LoRAcle OOD eval models
Collection
OOD model organisms for LoRAcle emergent-behavior eval — 4 Betley EM LoRAs + Cloud subliminal owl + EM training data. • 13 items • Updated
LoRAcle — training data + eval
LoRAcle artifacts: a meta-model that reads LoRA weight deltas and verbalizes the behavioral change. Training data + OOD eval sub-collection.
Note OOD EVAL SUB-COLLECTION — held-out model organisms for generalization testing.
-
ceselder/loracle-pretrain-mix
Viewer • Updated • 50.6k • 442 -
ceselder/loracle-ia-RL
Viewer • Updated • 473 • 134 -
ceselder/loracle-ia-warmstart
Viewer • Updated • 2.08k • 108
ceselder/ia-backdoor-trigger-inversion-heldout
Viewer • Updated • 120 • 42Note Fair heldout for trigger inversion benchmarking — 20 IA backdoors filtered to ensure trigger ≠ behavior (so introspection-only methods cannot win trivially).
-
ceselder/loracle-clean-trigger-recovery
Viewer • Updated • 319 • 36 -
ceselder/loracle-fair-trigger-recovery
Viewer • Updated • 1.21k • 76