What is this thread?
As everyone is exploring the wonderful new library and creating awesome extensions, there currently is not a list of everyone’s sub-libraries that they have made.
Well no more, welcome to the list!
To submit your own library to this list it needs to fulfill one requirement: It must be an extension/sublibrary.
Note: this thread is locked, but users can still edit this post
Simply follow this template format and submit yours into either a related section, or make a new one with two (##)'s:
- myLibrary github pip by @ your name, a quick TL;DR of what the goal of the library is and does (no more than 1-2 sentences max)
Natural Language Processing
- fasthugs github by @morgan, using fastai with HuggingFace’s pretrained transformers
- blurr github pip docs by @wgpubs, using fastai with HuggingFace’s pretrained transformers
- hugdatafast github pip docs by @Richard-Wang, integrating HuggingFace preprocessing to generate fastai
DataLoaders
Computer Vision
-
SemTorch github pip by @WaterKnight, bringing state-of-the-art segmentation architectures into fastai
-
faststyle github pip docs by @lgvaz, aiming to provide an easy and modular interface for Image to Image problems based on feature loss
-
UPIT github docs by @ilovescience, a fastai/PyTorch package for unpaired image-to-image translation
-
IceVision github pip docs by @lgvaz and @farid, an agnostic object detection framework.
Multi-Topic Encompassing Extensions
-
fastai2_extensions github pip docs by @rsomani95, a collection of interpretation, augmentation, and inference utility functions including for ONNX
-
fastinference github pip docs by @muellerzr, a collection of inference modules aimed at speeding it up, tabular and computer vision model interpretation, ONNX exportation and integration
-
fastai_xla_extensions github by @butchland and @tyoc213, allows for fastai/Pytorch models to run on TPUs using the Pytorch-XLA library
Self-Supervised Learning
- self-supervised github pip docs by @kcturgutlu, implementations of popular SOTA self-supervised learning algorithms as fastai Callbacks