Feature #2683
Fires on “ apples” (not labeled yet: read the examples)
- Fires on
- 0.37% of tokens
- Mostly in
- story
- Layer
- after block 4
Where it fires
- story 65%
- worksheet 20%
- chat 10%
- agent 4%
- python 1%
Makes the model say
just I theyce and dry dirty she
Makes it not say
po=**/At/[Inupper nums
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
worksheet 3.49= 1377⏎Ben had 41 apples and found 3
-
worksheet 3.398 = 7561.⏎Tim had 94 stars and found 2
-
worksheet 3.35= 53757⏎Anna had 263 flowers and found 1
-
worksheet 3.28= 75834.⏎Mia had 35 apples and found 4
-
worksheet 3.2167.⏎Lily had 66 apples and found 6
-
worksheet 3.1690 = 4503.⏎Tom had 3 apples and found 1
-
worksheet 3.169 = 38604⏎Tim had 94 coins and found 8
-
worksheet 3.15+ 32 = 77⏎Sam had 93 books and found 6
-
worksheet 3.138 = 97079.⏎Lily had 3 apples and found 8
-
worksheet 3.12a laptop by mini-lab.<|assistant_end|><|bos|>Anna had 21 books and found 5
-
worksheet 3.11= 19532.⏎Sue had 67 apples and found 6
-
worksheet 3.0955269.⏎Lucy had 298 books and found 5