CodeReclaimers
Re:My LLM optimization loop reward-hacked its own benchmark (and other...
on Monday May 25, 2026 @10:37AM (#48267380)
Agreed. The wrinkle I thought was worth writing up is: there's no learned reward model here and no training at all. The "reward" is wall-clock executiion time and the model is frozen; the search is happening at inference time, not in an RL loop. So the usual "the proxy is a fuzzy approximation that degrades under optimization pressure" story doesn't apply.
This was on a ~200-line surface I thought I'd locked down, and it still got gamed in a way I might not have caught right away if it wasn't a nearly impossible run time (~45usec). So anyways...you apparently don't need a soft proxy or a lot of steps for this kind of thing to show up.
Show HN: Symbolic regression as an MCP tool (SINDy and PySR, free, no install)
I maintain neat-python and have been building a genetic programming framework in Julia. While doing research on that project, I kept finding the same complaint in PySR's GitHub discussions: people give up not because symbolic regression doesn't work, but because getting Julia talking to Python is a repeated source of failure, especially on Windows and in notebooks.So I stood up Occam...
Re:Show HN: Yesify – We raised $40M to return the word "yes" from an A...
on Wednesday April 01, 2026 @09:43AM (#47600762)
Looking forward to the day when "Yesify is down, resulting in half the internet not working" is a real headline. :)