AWS AMI available for testing

(Phani Srikanth) #44

For my p2 limits request I have placed last week, they responded in 2 working days.

(Jeremy Howard) #45

I’ve updated the AMI to fix a couple of minor issues. The new ID is ami-8c4288f4 . Name is the same. I’ll updated the top post.

(Victor Alfonso Arias Vanegas) #46

Hi Mr. @jeremy, I don’t see the image

(Vikrant Behal) #47

@jeremy Should be setup AWS before lecture 2 or you’ll brief us?

(João Marcos Gris) #48

I was not able to find the image too.

(Ravindra Mahar) #49

I also search for “ami-8c4288f4” both in market place as well in community AMIs, no luck. If I am searching fastai only in community I am getting some results but they have different amAMI ID. I guess those are some old stuffs.

(Nandamuri Hari Naga Sumanth) #50

Is it required to clone the repository to my laptop or to aws instance?.
Some one please tell what need to be done after launching an instance in order to be able to run the course notebooks.
After running a notebook from an aws instance will the changes be saved and reflected in the notebook present in my local repo ?

(Phani Srikanth) #51

Most likely the new AWS AMI is set to private.

Incase you’re working on the aws instance, you might follow these steps.

  1. Launch the instance on AWS console.
  2. ssh -i your-aws-cert.pem ubuntu@ec2-instance-ip
  3. cd fastai; git checkout --environment.yml; git pull
  4. open tmux, launch jupyter jupyter-notebook --no-browser --ip='*'. Take note of the token. You need this in the next step.
  5. Visit the ec2-instance-ip:8888 on your web browser. Enter the token to see your notebooks.
  6. Work with the notebooks.

If you do port forwarding in step 2 using ssh -i your-aws-cert.pem ubuntu@ec2-instance-ip -L 8888:localhost:8888, then in step 5, you can access notebooks on localhost:8888 on your web browser.

If you make a few changes to your notebooks and you’d like to save them, you need to commit the changes to your aws-local-repo, push them to your remote fork and pull them to your laptop-local-repo.

Hope this helps.

Edit: Fixed the port-forwarding command.

(Ankit Goila) #52

Thanks for sharing that process!

Just a small doubt: what does this line do? git checkout --environment.yml

Not very familiar with git, so just wanted to clarify. :slight_smile:

(Vinay) #53

It reverts the changes made in environment.yml

(Mudit Verma) #54

same , unable to find the ami.


You need to make sure you are looking in the correct region. Are you using Oregon? Here is more info on AWS regions in case it might be useful.

(Chris Palmer) #56

Hi Jeremy

I am unable to find the AMI in the list of AMIs under the Images group - is this not the correct place to find it? The only AMI I can see which is dated in November is called V100-fastai. There is a fastai-gpu-v2 set up on October 31st, but that is not what you have identified in your post.


(K Sreelakshmi) #57

Unable to find the new AMI “ami-8c4288f4” in the Oregon region. The previous one “ami-5f2fe427” is also not available.

(Nikhil B ) #58

Phani, What did you put in the description section, did you request more than 2?
Its been a few days for me.

(Phani Srikanth) #59

I mentioned this in the description Requesting these instances for training deep neural networks for course. I requested for 4 and it was approved within 2 working days.

(Jeremy Howard) #60

Yeah looks like it was indeed set to ‘private’ still. I’ve made it public now - please try again! Note you need to set your region to ‘Oregon’. Sorry about the confusion…

(Nikhil B ) #61

great, thanks

(Benjamin DeKoven) #62

I am excited to finally have this AMI running on p2, thank you @jeremy.

However, I am concerned about the cost over the duration of the course.

(Jeremy Howard) #63

If you do 10 hours per week, for 7 weeks, it’ll cost you ~$70 total. If you got the AWS $500 credits, then you’ll have lots left over :slight_smile: