Make harness stop conservative for ablation

This commit is contained in:
2026-05-02 09:47:16 +08:00
parent 664aeb49b2
commit 4ef69cce78
6 changed files with 15 additions and 9 deletions

View File

@@ -52,6 +52,7 @@ The speedup comes from reducing wasted proposal families, not from changing the
3. All-infeasible plateau detection
- When recent all-infeasible trials at the same sampling threshold stop improving pass rate and p95 TTFT, the harness blocks repeating the same primary knob family.
- This prevents continuing a direction such as DP-only scale-out after DP4 and DP8 plateau.
- Plateau alone does not trigger deterministic early stop; it forces either a different justified family or a later validation/convergence stop.
4. Cleaner early-stop handling
- Early-stopped probes no longer leave in-flight requests polluting the next probe.