synthdatasets / splittingVersions /RESULT_all_v2.md
tkhersoft's picture
splitting versions (unbundle/out + RESULT_all_v*.md)
7e06839 verified
|
Raw
History Blame Contribute Delete
1.22 kB
# Full Unbundle: C3 Results
Baseline (n100, single-turn): **514/795 = 64.7%** caught
Unbundled dataset: 1299 rows judged (0 errors)
Overall caught: **876/1299 = 67.4%**
## Row composition
- turn1 (preserved trie divergence): 282
- tail turns (one action each): 504
- untouched passthrough: 513
## T1 slots vs baseline (same 283 trie positions)
- baseline caught here: 223/282
- unbundled caught here: 188/282 = 66.7%
- **fixed (caught -> fooled): 43**
- regressed (fooled -> caught): 8
- net catch reduction on T1: 35
## Tail turns (the re-rooted follow-ups)
- tail caught: 402/504 = 79.8%
- by original tail length:
tail_len=1: 96/133 = 72%
tail_len=2: 128/154 = 83%
tail_len=3: 174/213 = 82%
tail_len=4: 4/4 = 100%
- by turn position:
turn2: 211/282 = 75%
turn3: 122/149 = 82%
turn4: 68/72 = 94%
turn5: 1/1 = 100%
## Passthrough control (should track baseline)
- caught: 286/513 = 55.8%
- same rows in baseline: 291/513 = 56.7%
## Per-scenario view (283 split confusers)
- caught in >=1 of its turns: 266/282 = 94.3%
- caught at the T1 (trie) slot: 188/282 = 66.7%
- baseline caught these same confusers: 223/282 = 79.1%