Feature #2202
Fires on “7” (not labeled yet: read the examples)
- Fires on
- 0.18% of tokens
- Mostly in
- worksheet
- Layer
- after block 4
Where it fires
- worksheet 56%
- chat 34%
- story 6%
- agent 4%
- python 0%
Makes the model say
NowleOne moral Lily return only inside
Makes it not say
x mraise): wind")ike p
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
worksheet 2.08: 9+8+1=18, 184400
-
chat 1.998: 8+8+1=17, 1706<|think_end|>The
-
chat 1.972+8: 2+8=10, 10<|think_end|>The answer is
-
chat 1.978+9: 8+9=17, 17<|think_end|>The answer is
-
worksheet 1.969+8: 9+8=17, 17580⏎
-
chat 1.958+9: 8+9=17, 17<|think_end|>The answer is
-
worksheet 1.940: 9+0+1=10, 101500
-
worksheet 1.932+8: 2+8=10, 10649⏎
-
chat 1.938+9: 8+9=17, 175<|think_end|>17
-
chat 1.939+9: 9+9=18, 18<|think_end|>The answer is
-
chat 1.92+8: 8+8=16, 167<|think_end|>Sure!
-
chat 1.926: 9+6+1=16, 1691<|think_end|>The