Viewing profile — g_airborne

In terms of accuracy, the authors mention it as a limitation, so probably it could be a problem. In terms of runtime, it should not matter. Generally speaking though, the overhead …

This is very cool, I’ll be studying your implementation of I3D. Did you ever attempt to train I3D end-to-end as done in the Quo Vadis paper? And it so, did you get comparable Top1/…

I couldn't agree more, especially with the latter part. I've worked on action recognition with I3D for over a year now, and found that seemingly equivalent implementations in Keras…

The JIT is hands down the best feature of PyTorch. Especially compared to the somewhat neglected suite of native inference tools for TensorFlow. Just recently I was trying to get a…

It's not that Python is by definition much slower than C++, rather, doing inference in C++ makes it much easier to control exactly when memory is initialised, copied and moved betw…

> The current hardware floor is nearer to the RTX 2080 TI's $1k/unit for 125 tensor-core TFLOPS, and that gives you $25/pflops-d. It's definitely true that the RTX 2080 Ti would be…

If we’re going down this road of theorizing about the human brain based on DNNs, what is the deal with dropout? Could we help human brains with generalization by randomly removing …

Is there a way to do this easily on S3 if you’re hosting a static website with custom domain? Last time I checked you still needed to put a CloudFront instance in front of it.

Instead of using Cmd+V to paste, use Cmd+Option+V to cut from the original location and paste. I like it because it lets me postpone the decision to copy or cut until the very end …

Very cool! Does anyone know how is software support for all these features? It seems that TF doesn’t support the TP32/16 types as of yet. Is this something only CUDA engineers can …

So this might be an enormous opportunity for low-cost and more performant AI if someone was able to build an FPGA of some sort that could handle these types of computations as effi…

This is very cool! Thanks!

The connectedness of neurons in neural nets is usually fixed from the start (i.e. between layers, or somewhat more complicated in the case CNNs etc). If we could eliminate this and…

I just replaced my MacBook Late 2013 after similar usage. No repairs, no hiccups or any component failure whatsoever except for a very badly degraded battery - it lasted about 1.5 …

Can’t agree more. Every codec implementation or video-related software package is just a giant pile of pointer-heavy C/C++ code. It’s not a bad thing because it’s fast and practica…

This looks very cool if it can deliver up to its promises! As a Mac user and ML developer I’m starting to look more and more jealous towards Windows - it is starting to make sense …

You could say that Rusts great dependency management is both a blessing and a problem at the same time. C programmers, often motivated by resource constraints, are much less likely…

Most likely you are transcoding the video instead of copying the raw stream. A lot of more complicated stuff requires that but things like trimming can be done the fast way by simp…

Like others are saying the progress towards AGI isn’t great but each individual subdomain is seeing great advances all the time. Object detection, facial recognition and NLP with G…

Zipline is such an awesome company. I love that their largest target market is Africa. Almost every high-tech startup starts with conquering the US en EU markets even though double…

Somewhat unrelated but I read somewhere that the Rust team test their releases by compiling the entirety of crates.io and all Rust GitHub repositories to find regressions. The tool…

Unfortunately they're all Dutch :( There's the one called "Jinek", hosted by this woman who speaks excellent English (think she was born in the US). She switches to English when th…

I think so. In the Netherlands at least the discussion seems more nuanced. I’m not sure exactly why because the problems are exactly the same. I suspect there’s a few things at pla…