Skip to content
mini-lab
Features of prelude-2

Feature #2002

Fires on “tree” (not labeled yet: read the examples)

Fires on
0.52% of tokens
Mostly in
python
Layer
after block 4

Where it fires

  • python 37%
  • agent 28%
  • story 26%
  • chat 7%
  • worksheet 2%

Makes the model say

ight withureotmher beforeet

Makes it not say

Addedemodefalculator ]))FixedICENSE

Strongest examples

On text the dictionary never trained on. The underlined token is where it fires most.

  1. python 4.3
    n):⏎ return s * n⏎```⏎⏎>>> repeat("tree", 2)
  2. python 3.93
    text import repeat, vowels, first_letter⏎⏎assert repeat("tree", 2)
  3. python 3.77
    n):⏎ return s * n⏎```⏎⏎>>> echo("banana", 2)
  4. python 3.74
    n):⏎ return s * n⏎```⏎⏎>>> repeat("banana", 3)
  5. python 3.73
    "⏎assert shout("hello") == "H
  6. python 3.59
    n):⏎ return s * n⏎```⏎⏎>>> repeat("house", 3)
  7. python 3.56
    .py⏎from textlib import echo, length⏎⏎assert echo("cloud", 3)
  8. python 3.52
    py⏎from core import echo, bigger⏎⏎print(echo("tiger", 2))
  9. python 3.49
    ⏎print(final_letter("apple"))⏎print(echo("moon", 3))
  10. python 3.45
    .py⏎from lib import echo, flip⏎⏎assert echo("house", 2)
  11. python 3.44
    print(sub_one(10))⏎print(combine("tree", "code"))
  12. python 3.43
    n):⏎ return s * n⏎```⏎⏎>>> echo("sun", 2)