A latent text-to-image diffusion model

CLIP (Contrastive Language-Image Pretraining), Predict the most relevant text snippet given an image

Data augmentation for NLP

SimCLRv2 - Big Self-Supervised Models are Strong Semi-Supervised Learners

Debugging, monitoring and visualization for Python Machine Learning and Data Science

maximal update parametrization (µP)

Normalizing flows in PyTorch. Current intended use is education not production.

Named tensors with first-class dimensions for PyTorch

Code for the Paper "Restricting the Flow: Information Bottlenecks for Attribution"