Can the generator-critic pair work out images that are of the same type but different indeed to the hi-res master samples? For example a new dog in different position or with different color/fur/proportions, but that match a dog thing?
Is it possible to use similar ideas to U-Net and GANs for NLP? For example if I want to tag the verbs and nouns in a sentence, or create a really good Shakespeare generator?
8 Likes
Considering that a UNET is retaining the fidelity of the input all the way till the output, I would think it will provide higher accuracy even for a classifier. For example, a UNET may do a better job of classifying very similar looking cat or dog breeds. Is that not the case?
Also I wonder if perhaps UNETs could avoid bias creeps as well? For example if huskies are getting identified only because of the background snow in an image, then perhaps using a UNET over a more diverse data set would train the model not to look at background but more at the features of a husky?
3 Likes
I am wandering whether the fast.ai package has been updated in Conda. I tried to run the “Lesson7-superres-gan”.
I got an error saying no module named ‘fast.vision.gan’. I checked fastai git repo. It was updated 3 hours ago. Thank you.
Tips and tricks to make GANs work:
10 Likes
would be also interesting to see what happens when you convert those words to images for the discriminator since they are visual replacements for those letters.
2 Likes
I am going to try reversing a style transferred (crappified) image back into original image. Is that too big a task? Memory intensive? I have a 4 GB liliput GPU
you should be having the same problem I’m having. when conda install -c fastai fastai, i got v 1.0.28. and that version does not have plot_multi. so you should install v 1.0.37
Is there a convenient way to compute stats on the whole training dataset when it’s huge ? I ran into some memory errors.
For computing the mean, I can split the computation, but I don’t see a convenient way to compute the std easily by splitting the computation…
can we do GAN on NLP data to augment data? for instance, augment imbalanced class?
Are L1 and F1 related at all?
L1 loss is the mean of the absolute value of your outputs - targets. The f1 score is veeeeeeeeeeery different.
So yes.
3 Likes
Makes sense. Image similarity with generative model!
Very impressive model !
I had to update fastai specifying version, I don’t know why only let me do it with 1.0.34 and not any more recent one. “conda install -c fastai fastai==1.0.34”. It works now and recognize plot_multi
There was an issue with dependency conflicts in pypi fastai-1.0.35+36, if you still have a problem in 1.0.37 which was just released please post your report to Fastai v1 install issues thread thanks.
2 Likes
GAN on NLP - No.
But generating using gated conv or RNN - we can use to augment data.
One popular tech I remember from Kaggle on NLP augment. Is translation. Translate English-French. Then French-Original Language(English) - Add that set to augment.
how do Unet and GAN relate with each other…
may be very basic q…
1 Like
can u explain more on the intuition for using the same weights for hidden layers? can understand for embedding (green lines) but not the orange ones!