Feature #1253
Fires on “⏎” (not labeled yet: read the examples)
- Fires on
- 0.45% of tokens
- Mostly in
- story
- Layer
- after block 4
Where it fires
- story 55%
- agent 22%
- python 15%
- chat 6%
- worksheet 2%
Makes the model say
answer his thanue nodded soon off him
Makes it not say
became becomegestnums``` Goter`
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
python 2.97⏎⏎# demo.py⏎from arith import half, negate⏎⏎print(half
-
python 2.91⏎⏎# main.py⏎from number_utils import halve⏎⏎print(halve(
-
python 2.88⏎⏎# script.py⏎from math_utils import product⏎⏎print(product(
-
python 2.87.<|bos|># demo.py⏎from mathlib import power, cube⏎⏎print(power(
-
python 2.86+ "!"⏎⏎# main.py⏎from lib import hello⏎⏎print(hello("
-
python 2.861]⏎⏎# main.py⏎from textlib import loud⏎⏎print(loud("
-
python 2.82⏎<|bos|># demo.py⏎from calc import divides, decrement⏎⏎print(dec
-
python 2.8b.<|assistant_end|><|bos|># main.py⏎from items import biggest⏎⏎VALUES = [1
-
python 2.73app.py⏎from items import contains⏎from mathlib import fact⏎⏎print(contains([
-
python 2.71⏎⏎# main.py⏎from calc import positive, product⏎⏎print(product(
-
python 2.7s)⏎⏎# demo.py⏎from words import length⏎⏎WORD = "pizza
-
python 2.7+ "!"⏎⏎# main.py⏎from core import positive⏎⏎LIMIT = 1