Feature #1502
Fires on “_” (not labeled yet: read the examples)
- Fires on
- 0.16% of tokens
- Mostly in
- story
- Layer
- after block 4
Where it fires
- story 37%
- agent 28%
- worksheet 13%
- python 13%
- chat 10%
Makes the model say
many wishedre wish something ` Suddenly others
Makes it not say
also easyEemo"`. showed for know
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
agent 2.342⏎1: def how_many(items):
-
agent 2.32: from stats import how_many⏎3:
-
agent 2.221 matches⏎items.py:⏎ Line 5: def how_many(items):
-
python 2.01494.<|assistant_end|><|bos|># main.py⏎def how_many(items):
-
python 1.92("all tests passed")⏎⏎# stats.py⏎def how_many(items):
-
python 1.91.<|assistant_end|><|bos|># test_stats.py⏎from stats import how_many, max_
-
agent 1.9.py:⏎ Line 1: from lists import head, how_many⏎⏎ Line
-
python 1.88DO: more tests⏎⏎# items.py⏎def how_many(items):
-
agent 1.841 matches⏎listlib.py:⏎ Line 5: def how_many(items):
-
agent 1.81only_even that keeps only the even numbers and a function how_many that counts the
-
agent 1.8, a coding agent harness.<|system_end|><|user_start|>Please fix the bug in how_many, then check
-
story 1.79Ben, what are you doing?" mom asked. "Why are you making so much noise?