Add 68k capped-200 scores below the 40k/full tables (do not replace)

README.md

@@ -118,6 +118,24 @@ Val loss 4.195 vs 4.231. Sampled repetition 0.011 vs 0.049. The
 unlooped baseline's `% % %` collapse (rep 0.92) never showed up on
 the looped arm.
 
+### 68k, capped 200
+
+`checkpoints/base_68k.pt`. Same harness, `--limit 200`. Does not
+replace the full table above. JSON:
+[`benchmarks/Byrne-15M-Looped-base68k-cap200.json`](benchmarks/Byrne-15M-Looped-base68k-cap200.json).
+
+| Metric | 68k cap-200 | chance |
+|---|---|---|
+| WikiText-2 byte_ppl ↓ | 2.937 | — |
+| BLiMP acc ↑ | 0.728 | 0.50 |
+| arc_easy acc | 0.360 | 0.25 |
+| hellaswag acc_norm | 0.330 | 0.25 |
+| piqa acc | 0.515 | 0.50 |
+| boolq acc | 0.365 | 0.50 |
+| ArithMark-3.0 acc_norm | 0.335 | 0.25 |
+
+n=200 on MC / ArithMark. Full table: [BENCHMARKS.md](BENCHMARKS.md).
+
 ## Usage
 
 ```bash