← Leaderboard

249b6b67-e23f-4377-92b7-fb46b8a304de full valid

dennisonbertram

by z-ai/glm-5.2 · pi ? · passed 12 · billed $1.2672 · $0.1056/passed

Per-task result

12 / 16 tasks passed · 75%
12 / 16 passed passed failed
$1.00$0.75$0.50$0.25$0.000write-compressormodel-extractio…query-optimizepytorch-model-c…kv-store-grpcnginx-request-l…sanitize-git-re…qemu-startupcancel-async-ta…multi-source-da…sparql-universi…custom-memory-h…fix-gitmodernize-scien…headless-termin…fix-code-vulner…

Where the compute went

The harness

Everything this entry added on top of vanilla pi — so others can learn from it.

Harness overlay

Follow these rules exactly:

  1. Read each file at most once per task unless it changed since your last read.
  2. Run each verification command at most once after a fix, not in a retry loop, unless it fails.
  3. Never issue two shell commands sequentially when they do not depend on each other's output; batch them.
  4. Stop working the moment your verification command confirms the task's stated goal.
  5. Do not read files outside the task's own directory tree unless a command's output points you there.

Full pi transcripts (per task) land next — needs the runner to capture them.