Feature #563
Fires on “def” (not labeled yet: read the examples)
- Fires on
- 0.19% of tokens
- Mostly in
- agent
- Layer
- after block 4
Where it fires
- agent 41%
- python 40%
- story 15%
- chat 4%
- worksheet 0%
Makes the model say
to time(,"!",⏎ ver
Makes it not say
old loud at shiny dirty plan evenDon
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
story 4.46?"⏎Ben smiled and said, "Of course we can. But next time, let's
-
story 4.35okay. We can still make cocoa. Just be more quiet next time." They made
-
python 4.23⏎⏎>>> positive(6)⏎True⏎⏎```python⏎def factorial(n):
-
story 4.21She was not mad. She told Tim that it was okay, but next time, he should
-
story 4.11. You are too little. The show is for big kids. Maybe next year." Mom said
-
python 4.06TIGER")` returns `"tiger"`.⏎⏎```python⏎def sum_up_
-
python 4.03, 3])⏎4.75⏎⏎```python⏎def excite(s):
-
python 3.98, 4, 2])⏎4⏎<|bos|>```python⏎def odd(x):
-
python 3.96⏎⏎>>> vowels("lemon")⏎2⏎⏎```python⏎def divides(a,
-
python 3.91>>> is_odd(1)⏎True⏎⏎```python⏎def double(x):
-
chat 3.91But you can learn from this experience and be more careful next time."⏎Lucy
-
python 3.915])⏎[8, 8]⏎⏎```python⏎def last(items):