Projects sharing features with Rlcd

Align-anything is a multi-modal large language model alignment framework designed to fine-tune models across text, image, video, and audio. It functions as a distributed training orchestrator and toolkit for implementing preference-based learning to ensure model behaviors match human intentions and values. The framework provides specialized pipelines for Supervised Fine-Tuning and Direct Preference Optimization. It includes a high-performance inference engine wrapper for actor models to reduce sequence generation time and a dedicated training environment for refining vision-language-action mo

Ludwig is a declarative machine learning framework designed for training neural networks and large language models using configuration files instead of manual coding. It functions as a multimodal model builder and a low-code tool for supervised fine-tuning, allowing users to build models that process mixed inputs of text, images, audio, and tabular data. The project distinguishes itself through an automated hyperparameter optimizer and a system for large language model fine-tuning using parameter-efficient adapters. It features a multimodal data pipeline and the ability to automatically gener

An unofficial implementation of Self-Alignment with Instruction Backtranslation .

Self-Knowledge Guided Retrieval Augmentation for Large Language Models (EMNLP Findings 2023)

A large-scale, fine-grained, diverse preference dataset

The official implementation of Self-Play Fine-Tuning (SPIN)

Self-instruct is a framework for generating synthetic instruction datasets and fine-tuning large language models to improve their instruction-following capabilities. It provides a pipeline for aligning pretrained models with human intentions through a supervised fine-tuning workflow. The system utilizes a synthetic data generator that uses a seed set of tasks to prompt a model to create new instructional data. It includes an instruction dataset curator to remove redundant or low-quality entries, maintaining dataset diversity through a filtered task pool. The framework covers the full alignme

1. STaR 2. Mesh Transformer JAX 1. Updates 3. Pretrained Models 1. GPT-J-6B 1. Links 2. Acknowledgments 3. License 4. Model Details 5. Zero-Shot Evaluations 4. Architecture and Usage 1. Fine-tuning 2. JAX Dependency 5. TODO

Inspired by Meta's mid-year paper "Self-Alignment with Instruction Backtranslation," our project embarks on an innovative approach to enhance language model training using a novel data augmentation paradigm. This method, rooted in the principles of self-alignment, involves a meticulous process…

This repository contains PyTorch code for Motif, training AI agents on NetHack with reward functions derived from an LLM's preferences.

This repository is the official code repository for our paper Impossible Distillation: from Low-Quality Model to High-Quality Dataset & Model for Summarization and Paraphrasing.

Implementation of the training framework proposed in Self-Rewarding Language Model , from MetaAI

Let ChatGPT teach your own chatbot in hours with a single GPU!

RAIN is an innovative inference method that, by integrating self-evaluation and rewind mechanisms, enables frozen large language models to directly produce responses consistent with human preferences without requiring additional alignment data or model fine-tuning, thereby offering an effective…