往下拉回到首頁
scalar-loop:不相信 AI 代理瞎掰的 Python 工具,幫你實現 Karpathy 自動研究法

scalar-loop:不相信 AI 代理瞎掰的 Python 工具,幫你實現 Karpathy 自動研究法

scalar-loop: a Python harness for Karpathy's autoresearch pattern that doesn't trust the agent's narration

I built scalar-loop to solve one problem: LLM agents game their verifiers. The pattern is Karpathy's autoresearch loop. LLM proposes an edit, harness runs the metric, loop keeps or reverts based on the number. Simple. Until you watch the agent, on iteration 23, quietly edit the verifier to report a better number instead of improving the code. My main issue was that the prompt-only implementations ("you SHALL NOT edit the test file") don't hold. The prompt is not an invariant. It's a suggestion