Welcome to forums.fast.ai

Hello Fast Ai Users,
I am actually new to the deep learning stuff and community, I started right away with the coding part using Google Colab but am facing some issues.
Currently I am working on the Lesson 1 of Pets.
Sorry For Posting here as I didn’t know the right channel to post, if you could help me out wrt that it will be good.

I am not able to use the basic commands like:
help(untar_data)
It gives the following error

NameError                                 Traceback (most recent call last)
<ipython-input-6-896356bb8fd9> in <module>()
----> 1 untar_data

NameError: name 'untar_data' is not defined

Though I had used this:

#To install Necessary Packages.

!curl -s https://course19.fast.ai/setup/colab | bash

Hi Yuvraj hope all is well!

Not sure why you have chosen the old version of fastai.

I would suggest you start with version 2.0 which is the latest and current version.

This would be the place I would start if I were starting now.

Also searching the forum will provide a lot more help and information and help in choosing which thread to post in.

Cheers mrfabulous1 :smiley: :smiley:

Thank You @mrfabulous1 for responding.
I am currently working on this course https://course19.fast.ai/ which shows me it is version 3, also there were more lessons in it.

I will start with the new one then, if the V3 is older one.
Just one more thing could you pls provide the links for the forums that will be helpful.

Hi Yuvraj

For many people the best link is the one that solves any issue they may have!

Use the search bar to find links that can help solve any problems. Narrow down the search space to fastai2, then select and search within ‘Part 1 2020’ which has threads for fastai 2.

Also if you follow all the links in the first thread from the first link in my previous post, you see there are video lessons, sections on what notebook book platforms to use and deployment.

I believe this link acttuallly refers to fastai 1 version 3 as the latest version is fasta2.

Hope this helps.
mrfabulous1 :smiley: :smiley:

Hey all,
I am not a user but I would like to add this course to our Personal Development offer in our company.
Can anybody tell me, how much time it requires to finish the course?

Thanks a lot!

Hi LisaLi hope all is well!

Jeremy ran the current course Welcome to forums.fast.ai over an 8 week periood. However in the book there 20 lessons/chapters the course covers 8 of these. Jeremy advises A week of study per lesson.

Depending on factors such as time, previous knowledge and skill,I have seen people take 6 months to a few years to master the material. The book contains many furtther reading exercises.

However may people who have some programing experience or aptitude build their first image classifier in a notebook, in tthe first week or two and many of these are in the form of a simple app. (Share you work here - highlights)

Hope this helps
mrfabulous1 :smiley: :smiley:

Same issue here, is it written somewhere? I’m a bit rushy with a question

Thank you @mrfabulous1 for your help, I will take the needful action.
Have a great weekend.

1 Like

I am unable to get started with fast.ai course as all attempts to create a notebook via gradient is met with the error that says -This notebook is not supported by Gradient yet. Please open in Jupyter.
Opening in Jupyter shows an empty notebook and not the one shown in the instructions.
I’d appreciate any help including pointing me to the correct sub forum if this is the wrong forum to ask the question.
Thank you

I think this is because there is an error in pulling the fastai git repo. I see this error in the logs when I create the project using fastai base container
26: Cloning into ‘/notebooks’…
27: git: ‘remote-git+https’ is not a git command. See ‘git --help’

I worked around this problem by manually cloning the repo. In the empty Jupyter notebook that you get open a terminal and clone the fastbook repo
git clone https://github.com/fastai/fastbook.git

Hello everyone,

I am just playing around with Segmentation example in the first notebook. Unfortunately my results quite differ from ones in the book even though I’m using exactly the same code.

These are my loss values:
image

Values in the book:

This results in very bad segmentation prediction.

I Understand that results do vary from training session to training session but this seems too much.
How is it possible to gain such different results with same parameters?

Thank you in advance :slight_smile:

Hi everyone!
Commenting to be able to post :slight_smile:

Even I have received similar results like yours. It might be because of random batches ig. There might be few combinations where It is learning alot compare to the others.

Thank you for your reply!
I’m glad I’m not the only one getting those results. These small losses are probably result of randomness.
Still I think that it’s misleading to have results like these in the book if they are indeed less probable.

Hi! I’m working through Lesson 1 of Practical Deep Learning For Coders on Paperspace Gradient. I am not very familiar with Jupyter, but it looks like they are using a Juypter-like system, not Juypter itself. When I first start the course a few months ago (dropped due to being busy), they seemed to actually be using Jupyter.

This is causing me some problems that may or may not be related.

When I run doc(ImageDataLoadser.from_name_func), I don’t get output like shows in Lesson 1. Similarly, I can’t run untar_data without parameters to see where the method came from inline.

Anyone else have issues with this?

thanks for the edit, I was lost for a while

I guess I need to write a useless comment here first before I can go ask a concrete question? Great system you guys have set up. /s

if you were looking for ultimate guide for fortnite please visit www.proboosting.net

3 years later this comment saved me

Hi, I hope I’m posting in the right channel (if not please let me know: it’s for an open science/open data cause!).

We are happy to host the COVEO DATA CHALLENGE @ SIGIR 2021 - you can test your ML skills on a huge, new, rich eCommerce dataset that we just released to the AI community, and you’ll get a chance to write a research paper and present it at the conference!

DATES

  • Registration ends: June 1st, 2021
  • Final leaderboard: June 15th, 2021
  • SIGIR eCom Full day Workshop: July 15th, 2021

If you want to know more, ping me anytime - we also have a Slack channel you can join!