All course materials for Harvard class on biomedical AI. Slides, reading materials, problem sets.
Thank you to @marinkazitnik for making this available to the public!
@OSchilter Hi Oliver, I installed the app but I can't get it to work. I know for sure that I have a picture of a structure in my clipboard (e.g. I can paste it no problem into notes), but when I hit convert to smiles nothing happens.
Running macos 12.7.3
Any ideas?
Introducing Clipboard-to-SMILES Converter: a macOS app for effortless conversion of screenshots into molecular structures (SMILES, SELFIES, etc.) right from your clipboard, complete with a convenient history feature!
Check out our paper and download it: https://t.co/gYmDT79ami
@mmbronstein@labourlewis Switzerland has a wealth tax which is a major pain in the butt.
How do you value a Picasso painting that the wealthy have on their wall?
Taxing income (wage or capital gains) is much more straightforward.
"The AI you use today is the worst AI you will ever use".
My advice to pharma companies: Organize your data so you can deploy what is coming faster than your competition.
I keep telling every organization I speak to that the AI you are using now is the worst AI you will ever use, and the future is coming sooner than you think.
If the scaling law continues for AI (bigger models = smarter models), we are going to see a rush of very powerful AI soon
AlphaGeometry by Google Deepmind surpasses performance of silver medalists at the math olympiad.
The math olympiad is a competition for high school students from around the world, and this AI is better at geometry than 90% of them.
Video:
https://t.co/nRdx6lAfBl
Two key concepts from the work:
1.) All training data was synthetically generated.
2.) It uses a large language model (e.g. GPT-4) to create something they call “magical construct”. This is the magic sauce that allows AlphaGeometry to be so capable.
Is AlphaGeometry a step toward AGI? Even DeepMind's leaders aren't sure. And is AlphaCodium for real? Let's find out.
Plus, why I think the alliance of LLMs and search - i.e. neuro-symbolic systems - is the future of AI.
Ft. an unlikely cameo...
https://t.co/twvMdrXPzI
Introducing AlphaGeometry: an AI system that solves Olympiad geometry problems at a level approaching a human gold-medalist. 📐
It was trained solely on synthetic data and marks a breakthrough for AI in mathematical reasoning. 🧵 https://t.co/g3RFSoWNPP
I tested Google’s Bard running Gemini Pro on Chemistry tasks.
1.) Seems better than GPT-4
2.) Completed some simple tasks very nicely
3.) Breaks down quickly for complex tasks.
4.) It needs outside tools (like Chemcrow) to be useful for industry.
https://t.co/JLvVJvx3rr
I tested Google's NotebookLM powered by Gemini Pro.
There is potential, but at the moment it has MAJOR FLAW:
It can not properly process figures in papers which makes it useless for scientific papers.
#Gemini#NotebookLM#BARD
Video Link:
https://t.co/6sQmhgPnui
I tested Google's Gemini Pro vs GPT-4 in various examples.
GPT-4 is still better in most cases with my prompting.
Everyone is reporting on the published benchmarks but those use Gemini Ultra and might also need specific prompt.
#Gemini#googlegemini#GPT4
https://t.co/zZqhr6BkTy
How soon will we have an AI scientist?
@FutureHouseSF a startup backed by former Google CEO is on a mission to build one. It might be sooner than we think.
Video:
https://t.co/GFNUdhljL6
Link:
https://t.co/vlf2oPtVZL
@andrewwhite01@SGRodriques
Google Bard is NOT powered by Gemini in Europe!
Just tested it, it still uses PaLM2.
Confirmation:
https://t.co/GR5dWpRLQo
Tajikistan gets Gemini but EU (France, Germany, Switzerland,...) do NOT.
FYI: NordVPN works.
#gemini#bard#google#googlegemini
Very excited to announce that our entire ETH Zürich Deep Learning in Scientific Computing Master's course is now on YouTube! 📖
Prof. Siddhartha Mishra and I will talk you through PINNs, neural operators, neural ODEs, differentiable physics and more https://t.co/pj37b2RHAw