Minimal implementation of a Transformer and a pretraining script for language modeling in PyTorch. It supports compile, DDP.

Meta Lingua: a lean, efficient, and easy-to-hack codebase to research LLMs.

Efficient Triton Kernels for LLM Training

[CVPR 2022 (oral)] Bongard-HOI for benchmarking few-shot visual reasoning

Autoregressive Model Beats Diffusion: 🦙 Llama for Scalable Image Generation

PyContinual (An Easy and Extendible Framework for Continual Learning)

[ICLR'25] BigCodeBench: Benchmarking Code Generation Towards AGI

Release for Improved Denoising Diffusion Probabilistic Models

Official PyTorch Implementation of "Scalable Diffusion Models with Transformers"

A PyTorch implementation of the paper "ZigMa: A DiT-Style Mamba-based Diffusion Model" (ECCV 2024)

A minimal yet resourceful implementation of diffusion models (along with pretrained models + synthetic images for nine datasets)

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

AI Image SIgnal Processing and Computational Photography - Bokeh Rendering , Reversed ISP Challenge, Model-Based Image Signal Processors via Learnable Dictionaries. Official repo for NTIRE and AIM …

ReFT: Representation Finetuning for Language Models

A family of lightweight multimodal models.

A mini-library for training consistency models.

[ICLR-2023] Rarity Score : A New Metric to Evaluate the Uncommonness of Synthesized Images

NeurIPS 2023 - TopP&R: Robust Support Estimation Approach for Evaluating Fidelity and Diversity in Generative Models Official Code

E-LPIPS: Robust Perceptual Image Similarity via Random Transformation Ensembles

[Mamba-Survey-2024] Paper list for State-Space-Model/Mamba and it's Applications

Continual Forgetting for Pre-trained Vision Models (CVPR 2024)

ViT Prisma is a mechanistic interpretability library for Vision Transformers (ViTs).

[CVPR 2024 Highlight] Logit Standardization in Knowledge Distillation

[ECCV2024] MambaIR and MambaIRv2!

Easily benchmark PyTorch model FLOPs, latency, throughput, allocated gpu memory and energy consumption

Official implementation of "When Adversarial Training Meets Vision Transformers: Recipes from Training to Architecture" published at NeurIPS 2022.