search

zackchase / mxnet-the-straight-dope

An interactive book on deep learning. Much easy, so MXNet. Wow. [Straight Dope is growing up] ---> Much of this content has been incorporated into the new Dive into Deep Learning Book available at https://d2l.ai/.
https://d2l.ai/
Apache License 2.0
2.56k stars 727 forks source link

Training on multiple processes #362

Open dai-dao opened 6 years ago

dai-dao commented 6 years ago

Is there any examples / tutorials on how to use the tools/launch.py that supports training on different processes, either on different machines or on a single machine?

The documentation touched very briefly on this and barely mention the "tools/launch.py" script so I'm not exactly sure how to proceed on this.

What I'm interested in is implementing A3C Reinforcement learning with Gluon, so maybe Asynchronous SGD would be needed (on the Roadmap), when can I expect to have this feature available?

Thanks

eric-haibin-lin commented 6 years ago

We're adding a tutorial for distributed training in general (not specific to gluon) here https://github.com/apache/incubator-mxnet/pull/9152