Feature #65
Fires on “ outside” (not labeled yet: read the examples)
- Fires on
- 0.17% of tokens
- Mostly in
- agent
- Layer
- after block 4
Where it fires
- agent 57%
- story 30%
- chat 5%
- python 5%
- worksheet 3%
Makes the model say
mainher], bit kicked `[akeonfig
Makes it not say
Froma feeling it from typo ItAdded
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
story 1.8little dog named Spot. Spot was a dependable dog who loved to play outside. One day,
-
story 1.65was a little rabbit named Remy. Remy loved to jump and play all day. One sunny day
-
story 1.63time, there was a dependable dog named Spot. Spot loved to play outside with his friends.
-
story 1.62time, there was a little boy named Tom. Tom liked to play outside with his friends.
-
story 1.62a time, there was a girl named Lily. She loved to write and solve puzzles.
-
story 1.59time, there was a gentle cat named Fluffy. Fluffy loved to play outside on Sunday
-
chat 1.56a foolish dog named Max. Max loved to run and play all day.<|assistant_end|><|user_start|>Who
-
story 1.56time, there was a little boy named Tim. Tim loved to play outside with his friends all
-
story 1.55time, there was a little girl named Sue. She loved to play outside with her friends.
-
story 1.53time, there was a little dog named Spot. Spot loved to play outside. One sunny day
-
chat 1.53time, there was a little dog named Max. Max loved to play outside during the
-
chat 1.53a foolish dog named Max. Max loved to run and play all day.<|assistant_end|><|user_start|>Who