New release cycle
Starting from the release of v1.18.0, we will have one release every four weeks instead of the current cycle of one release every two weeks.
We had been keeping the biweekly release cycle since the first release in June, 2015. This release cycle worked well at the initial stage of the development when the number of features was small and the user base was not so large. After more than one year of development, we now have a lot of features and a relatively large user base, for which the current fast release cycle is no longer a fit to several demands, including the following:
- It is hard for users to catch up with the latest updates every two weeks.
- The release management is becoming a bottleneck of the development of Chainer.
- It is becoming hard to maintain the quality of each release.
Therefore, we have decided to change the release cycle. Starting from the release of v1.18.0, there will be a release every four weeks instead of every two weeks. Note that we may still have an accidental hot fix release at any time to fix critical issues.
Chainer is a Python-based, standalone open source framework for deep learning models. Chainer provides a flexible, intuitive, and high performance means of implementing a full range of deep learning models, including state-of-the-art models such as recurrent neural networks and variational autoencoders.
- Chainer v2.0.0 and our future development plans
- Performance comparison of LSTM with and without cuDNN(v5) in Chainer
- ChainerRL - Deep Reinforcement Learning Library
- Performance of Distributed Deep Learning using ChainerMN
- Research projects using Chainer
- Plan of v2
- New release cycle
- About Chainer Blog