Feature #9
Fires on “print” (not labeled yet: read the examples)
- Fires on
- 0.58% of tokens
- Mostly in
- python
- Layer
- after block 4
Where it fires
- python 55%
- agent 24%
- story 10%
- chat 7%
- worksheet 5%
Makes the model say
( again printsOnceInside many inside```
Makes it not say
hmainSamonly me?I positive
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
python 3.72⏎NUMBER = -9⏎print(magnitude(NUMBER))⏎print(lower_of
-
python 3.68(min_item([4, 9, 8]))⏎print(diffe
-
python 3.64, opposite⏎⏎print(next_number(11))⏎print(opposite(4
-
python 3.61, product⏎⏎print(product(7, 4))⏎print(positive(-9
-
python 3.6text import loud, initial⏎⏎print(initial("robot"))⏎print(loud("house
-
python 3.6_one⏎⏎print(sub_one(10))⏎print(combine("tree
-
python 3.59⏎TEXT = "robot"⏎print(loud(TEXT))⏎print(sum_up
-
python 3.572, 7]⏎print(previous(7))⏎print(head(SCORES
-
python 3.56Ada"⏎print(hello(USER))⏎print(contains([9
-
python 3.56= 6⏎print(larger(9, 7))⏎print(sub_one
-
python 3.55, 3]⏎print(min_item(VALUES))⏎print(total([4
-
python 3.55_one⏎⏎print(add_one(12))⏎print(product(5