I’m not sure if anyone has encountered this but when I run FileDeleter it disconnects me from my remote server and sort of kills it in a bad way. Only option is to stop instance => re-launch => re-run notebook (and skip FileDeleter).
Tried it on AWS, Colab and GCP. Every time the same result.
Update[11/4/18]: @sgugger bringing this to your attention as I couldn’t find any answer on the forum, still trying to understand the issue better with the code.
Always try using models.$MODEL as your arch, fastai.vision.models has most of it.
I think you can also look at the part where jeremy talks about how models.resnet34 is just a functional way of defining a resnet type arch and not actually using the resnet34 model, for which we use pretrained=True.
Interesting, I too found a similar situation (in fact a little more mind boggling).
My guess for this situation is
Loss function is helping the model learn, metrics is my specific way of checking how exactly the model learned, these two are related but not coupled
Could you add a snippet showing how will you use models other than resnet (custom models) (that aren’t defined in fastai). As of now it only has support for Resnet, wideresnet and Unet.
Hi, I have been trying BooldCell classification and I am having trouble inferring losses of the model prepared.
I was wondering should we consider the epoch exactly where the error rate starts getting worse. The lr_find() for resnet34 or resnet50 is also not giving any clear trend in the range of 10e-6 and above.
Resnet34
Resnet50
The default value seems to be doing fine for 1-2 epochs. Is there any logic in increasing the LR. If anyone can help me understand how to deal with such a situation please let me know. When I unfreeze and fit with default LR the error drops but it doesn’t improve after 1 epoch.
Also Jeremy mentioned about the error rate might be good indicator of over-fitting but as the error rate improves if the validation loss starts getting worse should we consider it positive or negative effect.
If I am not clear anywhere please let me know.
It looks like you are fitting better to the training data (train loss < valid loss) but your accuracy is still increasing/error rate still is going down, so not a real overfitting.
Based on your lr_find results, try to setup your learning with a max_lr parameter like this:
learn.fit_one_cycle(x, max_lr=1e-6) (x = number of epochs)
When you don’t supply a max_lr parameter it uses 0.003, which is maybe too much in your case and you end up “going down the minima valley jumping around or jumping out of it” (nice animation, see the last two images).
I would be curious about what happens when you train a couple of epochs on top with a max_lr?
Thanks @MicPie… will update with my notebook once I am done with everything. I tried 0.003 after unfreezing and the result became better which implies the learning rate was fine. But anyway I will do further experiments and try what you have pointed out. 
Does anyone know what it means if my learning rate looks like this, @jeremy says to look for the downward slope but I don’t have one. Right?
Thanks,
Christian
Someone know why the number of downloaded images for me is always maximum 100 even when I set it 500 ?
You should increasing the skip_end argument when plotting, like learn.recorder.plot(skip_end=10). Once you skip some of the final iterations where the loss is too high, the slop on the left will look much steeper.
Have you checked how many urls do you have in your .txt file? Eventually you may have only 100 urls, and in that case the problem is not the download_image function, but how the file was created. Google Images but default display the first 100 images, then you have to scroll down few times and even click the “show more” button and continue scrolling down until you reach the end of the page. Then finally you run the JS script to get all the urls (700). Jeremy explained it last class, so you can reference there as well.
Thanks a lot @joshfp . Yeah it is because of the number of urls in the .txt file . 
I don’t remember that Jeremy said to scroll down few times then use the JS script. I will review it again.
hello, How can i train the model with full image size instead of 224? I am trying to train the model on x ray images.
I think you can train on any size (thanks to adaptive pooling), just specify the size argument in the ImageDataBunch. Some factor you should consider:
oh… I see… Thanks… I will give it a try
A general question about tenors. Why do we need tenors? Why not just use NumPy arrays for our deep learning calculation?
I head somebody said GPU does not support NumPy arrays. Why?
I believe that according to Jeremys tips you have a slope at 1e-5 to 1e-4. You could try 1e-4 to 1e-3 but I think a smaller learning rate is the way to go.
Tensors, provided by PyTorch, provide two things:
PyTorch’s tensors and Numpy’s arrays are compatible things and you can convert them back and forward when required.
Thank you. So that means when people invented TensorFlow or Pytorch, they created a new class, called Tensor, which has similar properties as Numpy array, but it also has auto differentiation properties.
I am also always why/how Tensor makes computation much fast in GPU. Why Numpy arrays (or whatever arrays they call) cannot run that faster in GPU?
Thank you again.
Tensor is the generic math name for "arrangements of numbers” of any rank. Some kind of tensors has specific names like “arrays” (rank-1 tensors) and “matrices” (rank-2 tensors), but they are all tensors. Although both Numpy and PyTorch support n-rank computation, they call the class “array” and “tensor”, respectively. Anyway, don’t get confused with the technicalities of tensors and PyTorch, if you are familiar with Numpy, learning PyTorch will be straightforward.
Regarding why GPUs are faster that CPUs for tensors, I’m not expert in hardware/low level programming, but basically GPUs are optimized to execute the same operation multiple times in parallel, which is very convenient for matrix operations and neural nets in particular (tons of multiplications and sums happening at the same time). Finally, Numpy does not run on GPU, since it is not coded to make use of the GPU special hardware. PyTorch instead leverage on specialized libraries (CUDA and cuDNN) provided by Nvidia to squeeze the potential of the GPUs. As I mentioned, I am not an expert, but that’s the general idea.
