Use PEFT or Full-parameter to finetune 400+ LLMs or 100+ MLLMs. (LLM: Qwen2.5, Llama3.2, GLM4, Internlm2.5, Yi1.5, Mistral, Baichuan2, DeepSeek, Gemma2, ...; MLLM: Qwen2-VL, Qwen2-Audio, Llama3.2-Vision, Llava, InternVL2, MiniCPM-V-2.6, GLM4v, Xcomposer2.5, Yi-VL, DeepSeek-VL, Phi3.5-Vision, ...)

MedicalGPT: Training Your Own Medical GPT Model with ChatGPT Training Pipeline. 训练医疗大模型，实现了包括增量预训练(PT)、有监督微调(SFT)、RLHF、DPO、ORPO。

A library with extensible implementations of DPO, KTO, PPO, ORPO, and other human-aware loss functions (HALOs).

Easy and Efficient Finetuning LLMs. (Supported LLama, LLama2, LLama3, Qwen, Baichuan, GLM , Falcon) 大模型高效量化训练+部署.

tensorflow를 사용하여 텍스트 전처리부터, Topic Models, BERT, GPT, LLM과 같은 최신 모델의 다운스트림 태스크들을 정리한 Deep Learning NLP 저장소입니다.

Implementation for "Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMs"

An Efficient "Factory" to Build Multiple LoRA Adapters

SiLLM simplifies the process of training and running Large Language Models (LLMs) on Apple Silicon by leveraging the MLX framework.

Align Anything: Training All-modality Model with Feedback

Notus is a collection of fine-tuned LLMs using SFT, DPO, SFT+DPO, and/or any other RLHF techniques, while always keeping a data-first approach

Step-aware Preference Optimization: Aligning Preference with Denoising Performance at Each Step

Technical anaysis library for .NET

IterComp: Iterative Composition-Aware Feedback Learning from Model Gallery for Text-to-Image Generation

A RLHF Infrastructure for Vision-Language Models

This repository contains the code for SFT, RLHF, and DPO, designed for vision-based LLMs, including the LLaVA models and the LLaMA-3.2-vision models.

CodeUltraFeedback: aligning large language models to coding preferences

SuperCorrect: Supervising and Correcting Language Models with Error-Driven Insights

[ACL 2024] Self-Training with Direct Preference Optimization Improves Chain-of-Thought Reasoning

[NeurIPS 2024] Official code of $\beta$-DPO: Direct Preference Optimization with Dynamic $\beta$

基于DPO算法微调语言大模型，简单好上手。