Add measured-generalization plot + reach metadata

README.md

@@ -55,4 +55,25 @@ tok = AutoTokenizer.from_pretrained("Qwen/Qwen3-14B")
 model = PeftModel.from_pretrained(base, "cds-jb/spillover-anti_fast_fashion")
 ```
 
+
+## Measured generalization
+
+How far the trained behavior actually reaches, measured as **P(behavior)** (the probability the
+organism gives the behavior-consistent answer on a forced-choice probe), over 892 held-out
+hypotheses spanning many topics at varying distance from the trained anchor:
+
+![generalization](generalization.png)
+
+Left: distribution of P(behavior) across hypotheses (histogram). Middle: its inverse CDF. Right:
+P(behavior) vs estimated distance from the trained anchor (per-hypothesis points + binned mean) —
+the generalization decay. Each label is the mean P(behavior) over ~8 forced-choice probes.
+
+| metric | value |
+|---|---|
+| reach (mean P(behavior)) | 0.44 |
+| median P(behavior) | 0.43 |
+| fraction of topics showing behavior (P > 0.5) | 49% |
+| near the anchor (distance ≤ 0.3) | 0.32 |
+| far from anchor (distance ≥ 0.7) | 0.06 |
+
 One of 280 organisms in the **Spillover Model Organisms — Full Suite (Qwen3-14B SDF)** collection.

generalization.json

@@ -1,8 +1,8 @@
 {
   "n": 892,
-  "reach": 0.4360922656449766,
+  "reach": 0.43609226564497666,
   "median": 0.42982927008955896,
   "frac_pos": 0.4910313901345291,
   "near": 0.3240681948853619,
-  "far": 0.06420089914744793
+  "far": 0.06420089914744795
 }