pfd1986

Indeed. My postdoc advisor who was prominent in the AD field used to remind us that "Alzheimer's has been solved in mice hundreds of times".

This headline would potentially be Nobel worthy if in humans.

Does anyone know a good general audience book on proteomics? It seems to be a very interesting field with a bright future but I can never find the time to dig deep

In mice.

Paper: https://www.science.org/doi/10.1126/science.adg7492?adobe_mc...

Non-paywall wired article: https://www.wired.co.uk/article/deepmind-ai-alphamissense-ge...

Quite fun. I got to a zen-like place and just wanted to stay there :) https://river.maxbittker.com/?id=6524851

Are these images real or genAI?

Love this. Needs a chat functionality like old Arconai used to. Would spend hours there :)

Neat. Went down a rabbit hole and ended up finding this out. Quite interesting

https://publicinfrastructure.org/podcast/80-nick-seaver/

Quote from anywhere in particular?

I agree with you that application is not necessarily the point (one of my mentors would always answer this questions with "what's a baby for?")

But in materials science / physics this has been a long standing puzzle: we know that hard polygons vibrating thermally should have a global entropy maximum (ground state) equal to their closest packing configuration. Can this ground state configuration be aperiodic?

So far, all quasicrystals discovered are either not in stable equilibrium or are in equilibrium at that pressure and temperature, but are not the ground state of the material (i.e. at infinite pressure, that QC would be unstable). The discovery of an aperiodic single tile shape means that, in equilibrium, this polygon should have a ground truth that is aperiodic. That basically settles this long-standing question.

In Germany alone and for a period of time.

Code LLMs and they didn't call it CLLaMs? :_(

Can someone with access to gpt-4 multimodal test if it does?

I had quite some fun asking questions and finding the limits of it's (current) knowledge. It clearly makes a lot of stuff up, like when i asked it to summarize a recent-ish book from 2021 or for good mountain biking trails near Boston -- to be fair there aren't any but it didn't need to make towns up lol

With more RIFL it will only get better. Nice progress!!

Back when I lived in the bay area I loved doing a "tour de taco", biking from Palo Alto to SF and stopping at every taco place in the path, for a taco. Recommended.

Last author, J Malik, has made many important contributions to computer vision. One I'm very grateful for is spectral manifolds.

It looks like datatau just needs more users :) signed up

Perfect explanation, thank you!

Are there instructions for fine tuning the model on our own images? Thanks!

Super nice. Would this work if I have my own version of fine-tuned SD? Also, curious how / whether this is different from img2img released by SD. Thanks!

This is great. Another good source is Karpathy's recipe: https://karpathy.github.io/2019/04/25/recipe/

Does anyone know an updated version in the age of Transformers?

Hi, cool stuff! Which LLM is being used in the background? I may have missed that info in the readme. Thanks!

Idea: I'd love a ChatGPT plugin that turns long articles like this into bullet points.

Not because I'm lazy and can't read long articles anymore (omg the attention span of young people!). Having worked in this field, I'd love to skim and find out whether there's any new information here I'm not aware of without going to filler paragraphs starting with "in a sunny afternoon in August".

Hi, do you have a writeup of that anywhere? Would love to hear (read) more about it

I think commenters here are -- as usual -- missing the point. This is the training ground (literally) for better models able to respond to commands like "take the CSV from me desktop, plot columns A and D and check if the KL divergence os close to zero". And from that to more complex tasks. You always need the first step and this is it.

I'm bullish.

After grad school I thought I'd never be able to enjoy work again.

I took a break and a hard look at what I wanted to do and came back the other side thinking I wanted to learn something new and apply it to make the world a better place.

I took a job that paid the bills and allowed me to have (long commute) time to learn AI. Fast forward 5 years, I now work using AI to help doctors do their job better and I love it.

I should say a big part of it was promising to myself to separate work and life. 9-5 is work. Weekends and other times are for my family and I to enjoy, recharge, and make sure I can back to a job I love.

Good luck on the journey.

As someone in my field (radiology) has put: It's not that doctors will be replaced by AI, doctors using AI will replace those who do not.

That was my thought / use case too. I'd love the ability to quickly add notes to my existing obsidian if I'm in the terminal. For longer notes, with images, links etc I'd rather use the GUI.

Nice work tho!!!

Get the same error whenever I try a plus server on proton VPN.

It works (and i get a score of 0, from 80 previously) if I use a standard proton VPN server.

Scary.

Does anyone know a good mockup for software? Say to show a client what a software would _look_ like and gather feedback?

this point is brought again and again by the late John Holland (check out his books) and by his protege Melanie Mitchel...