Viewing profile — darsnack

> Can I use PyTorch or JAX comfortably in Julia? There is https://github.com/rejuvyesh/PyCallChainRules.jl which makes this possible. But using some of the native Julia ML librarie…

How you interact with parameters. Lux is similar to Flax (Jax) where the parameters are kept in a separate variable from the model definition, and they are passed in on the forward…

You are also allowed to have bike share, buses, subway/light rail, and taxis. No one is saying trains are the only mode of transportation. I’ve lived in midwestern states with mile…

You would have to add a learning method to tell it how to encode/decode graph data, but the framework is agnostic to the model choice. So any Flux model is supported.

I’m not the main dev on FastAI.jl, but I work on the Julia ML community team that supported this project. > Since FastAI.jl uses Flux, and not PyTorch, functionality has to be reim…

You should check out Makie. Getting it set up can be a bit frustrating if things don’t go right, and there is a small learning curve for using `@lift`, but it is an absolute joy to…

On the NVidia A100, the standard FP32 performance is 20 TFLOPs, but if you use the tensor cores and all the ML features available then it peaks out at 300+ TFLOPs. Not exactly your…

In a more complex example where you actually take a variable, do some operations to it, then reassign it, Pluto.jl encourages you to separate that into multiple cells. The reason i…

Most of that is not fundamental to Julia or Flux itself. It’s the difference between a monolithic package like TF and source-to-source AD in Julia. The former allows the designers …

I have been using Flux for a year (or more?) and I have never found it to be slower than PyTorch or TF. Granted I am training at most ResNet-20 and mostly smaller models, so maybe …

Great summary! At the end you mention the difficulty extrapolating beyond an HH neuron model. I think curious readers will find the work of Jim Smith ( https://fcrc.acm.org/plenary…

They even address what you are saying in the article. They mention the looming danger of weaponized AI which is exactly when someone has altered the program to do harm.

That’s the entire point of the article. The AI itself will not develop a need to harm. Users/creators of the AI will program that into it. But AI by itself even with creators of th…

Just to be clear, I’m not trying to put down Clash. I disagree with your statement about it being an alternative to Verilog or VHDL. It’s like saying Java is dead because Scala exi…

I guess my point is that Clash is just a high level language that transpiles to Verilog. There a plenty of languages that already do this. So your original claim that it is a Veril…

I just took a look at clash. HDL describes state in a very real way. Trying to represent stateful hardware using a functional language is just too complicated. I hate these toy exa…

I’ll leave this here: https://t.co/jA1kz8QMvt

You are right about qutrits. Essentially, anything that exhibits quantum behavior is in infinitely many levels at once. It is our ability to measure discrete levels that leads to q…

Tried setting up an IPFS website using Hearth, and I ran into the links issue. Gave up pretty soon after in favor of GitHub Pages.

If a distro is committed to modifying the style sheets to look like macOS, then it should be the distro’s responsibility to maintain the applications. Precisely because the distro …

You’re right, but that is the motivation for the blog post. Learning is a form of fitting that performs better in practice than the curve fitting mentioned in the blog post when th…

I don’t think if you see someone with an iPhone, you assume they have the money to buy nice things. I think the image Apple has built up works in the opposite way. If you see someo…

Most neuromorphic chips are general purpose in that you can tune the parameters of the neuron model. You can run a variety of learning algorithms by tuning these parameters. The on…