Code of Pyramidal Flow Matching for Efficient Video Generative Modeling

Next-Token Prediction is All You Need

Official Implementation of "Lumina-mGPT: Illuminate Flexible Photorealistic Text-to-Image Generation with Multimodal Generative Pretraining"

[ICLR 2024] Fine-tuning LLaMA to follow Instructions within 1 Hour and 1.2M Parameters

Official inference repo for FLUX.1 models

MINT-1T: A one trillion token multimodal interleaved dataset.

Toolset for creating animated React components built on Theatre.js and Framer Motion.

PyTorch native quantization and sparsity for training and inference

Code for Adam-mini: Use Fewer Learning Rates To Gain More https://arxiv.org/abs/2406.16793

[ECCV 2024] official code for "Long-CLIP: Unlocking the Long-Text Capability of CLIP"

Qwen2.5 is the large language model series developed by Qwen team, Alibaba Cloud.

[CVPR`2024, Oral] Attention Calibration for Disentangled Text-to-Image Personalization

InstantID: Zero-shot Identity-Preserving Generation in Seconds 🔥

Analyzing and Improving the Training Dynamics of Diffusion Models (EDM2)

Project Page for "LISA: Reasoning Segmentation via Large Language Model"

[ ECCV 2024 ] MotionLCM: This repo is the official implementation of "MotionLCM: Real-time Controllable Motion Generation via Latent Consistency Model"

DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model

Universal Tensor Operations in Einstein-Inspired Notation for Python.

Implementation for MatMul-free LM.

Official repository of the xLSTM.

Lumina-T2X is a unified framework for Text to Any Modality Generation

Tile primitives for speedy kernels

Your image is almost there!

[ECCV2024] This is an official inference code of the paper "Glyph-ByT5: A Customized Text Encoder for Accurate Visual Text Rendering" and "Glyph-ByT5-v2: A Strong Aesthetic Baseline for Accurate Mu…

Official repository for LightSeq: Sequence Level Parallelism for Distributed Training of Long Context Transformers

Memory optimization and training recipes to extrapolate language models' context length to 1 million tokens, with minimal hardware.

USP: Unified (a.k.a. Hybrid, 2D) Sequence Parallel Attention for Long Context Transformers Model Training and Inference