Feature #3631
Fires on “1” (not labeled yet: read the examples)
- Fires on
- 1.20% of tokens
- Mostly in
- worksheet
- Layer
- after block 4
Where it fires
- worksheet 55%
- chat 34%
- python 9%
- agent 1%
- story 1%
Makes the model say
='s<|assistant_end|>/So tests)`. filled
Makes it not say
to0One helpigx everyone1
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
worksheet 3.468+1=14, 470⏎0+1: 0+
-
chat 3.466+7=13, 38⏎4+1: 4+
-
chat 3.43: 9+9=18, 8⏎5+1: 5+
-
worksheet 3.419+7=16, 683⏎0+1: 0+
-
worksheet 3.40+7=7, 735⏎0+1: 0+
-
worksheet 3.4+1+1=2, 284⏎0+1: 0+
-
chat 3.383: 5+3=8, 8⏎6+5: 6+
-
chat 3.38+7+1=17, 72⏎2+1: 2+
-
worksheet 3.371+0=1, 1854⏎5+0: 5+
-
worksheet 3.35: 1+8=9, 97⏎0+7: 0+
-
chat 3.34: 2+0=2, 29⏎1+0: 1+
-
chat 3.34: 8+0=8, 84⏎5+0: 5+