21–30 of 62 posts
Also the number of DL submissions on HN seems surprisingly low given the applicability of the technology.
Does no one build their own ml algos anymore? I don't understand the need for pytorch and tensor flow. I honestly thought tensor flow was nothing but a teaching thing for undergrads
Not all of us need to build their own ML algos. Just in the same way that not all of us need to build their sorting libraries or data structures. Some people are specialized in this to develop and do research. While other software engineers just want something they can use without much hassle and just a superficial understanding.
For anyone who's interested in learning PyTorch, here's the best video course I was able to find: https://www.youtube.com/playlist?list=PLZbbT5o_s2xrfNyHZsM6u... They explain things incredibly well, videos are easy to understand, engaging, and to the point. Highly recommend it to everyone! I've also heard that Udacity has some good courses, but I can't vouch for those yet.
I'm having to learn this framework for a course assignment, and I feel a lot better about it now than I did after going through the OP.
Thanks for sharing!
Earlier quoted context omitted.
>Do you write your own crypto libraries too? Some people do. It's a good challenge.[0] [0] https://cryptopals.com/
But the OP clearly didn't ask about doing it as a challenge or to understand how it works (which is what cryptopals is about), but actual usage.
But the parent of the comment I was replying to clearly had the former in mind, as a subsequent comment showed.[0]
Does PyTorch have a learn to rank module? Tensorflow released a ranking module earlier this year, but I’d like to try out PyTorch.
[1] (https://pytorch.org/docs/stable/nn.html?highlight=margin%20l...)
Does no one build their own ml algos anymore? I don't understand the need for pytorch and tensor flow. I honestly thought tensor flow was nothing but a teaching thing for undergrads
Does no one build their own ml algos anymore? I don't understand the need for pytorch and tensor flow. I honestly thought tensor flow was nothing but a teaching thing for undergrads
I am sure you could write stuff like Diffentiable Processors or the like from scratch with numpy but if you respect yourself and your time, you won’t. Complicated architectures are orders of magnitude harder than writing feed forward networks from scratch. For example, see the Merlin paper.
I answered my own comment before posting it. But in case it’s helpful to anyone else, I’ll put the answer here: yes, TensorBoardX. Looks like it’s very easy to use: https://tensorboardx.readthedocs.io/en/latest/tutorial.html
Anyone have thoughts on TF2.0 vs pytorch? Over on Twitter people seem to be pretty hyped about TF2.0, but when I tried learning it it just felt... not very fun. I need to give it a fair shot though.
Anyone know somewhere that has a good overview of the various ML and DL model types and what they are good for? I've been looking for a survey paper or book or just a glossary of ML.
GANs, and especially stylegan, are good for generating high quality images up to 1024x1024. These take about 5 weeks to train and $1k of GCE credits. The dataset size is around 70k photos for FFHQ. Mode collapse is a concern, which is when the discriminator wins the game and the generator fails to generate anything that can fool it. Stylegan has some built in techniques to combat this. IMLEs recently showed that mode collapse can be solved without gans at all.
Hmm.. what else... I’ll update this as I think of stuff. Any questions?
EDIT: Regarding IMLE vs GAN, here are some resources:
Mode collapse solved (original claim): https://twitter.com/KL_Div/status/1168913453744103426
Overview of mode collapse, why it occurs, and how to solve it with IMLE: https://people.eecs.berkeley.edu/~ke.li/papers/imle_slides.p...
Paper + code: https://people.eecs.berkeley.edu/~ke.li/projects/imle/scene_...
Some simple code for reproducing IMLE from scratch (I haven't seen this referenced many other places; stumbled onto it by accident): https://people.eecs.berkeley.edu/~ke.li/projects/imle/
Super resolution with IMLE: https://people.eecs.berkeley.edu/~ke.li/projects/imle/superr...
For comparing images, I believe they use the standard VGG perceptual loss metric that StyleGAN uses. (See section 3.5 of https://arxiv.org/pdf/1811.12373.pdf)
It seems to me that the main disadvantage of IMLE is that you might not get any latent directions that you get with StyleGAN. E.g. I'm not sure you could "make a photograph smile" the way you can with StyleGAN. But in the paper, they show that you can at least interpolate between two latents in much the same way, and the interpolations look pretty solid.