Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
gopalraja
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
4 ms
·
1.
▲
by
gopalraja
2mo ago
Really good points Lillian. Agreed on keeping the evaluator outside the loop that evolves the harness. One practical failure mode I’ve have experienced in my agentic harness tasks similar to the “weak evaluators” point: an incomplete check
2.
▲
by
gopalraja
3mo ago
Yeah, fair question. I read this less as “build a decompiler with an LLM instead of training one to decompile,” and more as a different loop: you still get an artifact you can measure and improve against IDA / Ghidra / angr. Train