Frequently Asked Questions

From Deep Learning Course Wiki
Jump to: navigation, search

General

Do you have a Slack channel so I can ask questions? How can I join?
For most questions and discussions, we prefer to use Discourse, a free open-source platform for community discussions developed by Jeff Atwood and others. Our discourse instance is available at forums.fast.ai and open to anyone. Discourse has all the functionality of Slack, plus some great additional features:
  • Discourse is much easier to search (we find Slack history difficult to search, so questions often get asked multiple times on Slack and it can be hard to re-find a conversation of interest)
  • as you start typing a new topic into Discourse, a box will show you all similar topics.
  • Discourse has a system of different trust levels for users. As you spend more time on it, make posts, and receive likes, your trust level increases. There are different permissions at different trust levels.

That said, we also have a Slack channel. Once you read at least 3 different posts on the forums for a total of 10 minutes, a post will appear telling you how to join our Slack channel. Please continue to post your questions on the forums though, since they are more active and easier to search.

Will Part 2 be made available for free online?
Yes, we will release part 2 as a free online course this summer. There are still several benefits to taking the in-person course if you are able: being a part of the in-person community for the course and getting to ask questions in real-time.
Why don't you use Docker for your setup?
Docker won't install graphics drivers or set up CUDA for you, so it would not be very helpful for our use case of creating a GPU setup (which is what you want for deep learning). If you disagree and want to create a docker setup to share with the class, we welcome your contribution. The 1st step is for you to go through our recommended AWS Setup, so that you fully understand our technical requirements.

Getting Started

How can I access the notebooks and python files used in the videos?
We have provide the jupyter notebooks and python files that Jeremy uses in the class lectures in a github repository. Note that in the videos, Jeremy may reference files being available on platform.fast.ai, but we have since switched to github and the versions on github are most up to date. If you are new to Github, here is how to get started. Be sure to read our strategy for How to use the Provided Notebooks in order to get the most out of the course.
When I try to import Keras, I get 'ImportError No module named tensorflow.python'
Keras can use either Theano or TensorFlow for it's backend. In Part 1 of this course, we use Theano. If you are getting this error, you need to change the keras config file (~/.keras/keras.json) to have "theano" as the value for backend. Refer to the Keras documentation, for more details.

Questions about AWS

Why am I getting an InstanceLimitExceeded error?
You are not approved for g2 or p2 instances yet, so you will need to use a t2 instance instead. Read this for more details.
Why am I getting a VpcLimitExceeded error?
Each time you run a setup script, it creates a VPC and there's a limit of 5. Go to https://console.aws.amazon.com/vpc/home and delete all the existing VPCs.
When I type 'nvidia-smi', I get the message 'Failed to initialize NVML Driver/library version mismatch'
Restart your instance.
Why am I getting this other error?
We have compiled a list of some of the most common AWS errors you may encounter and their solutions. After reading through those, if you are still stuck and your AWS has ended up in a weird state, you may want to delete things and start over. See this guide on Starting Over with AWS.