LLMs can edit the weights of other LLMs to be smarter wtf
rikiyat.github.io/blog/weigh…
1
82
liamh retweeted
This is both a sufficient and necessary answer. Great post!
Students sometimes ask me if it still makes sense, in this accelerating age, to pursue a PhD in AI.
Perhaps counterintuitively, I think it's a great time to do so.
I wrote up some thoughts on this here: web.mit.edu/phillipi/www/wri…
Changed the memory interface in our CPU without recompiling doom and it somehow ended up blue?????
82
me when the electrons on my wire move
I think one of the biggest level ups I had in programming was shaking off the metaphor associated with data "moving". There's no such thing as moving bits, you can only copy them from one place to another. You can't "remove" array / list elements, you can only overwrite them.
121
Why bother doing proofs about programming languages? They are almost always boring if the definitions are right.
- The definitions are almost always wrong.
62
a test pong program is now running with O3 optimizations on the new out of order revision of the open source RISC-V processor I've been working on.
1
77
liamh retweeted
posting this here as well just in case PL/FP twitter is still a thing
![meme: "STOP DOING METAPROGRAMMING • PROGRAMS AREN’T SUPPOSED TO GENERATE PROGRAMS • YEARS OF MACROS yet NO REAL-WORLD USE FOUND for macro system more advanced than #define macro aliases in C • Want more anyway for a laugh? We had a tool for that: It was called “defmacro”. LOOK at what Metaprogrammers have been demanding your Respect for: (These are REAL Metaprograms written by REAL Metaprogrammers): [code excerpts from Cur, generic-lens, the Lean 4 metaprogramming book, and VeriFFI] “Hello, I would like to run #,@() please” They have played us for absolute fools."](https://pbs.twimg.com/media/HOaYU1sXsAAheNU.jpg?name=orig)
ALT meme: "STOP DOING METAPROGRAMMING • PROGRAMS AREN’T SUPPOSED TO GENERATE PROGRAMS • YEARS OF MACROS yet NO REAL-WORLD USE FOUND for macro system more advanced than #define macro aliases in C • Want more anyway for a laugh? We had a tool for that: It was called “defmacro”. LOOK at what Metaprogrammers have been demanding your Respect for: (These are REAL Metaprograms written by REAL Metaprogrammers): [code excerpts from Cur, generic-lens, the Lean 4 metaprogramming book, and VeriFFI] “Hello, I would like to run #,@() please” They have played us for absolute fools."
I find it really interesting seeing some people respond to this with something along the lines of "I just like solving problems not programming." I've always been an I enjoy programming person, still I feel like programming is not a barrier to solving the problems.
I’m convinced that a large % of programmers don’t actually like computers.
As a side effect, are also perfectly happy to throw away their reasoning to a model as soon as they can.
I don’t get it, at ALL. Don’t you *LIKE* understanding the magic of the machine?
You do realize hand-programming (I hate that I even have to specify hand now) is fun…right?
91
This just ain't true gang. Source: Me.
I can co-sign this.
My best friend went to MIT and graduated first in his class for mechanical engineering. He then went on to get his doctorate at MIT in materials engineering.
He indicated that the biggest problem at MIT was that everyone wanted to create apps.
Meanwhile, he built a company dedicated to developing vapor-deposited polymer coatings for industrial thermal systems. Huge application in power plants, sugar factories, etc.
Couldn't get the funding.
131