[ICML 2024] LESS: Selecting Influential Data for Targeted Instruction Tuning

Rewarded soups official implementation

[ACL'2024] Beyond One-Preference-Fits-All Alignment: Multi-Objective Direct Preference Optimization

A lightweight library for large laguage model (LLM) jailbreaking defense.

Official Repository for "Tamper-Resistant Safeguards for Open-Weight LLMs"

Official repository for ``Concept-based Interpretation Without Linear Assumption'' published in ICLR 2023.

Inference-Time Intervention: Eliciting Truthful Answers from a Language Model

A framework for few-shot evaluation of language models.

ConceptVectors Benchmark and Code for the paper "Intrinsic Evaluation of Unlearning Using Parametric Knowledge Traces"

A curated list of LLM Interpretability related material - Tutorial, Library, Survey, Paper, Blog, etc..

欢迎来到 LLM-Dojo，这里是一个开源大模型学习场所，使用简洁且易阅读的代码构建模型训练框架(支持各种主流模型如Qwen、Llama、GLM等等)、RLHF框架(DPO/CPO/KTO/PPO)等各种功能。👩‍🎓👨‍🎓

A recipe for online RLHF.

Improving Alignment and Robustness with Circuit Breakers

Comprehensive toolkit for Reinforcement Learning from Human Feedback (RLHF) training, featuring instruction fine-tuning, reward model training, and support for PPO and DPO algorithms with various c…

Directional Preference Alignment

Recipes to train reward model for RLHF.

Trains and compares a variety of preference models (reward models) with different losses and datasets.

LLM Unlearning

kaggle:otto competition

News Recommendation with Category Description by a Large Language Model

(WSDM 2024) Official implementation of the paper "ONCE: Boosting Content-based Recommendation with Both Open- and Closed-source Large Language Models"