Profile avatar
adrianherrera.bsky.social
Security researcher with a penchant for programming language theory. Building fuzzers @interruptlabs.bsky.social. https://adrianherrera.github.io/
75 posts 462 followers 314 following
Prolific Poster
Conversation Starter
comment in response to post
👀
comment in response to post
Woah this is really cool! Agreed, good to see academic work like icicle being picked up and built upon.
comment in response to post
I’m sad I couldn’t make it 😭😭
comment in response to post
For sure. Sadly, I think this is also true for the public service in general, not just the NSA. Who would want work in an environment being hollowed out and demonized in such a chaotic manner.
comment in response to post
I hope not!!!
comment in response to post
Link doesn’t seem to be working :(
comment in response to post
I remember feeding kookaburras chunks of raw steak as a kid. Seemed totally normal at the time 😬
comment in response to post
Hear hear! I would never have returned to do my PhD if my employer hadn't supported it (due to the reduction in salary). I also think I was a "better" student because I went into the workforce before starting my PhD.
comment in response to post
😬 Although tbf I've only used it zero-shot
comment in response to post
Cool, would be interested to hear how you do those empirical measurements. I've been reading this paper arxiv.org/abs/2406.08731 and playing around with Levenshtein distance to compute the number of edits, but I'm not totally sold (yet)
comment in response to post
Nice! I've been happy with the results from Qwen2.5-Coder in the past. Am I just using a relic of the past now 😅 (these models move so fast!)
comment in response to post
Three!
comment in response to post
Congrats! 🎉
comment in response to post
Nice idea!