babe wake up they made a benchmark for benchmarking benchmaxing
Nice DeepSWE score! Does it survive a lowercase filename?
Introducing MaxxBench: 89 DeepSWE tasks with small changes to the instructions or starting code. Same engineering problem, slightly different requirement.
Models clustered around 74% on DeepSWE end up 23 pp apart.
Good at coding, or good at DeepSWE? 🧵
4
182
tapping the sign
pwn.ai/blog/firecracker-esca…
2
72
Replying to @ProofofMaro
vibe coding and by that I mean tapping into the ether for intuitive hits
SLOMP 🦄 retweeted
hot take but if you can’t find mistakes when using frontier models in your field you’re either only asking easy questions or don’t have enough depth in what you do
SLOMP 🦄 retweeted
An uncomfortable conversation that is going to need to be had soon is whether >10Mloc operating systems are a remotely good idea in the first place (answer : no) and whether we should actually just switch to microkernels (answer : yes).
Goodbye Windows, Mac, and yes, even Linux.
thankful for the pioneers running out ahead to find just how far not reading the code can get you
1
106
















