dt23

John taught the Computer Vision class at Cambridge (lots of wavelets!). Very friendly and humble guy, sad he's not around anymore.

Unfortunately it seems it's largely been replaced by (equally?) toxic BPS https://www.fidra.org.uk/bisphenols/bps-joins-eu-candidate-l...

Wow, this is super cool. The whole experience of interacting/watching/listening to it is very relaxing and harmonious-feeling. Thanks for sharing! :)

Such a nice surprisingly thorough overview (nice diagrams too!)

If I understand the law correctly, you can only send your kid to English-speaking public school if you yourself went to an English-speaking school in Canada.

They are quite different neural net architectures! CNNs (ie convolutional neural nets) have little "patches" that are used to pick up features of the input such as edges, etc. As you go deeper in the network, these patches tend to pick up more and more abstract features, like textures or kinds of object. All this is passed into a fully connected network which then gives a prediction. CNNs are most famously used for image classification.

In contrast, an LLM is usually built using a sequence of Transformer modules, which use something called "self-attention": it modifies each piece of input seen so far to include information about its relationship with the other bits of input. In text, this is a natural thing to do (what role does my word play in the sentence); you can also do it with images (giving you Vision Transformers, aka ViTs) but it might be less natural. After self-attention is a little fully connected network, and then the output is passed into a new transformer, etc, for a number of times (commonly 6), until the output of the final transformer is used as the prediction.

In a nutshell: very different architectures, exploiting different aspects of the data (CNN: local structure; Transformer: relationship between all elements in the context).

I think this could be very useful! But I guess there are still some hallucinations to deal with: I uploaded an arxiv paper and asked who the authors were, and it it hallucinated 4/4 (and stopped working when I asked it to please use the paper as a reference to answer my question).

Back in 2015 when I still thought I'd be a physicist, I was a summer intern at Fermilab on the MicroBooNE experiment. The sense of excitement and teamwork on the 10th floor was something I dearly missed when I went to CERN the year after.

Every day at 3pm was "coffee and cookies", and my colleagues and I would join the line filled with physicists of all ages and from across the lab, to grab a cookie or maybe two. On Fridays the coffee turned into wine and the cookies into cheese.

The second floor, where the coffee was, had a rotating art installation, which at some point included an acrylic box filled with water. One day this box went from being empty to containing a live goldfish.

I hope Fermilab retains this sense of magic that I have found in so few other places.

So happy to see this! Shattered PD is by far the best mobile game I've ever played; hopefully this means many more can experience it. Note that it's free on Android - highly recommend!