Feature #374
Fires on “TEXT” (not labeled yet: read the examples)
- Fires on
- 0.59% of tokens
- Mostly in
- agent
- Layer
- after block 4
Where it fires
- agent 66%
- python 29%
- story 3%
- worksheet 2%
- chat 0%
Makes the model say
))txtai dirt %it name high
Makes it not say
." Can`assert oneof0⏎
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
python 2.7words("banana", "cloud"))⏎print(last_letter(TEXT))⏎⏎$
-
python 2.696))⏎print(even(2))⏎print(opposite(7))⏎⏎$
-
python 2.61, 8, 4]))⏎print(quiet("Tree"))⏎⏎$
-
python 2.53], 7))⏎print(raise_to(1, 2))⏎⏎#
-
python 2.38, 2, 1]))⏎print(next_number(3))⏎⏎#
-
python 2.38⏎print(count_vowels("hello"))⏎print(square(SIZE))⏎⏎#
-
python 2.38⏎print(count([8, 2, 2, 4, 2]))
-
python 2.36⏎print(square(9))⏎print(last_letter(TEXT))⏎⏎$
-
python 2.32(-1))⏎print(max_of(12, 6))⏎⏎#
-
python 2.32⏎print(first_letter("apple"))⏎print(cubed(6))⏎⏎#
-
python 2.32print(quiet("STAR"))⏎print(length("pizza"))⏎⏎#
-
python 2.297]⏎print(previous(7))⏎print(head(SCORES))⏎⏎#