Scanning all new published packages on PyPI I know that the quality is often quite bad. I try to filter out the worst ones and list here the ones which might be worth a look, being followed or inspire you in some way.

• geffnet
(Generic) EfficientNets for PyTorch. A ‘generic’ implementation of EfficientNet, MixNet, MobileNetV3, etc. that covers most of the compute/parameter efficient architectures derived from the MobileNet V1/V2 block sequence, including those found via automated neural architecture search. All models are implemented by GenEfficientNet or MobileNetV3 classes, with string based architecture definitions to configure the block layouts (idea from [here](https://…/mnasnet_models.py ))

• metamorphosis
An opinionated event streaming microframework. Metamorphosis is an event-streaming based microframework that builds on top of Flask, Graphene, Kafka, and Redis. It allows you to define microservices, which consist of a number of mutations (commands), queries, and event consumers.

• sequence-builder
Building sequences from time-series data

• tensorflow-serving-client-grpc
A prebuilt tensorflow serving client from the tensorflow serving proto files

• tensorly-musco
Tensor learning in Python.

• tf-helpers
A collection of useful functions for TensorFlow for training and visualization.

• torchray
TorchRay is a PyTorch library of visualization methods for convnets. The TorchRay package implements several visualization methods for deep convolutional neural networks using PyTorch. In this release, TorchRay focuses on attribution, namely the problem of determining which part of the input, usually an image, is responsible for the value computed by a neural network.

• hyperdim
Hyperdimensionality computing machine learning library