[NeurIPS 2023] ImageReward: Learning and Evaluating Human Preferences for Text-to-image Generation

CLIP+MLP Aesthetic Score Predictor

AlignProp uses direct reward backpropogation for the alignment of large-scale text-to-image diffusion models. Our method is 25x more sample and compute efficient than reinforcement learning methods…

Code for the paper "Training Diffusion Models with Reinforcement Learning"

A 100x faster SVD for PyTorch⚡️

Dee-seek_project

ICLR 2025 (Spotlight) One-Prompt-One-Story: Free-Lunch Consistent Text-to-Image Generation Using a Single Prompt

提取微信聊天记录，将其导出成HTML、Word、Excel文档永久保存，对聊天记录进行分析生成年度聊天报告，用聊天数据训练专属于个人的AI聊天助手

Official Implementation for "Consistency Flow Matching: Defining Straight Flows with Velocity Consistency"

从零手搓Flow Matching（Rectified Flow）

Code for the paper "Aligning Few-Step Diffusion Models with Dense Reward Difference Learning"

[ICML 2024] Code for the paper "Confronting Reward Overoptimization for Diffusion Models: A Perspective of Inductive and Primacy Biases"

Human Preference Score v2: A Solid Benchmark for Evaluating Human Preferences of Text-to-Image Synthesis

cuML - RAPIDS Machine Learning Library

Faster generation with text-to-image diffusion models.

PyTorch native quantization and sparsity for training and inference

xDiT: A Scalable Inference Engine for Diffusion Transformers (DiTs) with Massive Parallelism

Hackable and optimized Transformers building blocks, supporting a composable construction.

This is the pytorch implementation of Hindsight Experience Replay (HER) - Experiment on all fetch robotic environments.

My implementation of Hindsight replay in PyTorch: "Hindsight Experience Replay"

PixArt-α: Fast Training of Diffusion Transformer for Photorealistic Text-to-Image Synthesis

T-GATE: Temporally Gating Attention to Accelerate Diffusion Model for Free!

Hunyuan-DiT : A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding