Lesson 1 In-Class Discussion

(Jeremy Howard) #1

This thread contains the in-class discussion from Lesson 1. The Wiki links have been moved to this new thread. Please ask any questions about lesson 1 in the new wiki thread.

Clarity on homework / "what's due"
International fellowship thread
Up and running on AWS in less that 60 seconds
Idea: Lessons Pinned thread
(Jeremy Howard) #2

(Mudit Verma) #3

will there be any recorded version as well ? (of the live stream)

(Ankit Goila) #4

You can watch the streamed (live) video later through the same link. :slight_smile:

(naveen manwani) #5

@jeremy what are these group for.

(Victor Alfonso Arias Vanegas) #6

groups? what is this?

(Jeremy Howard) #7

You’ll see! :smiley:

(Divyansh Jha) #8

Will you help us setup aws today?

(Kevin Bird) #9

Working on going through lesson 1 and having trouble with this command:

It doesn’t look like I can use conda to install it. Is there a way to install it in my conda environment without conda?

I tried using pip3, but I’m wondering if it didn’t install to my conda environment.

(Rajat Gupta) #11

Hi Jeremy,

I am not able to find my name on the shared spreadsheet. Can you please help. :slight_smile:


(Sid Ravinutala) #13

Hello all,

I am a development economics researcher hoping to train NN on last 40 years of trade data to better understand industrialization pathways and to develop better policy recommendations. My undergrad is in Comp Sci and Electrical Eng and Masters in Development Econ.

I have really enjoyed doing ML over the last few years and hope to transition to a career in deep learning (ideally with applications in economic development) and this is my first step towards it.



Hi @kevinb,

Sorry for the multiple replies from my - trying to do multiple things at once and not working that well :slight_smile:

You could try giving this a go. I haven’t tried this myself but the docs say this is the way to go so maybe it will work :wink:

(Chris Palmer) #15

I have the same issue. Hopefully all will be revealed once class begins!

For what its worth, I copied the git repository into a sub-directory, which is the same structure as the repo. In the root if the directory is a yml file, I issued the following command and it built a new environment for me - there were no errors reported:

conda env create -f environment.yml

I then activated the environment (command source activate fastai), and navigated down to the dl1 sub-directory where today’s lessons Jupyter Notebook is located, and started the notebook server. Then, finally, linked to the running server and from lesson1 notebook I attempted to import from fastai.imports as outlined by KevinB, and got the same error.

Obviously missing an important step here :slight_smile:

(yinterian) #16

pip install isoweek

(Devan Govender) #17

As an alternative, the lesson is pre-loaded on Crestle.

(WG) #18

You can do a > conda list to see what packages are installed to see if the packages its complaining about are even there.

(Clayton Yochum) #19

Are you geting the new lessons on Crestle? I only have the old ones :frowning:

(Leon Adams) #20

I think you can get the environment set up with the following command:

conda env create -f environment.yml

Use this in the cloned directory that have the environment.yml file…


(WG) #21

Open the terminal and git clone the fastai repo:

git clone https://github.com/fastai/fastai.git

That should get you everything you need. Personally, I’m wating to see how the evening plays out … watch/participate in class and then put together an environment on AWS over the next couple of days.

(yinterian) #22

git clone https://github.com/fastai/fastai.git