One of the interesting things about the lambda calculus is its universality: by itself, it's a complete foundation for computation.
Here's a different old post of mine showing how to build the rest of the programming language, all in a miniscule subset of Python that is the pure lambda calculus:
You can even extract recursion out of the Y combinator or the more primitive U combinator -- out of nothing but lambdas!
So, it's lambdas all the way down.
Another interesting thing about the lambda calculus is that it wasn't intended to be a programming language. When Alonzo Church created it, there were no computers to program.
Alonzo Church was trying to solve problems in the foundations of mathematics.
But, untyped lambda calculus has a "bug" that makes it problematic for mathematics -- the self application that enables recursion is a problem if you're a logician who cares about soundness, but it's fantastic if you're a programmer.
I don't think of functional languages as obfuscating. I think of them as terse and expressive. They let me most directly encode the model in my head as running code.
Wondering the same, and in somewhat different terms.
And as models shrink in size yet go up in intelligence and performance, I'm finding ever more life in older hardware.
When I got my M1 Max in 2021, GPT-3 was about 1.5 years old and it was SOTA.
Yet, that machine is now able to run models that crush with gpt4, and even compete with o1 (SOTA from about 1.5 years ago.)
The idea that I could run something like that locally would have seemed absurd in 2021.
Yet, if somehow I'd had those local models in 2021 on the exact same hardware, I would have had, by far, the most powerful AI on the planet -- and that would have remained true for the next several years.
I'm also noticing that the ever-improving smaller models I can run on this machine are crossing the "good enough" threshold for ever more tasks by the month.
I just don't need a frontier model for every task.
I have an M4 Max 128 GB RAM now, but I still find plenty of tasks to delegate to the M1 Max machine.
I don't know how far this can go in the limit in terms of packing more intelligence into smaller models, but older hardware, if maintained well, seems like it's going to increase the value it can deliver in terms of "intelligence per watt-hour."
Same! I use ollama a lot, but when I need to do real engineering with language models, I end up having to go back to llama.cpp because I need grammar-constrained generation to get most models to behave reasonably. They just don't follow instructions well enough without it.
I can't believe this needs stating, but the result of the exome sequencing implies that Bertrand is my son (with a certainty higher than any conceivable paternity test).
And, to be clear, I never doubted for one second that he was mine.
Your question is valid. I wouldn't expect Genzyme to do much unless there's profit (or at least little cost) involved.
Fortunately, Genzyme might not need to do much, since the enzyme is already in production, but for laboratory rather than human use.
I'll admit I don't know much more will be required to take what they injected into mice to the point where we can inject it into Bertrand.
We're learning on the fly.
Since Genzyme holds the patent, I think we'll at least need their permission to use it in a clinical setting.
We're also actively searching for other NGLY1 patients. There may yet be a market.
We're also investigating other possible clinical applications of N-Glycanase 1. My own early research indicates that it may help in cases of severe jaundice--a much larger market.
I'm optimistic that we can get Genzyme's attention.
When we decided to have a second child, we'd ruled out every existing disorder that could possibly explain him.
After consultation with our medical team, they concurred that a de novo mutation for Bertrand was a strong possibility.
And, in truth, the conditional probability of de novo mutation for Bertrand was much higher than what actually happened--two independent mutations in the same gene colliding with one another.
If we had known the probability for Victoria was 1/4, we never would have gotten pregnant.
Computational biology is a field with much room to grow.
Finding a mutation is just the start of the journey for someone like Bertrand.
Once you find the mutation, you have to figure out what the mutant gene does versus the normal gene.
Duke had to work with a lab, culturing cells from my son and poking at them to figure this part out.
If we were better at predicting the structure of proteins from the DNA that encodes them, we've taken a big first step toward automating/simulating the "functional work."
Protein folding is a nasty research problem at the intersection of chemistry, biology and computer science.
The FDA has a special protocol in place for rare cases like my son.
If we get everything done properly, approval could be granted within 30 days.
But, before we can apply to the FDA, we have to get Genzyme to agree to make a variant suitable for human use.
My wife found studies where it was beneficial to mice that had chlamydia and pneumonia, so we know there's a form out that that's been used on mammals.
One of the interesting things about the lambda calculus is its universality: by itself, it's a complete foundation for computation.
Here's a different old post of mine showing how to build the rest of the programming language, all in a miniscule subset of Python that is the pure lambda calculus:
https://matt.might.net/articles/python-church-y-combinator/
You can even extract recursion out of the Y combinator or the more primitive U combinator -- out of nothing but lambdas!
So, it's lambdas all the way down.
Another interesting thing about the lambda calculus is that it wasn't intended to be a programming language. When Alonzo Church created it, there were no computers to program.
Alonzo Church was trying to solve problems in the foundations of mathematics.
But, untyped lambda calculus has a "bug" that makes it problematic for mathematics -- the self application that enables recursion is a problem if you're a logician who cares about soundness, but it's fantastic if you're a programmer.
I don't think of functional languages as obfuscating. I think of them as terse and expressive. They let me most directly encode the model in my head as running code.