I apologize in advance if this question has been asked before, but why does the 1st video stop at LR finder explanation, whereas the 1st notebook has a lot more stuff after that?
Jeremy believes I guess that something should be left for self exploration otherwise it’s like spoon fed…
Also don’t go by the cell number, it’s not that way in the later notebooks…
Model will have parameters?
Model . summary ()?
Just print the learner (m)
(Forgotten the proper call, haven’t run the DL notebooks in a while)
Ok, I just believed the nbs were not yet synced with the videos.
Also, thank you so much for v2. I thought v1 was incredible but struggled with setup (windows native) and also with modifying notebooks to different datasets while maintaining performance. I have only completed the first v2 lesson and used one Kaggle dataset, but the process has been so much smoother and faster. I look forward to digging into the details as I progress!
Thank you so much for what you and Rachel are doing for the community, I cannot believe this is all free. I hope more people will take advantage of such an incredible resource in order to improve lives around the world.
Could anyone reply with a link to upload your dataset into Crestle in order to practice lesson 1
please, is it possible to accelerate the pip version update of fastai library?
https://pypi.python.org/pypi/fastai/0.6 (that is from 2017-11-29).
Indeed, I found it more stable in some environments, but the new updates such as the opencv dependency removing in favor of PIL have to be taken into account.
Could we have an idea of when that will be done, please?
Hi, can you please send another link? This link has expired too! I am desperately looking for a study group!
Hello, Just saying HI! I’m about done with the first week, and really enjoying the course! Thanks a million for putting this stuff online!
Data Science @ College Vine
Can Lesson 1 code be used for Google’s Landmark Recognition Challenge? It contains nearly 1.2M training images and and close to 15K classes.
[ref: lesson 3 video @1:56:56 ]
I am part of the AI class here in Lagos, it has been overwhelming and wonderful
Had not realised you have run the Coders course again. When I found it I played the lesson 1 video and was really pleased with what I saw and heard. It is a privilege to watch a very polished presentation and has rekindled a desire to run through the lessons again. Especially with the new improvements which this issue has been built upon.
I would like to start out right at the beginning running the code on my Ubuntu 16.04 server with gtx1080 ti. I used this while running the 2017 version. My server has not been run for a while so on starting it up there will be a lot of software up grades to perform and this may break my Cuda system. Before I do these upgrades is it written somewhere how to set up the PyTorch environment, or what the requirements are for PyTorch. I guess I’ll head there first, but some pointers would be welcome.
Thanks to yourself and Rachel for your commitment.
Everything is available at Reshma’s GitHub repository…
Or At fast.ai GitHub’s repository
Searching the forum will help…
I’ve overcome the biggest hurdle and thats updating my local box with the latest drivers etc for GPU since I have’nt used it in months. I am running the yml script which returns dependency issues for pytorch >= 0.2.0 but seem to have solved that with a conda update of my base anaconda3 environment. Have installed cuda-9-1 for my gpu engine so perhaps there may be some issues with dependencies in the yml file.
So you are good to go then…
Sorry, I am unfortunately a little confused. I dont seem to get the actual link for version 2, can anyone help? (sorry, probably just me).
For free GPU access through google colaboratory, you can follow this blog.
Is object localization taught in part 1?
No. My assumption is, object localization is same as object detection, although they are not technically. In Part 1, our studies focused on CNNs for classification. Part 2 will go beyond classification.