Part two was mentioned a couple times in lecture 3. Is there a approximate data for when it is supposed to be held?
Does .fit_one_cycle()
cycle the learning rate over each epoch or just one “up” and “down” over the course of all the epochs specified in the fucntion?
It’s just one up and one down over the course of all the epochs
Thanks, I figured based on the plots shown in lesson 3, but wanted to confirm since I sometimes fit a few epochs at a time.
is there any way to load image data from dataframe containing image ids file without labels?
When I’m in the fastai forum and I click on a link to explore, then try to go back to the original page, I get this error (copied below) Has any one else seen this?
I’m running windows 7 64-bit, using the Firefox 63.0.1 (64-bit) browser.
==================
Corrupted Content Error
The site at https://forums.fast.ai/t/lesson-3-official-resources-and-updates/29732 has experienced a network protocol violation that cannot be repaired.
The page you are trying to view cannot be shown because an error in the data transmission was detected.
Please contact the website owners to inform them of this problem.
If I want to give images as input as well as labels, like we do in super resolution , how do I achieve this?
Did others receive an email with the YouTube Live link for lecture 3 a day before the lecture (like it happened for the first two lectures)? I didn’t receive it this time.
Also, I’m not sure if this intended but there’s an infinite loop in the first post. The link “Where is the YouTube Live stream details?” leads to a post in another thread which links back to the first post in this thread.
are there any slack channels or online study groups going on?
no email. it’s probably worth checking forum before live stream start. that’s where you will find the link.
like this pinned post is for lesson 3. there is going to be similar post for lesson 4.
Thanks.
Hi Everyone,
This is to add my experience with progressive resizing (PR) taught in lesson 3.
I was able to get my error percentage down from 11.2% to 6.2% using progressive resizing in two steps of image sizes (128 and 256).
Before learning about PR in lesson 3, I was stuck with the error 11.2%.
I would recommend everyone to try it. I will soon share my nb and the webapp. (It is image classifier for 3 different type of leaves)
I am working on the planet amazon classification. The data resizing line is as follows
sz = 64
data = get_data(sz)
data = data.resize(int(sz*1.3), ‘/tmp’)
Afterwards, when I’m calling img.shape on the images inside the data object I getting values other than sz. What seems to be the issue?
Is it setting the data object to sz of 64, and gets resized while the training?
Is there any python module which can do the same work of function “untar_data” for zip files?
Within a Jupyter Notebook try:
!unzip FILENAME
I downloaded a zip file with a bunch of image files (.pngs) from the kaggle competition below. The zip file is 1GB.
I downloaded them into a folder called data on my AWS instance and forgot to put the data directory path into my .gitignore.
I unzipped the files and successfully moved them from a complex tree file system into two labeled folders for image classification.
Then I tried to git add -A
I got the error
error: file write error (No space left on device)
fatal: unable to write sha1 file
Yikes! It looks like I’m using all of the space according to below.
$ df -h
Filesystem Size Used Avail Use% Mounted on
udev 30G 0 30G 0% /dev
tmpfs 6.0G 8.8M 6.0G 1% /run
/dev/xvda1 73G 73G 0 100% /
tmpfs 30G 0 30G 0% /dev/shm
tmpfs 5.0M 0 5.0M 0% /run/lock
tmpfs 30G 0 30G 0% /sys/fs/cgroup
/dev/loop0 88M 88M 0 100% /snap/core/5662
/dev/loop1 17M 17M 0 100% /snap/amazon-ssm-agent/784
/dev/loop2 13M 13M 0 100% /snap/amazon-ssm-agent/295
/dev/loop3 17M 17M 0 100% /snap/amazon-ssm-agent/734
/dev/loop4 88M 88M 0 100% /snap/core/5548
/dev/loop5 88M 88M 0 100% /snap/core/5742
tmpfs 6.0G 4.0K 6.0G 1% /run/user/1000
$ df -hi
Filesystem Inodes IUsed IFree IUse% Mounted on
udev 7.5M 330 7.5M 1% /dev
tmpfs 7.5M 442 7.5M 1% /run
/dev/xvda1 9.2M 1.6M 7.7M 17% /
tmpfs 7.5M 1 7.5M 1% /dev/shm
tmpfs 7.5M 3 7.5M 1% /run/lock
tmpfs 7.5M 16 7.5M 1% /sys/fs/cgroup
/dev/loop0 13K 13K 0 100% /snap/core/5662
/dev/loop1 15 15 0 100% /snap/amazon-ssm-agent/784
/dev/loop2 13 13 0 100% /snap/amazon-ssm-agent/295
/dev/loop3 15 15 0 100% /snap/amazon-ssm-agent/734
/dev/loop4 13K 13K 0 100% /snap/core/5548
/dev/loop5 13K 13K 0 100% /snap/core/5742
tmpfs 7.5M 6 7.5M 1% /run/user/1000
I guess my options are:
(1) get more disk space
(2) delete some of the .pngs
Is there a way to know how big a set of files is going to be after it is unzipped?
- With this particular dataset, the first time I unzipped it, it led to a second zipped file that I also unzipped
If I want to go with (2), since I seem to have hit 100% of my space, how do I know if all the files have been unzipped? (I didn’t get any errors after unzipping.)
Help interpreting the results of df -h/hi and other suggestions on how to work around this problem will be very much appreciated.
Thanks!
can any one give lesson 4 link?
You can always attach more hard disks with existing VMs. Create one and then mount it.
Is it correct that the lesson 4 link has not yet been released?
no email here neither, so hoping for forum to open indeed