ML Model Development Framework & Model Repositories
There are hundreds of machine learning tasks. To do these tasks there are thousands of datasets created by individuals, governments, and corporations. We need to develop AI models using these datasets. There are thousands of models 1, 2, 3 and many model development frameworks. It is practically mind-blowing to track this whole body of work and understand all this work in its entirety. But if you dive deeper into the following frameworks you will get a fair idea about the overall direction of the work. These frameworks are used to maintain pre-trained model repositories and download pre-trained models. You can develop your own finetuned model using those pre-trained models.
List of Frameworks
- Model Repo - tfhub.dev : AI Model repository from google.
- tensorflow.org Website : an open-source software library for machine learning from Google
- Microsoft AI on github
- Microsoft Cognitive Toolkit Website (previously known as CNTK), an open source toolkit for building artificial neural networks.
github : github repo of CNTK
- Model Repo - pytorch.org/hub : An open-source Tensor and Dynamic neural network in Python from facebook
Framework - pytorch.org
- Keras, a high level open-source software library for machine learning (works on top of other libraries).
- huggingface.co/models : AI Model Repository from huggingface.
- modelzoo.co : Discover open source deep learning code and pretrained models. Repository is created by Jing Yu Koh
- modelplace.ai : Ready-to-Integrate Computer Vision Neural Networks
- Apache Mahout, a library of scalable machine learning algorithms.
- Deeplearning4j, an open-source, distributed deep learning framework written for the JVM.
- OpenNN Website, a comprehensive C++ library implementing neural networks. OpenNN github repo
- Theano, a Python library and optimizing compiler for manipulating and evaluating mathematical expressions, especially matrix-valued ones.