Best practice for training LLaMA models in Megatron-LM

OpenVLA: An open-source vision-language-action model for robotic manipulation.

An open-source implementation of Scaling Laws for Neural Language Models using nanoGPT

Easily compute model embeddings and save the embeddings.

The simplest, fastest repository for training/finetuning medium-sized GPTs.

Ongoing research training transformer models at scale

Reproducible scaling laws for contrastive language-image learning (https://arxiv.org/abs/2212.07143)

The official implementation of Latte: Latent Diffusion Transformer for Video Generation.

Utilities for Training Very Large Models

Python 3 support for the MS COCO caption evaluation tools

bio-datasets

Locally run an Instruction-Tuned Chat-Style LLM

Turn any collection of files into a dataset

Inference code for LLaMA models

Fork of Facebooks LLaMa model to run on CPU

A set of examples around pytorch in Vision, Text, Reinforcement Learning, etc.

Tools to train a generative model on arbitrary audio samples

Optimized Stable Diffusion modified to run on lower GPU VRAM

A better PyTorch implementation of image local attention which reduces the GPU memory by an order of magnitude.

Probing the representations of Vision Transformers.

Colab notebook to finetune GLIDE.

Ongoing research training transformer language models at scale, including: BERT & GPT-2

An open source implementation of CLIP.

This is an official implementation of CvT: Introducing Convolutions to Vision Transformers.

Contrastive Language-Audio Pretraining