Feature #70
Fires on “ x” (not labeled yet: read the examples)
- Fires on
- 0.99% of tokens
- Mostly in
- python
- Layer
- after block 4
Where it fires
- python 59%
- agent 28%
- chat 6%
- worksheet 5%
- story 1%
Makes the model say
x Python tests <Fix >alculator doubled
Makes it not say
`⏎," to idea runJohn adventure
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
python 8.560⏎```⏎⏎`is_positive` returns True if x is positive. For
-
python 8.5x > 0⏎```⏎⏎`positive` returns True if x is positive. For
-
python 8.32x > 0⏎```⏎⏎`positive` returns True if x is positive. For
-
python 8.31robot"`.⏎⏎`includes(items, x)` returns True if x is in items.
-
python 8.27```⏎⏎Bug: `is_positive` should return True if x is positive, but
-
python 8.23```⏎⏎Bug: `is_even` should return True if x is even, but
-
python 8.16result⏎```⏎⏎Bug: `includes` should return True if x is in items,
-
python 8.165.<|assistant_end|><|bos|>`is_positive(x)` returns True if x is positive.⏎
-
python 8.13```⏎⏎Bug: `has_item` should return True if x is in items,
-
agent 8.112 == 1<|tool_end|><|assistant_start|>`odd(x)` returns True if x is odd.<|assistant_end|>
-
python 8.1⏎⏎Bug: `is_even` should return True if x is even, but
-
python 8.09```⏎⏎Bug: `has_item` should return True if x is in items,