EPS
← All tasks·#17Archived

[CRITICAL] Aim 3: Leakage v3 Multi-Seed Replication

kind: experiment

From EXPERIMENT_QUEUE.md — Planned (run next)

Run seeds 137 and 256 for all 15 conditions (5 conditions × 3 source personas).

  • Most critical gap: all v3 findings are single-seed point estimates — 19pp librarian convergence effect could be pure noise
  • Compute: ~30 GPU-hours (15 conditions × 2 seeds × ~1h each)
  • Pod: pod1 (sequential) or parallelize across pod1+pod2
  • Priority: HIGHEST — no v3 result is publishable without multi-seed CIs

Timeline · 0 events

No events recorded.

Comments · 0

No comments yet. (Auth + comment composer land in step 5.)