 Dataset type** prepare your dataset in the `ChatDataset`, examples available [here](docs/dataset_example.md#-chat-dataset) format.
- **📝 Configs Example**: [sft.json](tutorials/sft/sft.json)
- **🖥️ CLI launch command**
```bash
python -m turbo_alignment train_sft --experiment_settings_path tutorials/sft/sft.json
```
<a name="-preftune-train"></a>
## Offine Preference Tuning
<a name="-rm-train"></a>
### Reward Modeling
- **📚 Dataset type** prepare your dataset in the `PairPreferencesDataset` format, examples available [here](docs/dataset_example.md#-pair-preferences)
- **📝 Configs Example**: [rm.json](tutorials/rm/rm.json)
- **🖥️ CLI launch command**
```bash
python -m turbo_alignment train_rm --experiment_settings_path tutorials/rm/rm.json
```
<a name="-dpo-train"></a>
### DPO
- **📚 Dataset type** prepare your dataset in the `PairPreferencesDataset` format, examples available [here](docs/dataset_example.md#pair-preferences)
- **📝 Configs Example**: [dpo.json](tutorials/dpo/dpo.json)
- **🖥️ CLIExcerpt of 9,023 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:7fbeba92ceba303b, name:alignment, desc:alignment