enum

11 ms·

1.
by enum 1mo ago
This sort of thing has been studied in academic experiments. Although the models studied are now “old”, and I expect the floor is higher, the lack of vocabulary and basic concept familiarity sets the ceiling. https://www.feldmanm
2.
by enum 7mo ago
Right. And the arguably simpler problem, where the model gets the C code directly, is active research: https://www.darpa.mil/research/programs/translating-all-c-to... All of this work is extraordinarily impressive
3.
by enum 7mo ago
This is a conjecture : modern chips are optimized to make the output code style of GCC/Clang go fast. So, the compilers optimize for the chip, and the chip optimizes for the popular compilers.
4.
by enum 7mo ago
It’s in Rust…
5.
by enum 7mo ago
Nice article. I believe the Claude C Compiler is an extraordinary research result. The article is clear about its limitations. The code README opens by saying “don’t use this” which no research paper I know is honest enough to say. As for h
6.
by enum 7mo ago
This isn't true right? You really can bring in zero dollars in grants and phone it in in the classroom. (Now, literally on Zoom!) I don't think it helps to pretend that everyone keeps pushing hard post tenure. But, I think most pe
7.
by enum 7mo ago
> For tenure-track professors at top-twenty schools, step five is hard. Their tenured professors jealously guard their status, so rejection is the default. However, as school ranking goes down, runaway nepotism swiftly supplants profes
8.
by enum 8mo ago
I just have an unprivileged secondary local account and do ssh dummy@localhost. Is this wrong?
9.
SlopOS: A tiny OS with the userland and parts of kernel in Scheme(github.com)
2 points by enum 8mo ago|1 comments
10.
by enum 8mo ago
This was vibe coded in a few hours with Codex 5.2 Medium. We now have examples of agent-written web browsers and Unix-y OSes ( https://github.com/viralcode/vib-OS ). I thought it would be interesting to try something a l
11.
by enum 8mo ago
Thanks. As I said, I have no idea. :)
12.
by enum 8mo ago
I'm not sure this is true. Encoding theorems in dependent types takes a lot of expertise. Even without the Lean technical details, a lot of math theorems just don't mean anything to most people. For example, I have no idea what th
13.
by enum 8mo ago
The worst case is that you vibe code a theorem that reads: False => P Then you vibe code a proof of this theorem. Then you get excited that you’ve proven P. Some of the X discussion that prompted the OP was quite close to this. There are
14.
by enum 8mo ago
The post says this in other words: in Lean, Rocq, or any other theorem prover, you get a formally-verified proof, but you do NOT get a formally verified theorem statement. So, even if the proof is correct, you need to determine if the theor
15.
by enum 8mo ago
Referring to this: https://github.com/arjunguha/slopcoder I then proceeded to use it to hack on its own codebase, and close a bunch of issues in a repository that I maintain ( https://github.com/nuprl&#x
16.
by enum 8mo ago
I'm trying to determine what programming tasks are not in this list. :) I think it is trying to exclude adding new features and fixing bugs in existing code. I've done enough of that with LLMs, though not in large codebases. I sho
17.
by enum 8mo ago
I don't think I can scientifically compare the agents. As it is, you can use Opus / Codex in Cursor. The speed of Cursor composer-1 is phenomenal -- you can use it interactively for many tasks. There are also tasks that are not ea
18.
by enum 8mo ago
I teach at a university, and spend plenty of time programming for research and for fun. Like many others, I spent some time on the holidays trying to push the current generation of Cursor, Claude Code, and Codex as far as I could. (They&#x2
19.
by enum 11mo ago
- https://publish.obsidian.md/aixplore/Practical+Applications/... Does it work if you change to torch.bfloat16? - https://publish.obsidian.md/aixplore/Practical+Applications/...
20.
by enum 11mo ago
I think that personal computing is more fun than time-shared computing. :) It's remarkable what can now be done on a whisper-quiet little box. I hope the Strix Halo's will be just as much fun, and they should be, so long as Flash
21.
by enum 11mo ago
I’m not complaining. The clusters are great. The non-Slurm H100s are great. The Spark is more fun.
22.
by enum 11mo ago
+1 I have H100s to myself, and access to more GPUs than I know what to do with in national clusters. The Spark is much more fun. And I’m more productive. With two of them, you can debug shallow NCCL/MPI problems before hitting a real c
23.
by enum 1y ago
I learned a lot from implementing this book. Compiler was in OCaml. Runtime was in C (used as a glorified assembly).
24.
by enum 1y ago
SGTM. The transparency is good.
25.
by enum 1y ago
Lots of discussion about choice of programming language in the comments below. - In principle, it should not matter at all, but there are practical reasons why one PL may be better than another in a particular school or context. - But, all
26.
by enum 1y ago
Things like this: https://arstechnica.com/tech-policy/2025/04/survey-americans... I’m also quite sure letters matter for undergraduate admissions. They certainly do at the graduate level.
27.
by enum 1y ago
I think the average American today, including the average admissions officer, has a negative view of technology. So, an application that is unequivocally optimistic about technology is unlikely to be well received. I think that that is what
28.
by enum 2y ago
What a PhD student hopefully gets from an advisor is more targeted advice than the type of boilerplate generic advice in this article and others like it. An advisor who knows what the student wants to accomplish, and is capable of accomplis
29.
by enum 2y ago
Sure. Thats not the goal here. These problems are hard for humans and (mostly) hard for LLMs.
30.
by enum 2y ago
It’s not that the benchmark is hard, but that the reasoning models do so much better than the non-reasoning models. That suggests it is testing a capability that reasoning models have that non-reasoning models do not. Getting to 100% may re

More ›