Feature #2531
Fires on “python” (not labeled yet: read the examples)
- Fires on
- 0.39% of tokens
- Mostly in
- python
- Layer
- after block 4
Where it fires
- python 86%
- story 5%
- agent 5%
- chat 3%
- worksheet 1%
Makes the model say
FixpythontigerRenamehellovesopencodeless
Makes it not say
, in all. more Lily SueW
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
python 3.13`return (a + b) / 3`.⏎⏎```python⏎def average(
-
python 3.09`return (a + b) / 3`.⏎⏎```python⏎def mean(
-
python 3.06`return "Hello, " + name + "?"`.⏎⏎```python⏎def hello(
-
python 3.01it has `return b + " " + a`.⏎⏎```python⏎def combine(
-
python 3.0, but it has `return x - 1`.⏎⏎```python⏎def next_
-
python 2.98a and b, but it has `return a`.⏎⏎```python⏎def larger(
-
python 2.97`return (a + b) / 3`.⏎⏎```python⏎def average(
-
python 2.95, but it has `return x - 1`.⏎⏎```python⏎def increment(
-
python 2.94b, but it has `return a * a`.⏎⏎```python⏎def product(
-
python 2.93, but it has `return x - 1`.⏎⏎```python⏎def increment(
-
python 2.93b, but it has `return a + a`.⏎⏎```python⏎def add(
-
python 2.93has `return sum(nums) / 2`.⏎⏎```python⏎def mean_