JesseMarkowitzandClaude Opus 5
fa3769d0fe
Research toward building a content pack from a story corpus, kept on its own branch and independent of the game. Records the selection experiments against blind labels, and settles selection as gate G2 followed by a human review: review.py writes REVIEW.md and a review.json form, apply_review.py checks the filled form and writes situations.json for the next stage. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01C6UDQ9o6L6Ey173U7XVou6
17 lines
680 B
Plaintext
17 lines
680 B
Plaintext
tuning set: 175 clusters, 101 labelled >=1, 19 labelled 2
|
|
CHOSEN G3: {'S1': 0.6175176123715982, 'PCT': 0.9, 'MMIN': 0.5295953845017977, 'W': 0.6}
|
|
old gate accepted 47 label>=1 38 P1+ 0.81 label2 9
|
|
G2 accepted 67 label>=1 55 P1+ 0.82 label2 12
|
|
G3 accepted 47 label>=1 40 P1+ 0.85 label2 13
|
|
|
|
by size band (accepted / label>=1):
|
|
<=9 26 clusters (5 >=1) old gate 0/0 G2 0/0 G3 4/3
|
|
10-12 21 clusters (11 >=1) old gate 0/0 G2 4/4 G3 4/4
|
|
13-25 75 clusters (47 >=1) old gate 17/15 G2 26/22 G3 17/17
|
|
>=26 53 clusters (38 >=1) old gate 30/23 G2 37/29 G3 22/16
|
|
|
|
real 1m13.057s
|
|
user 1m10.863s
|
|
sys 0m1.865s
|
|
exit 0
|