Skip to content
mini-lab
Features of prelude-2

Feature #1742

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

Fires on
1.24% of tokens
Mostly in
python
Layer
after block 4

Where it fires

  • python 44%
  • agent 42%
  • story 9%
  • chat 3%
  • worksheet 2%

Makes the model say

([countlasth countreturn_ numbers

Makes it not say

very?⏎ big mom baby round gift

Strongest examples

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

  1. python 3.51
    shout("hello") == "HELLO"⏎assert first_letter("robot
  2. python 3.42
    assert loud("cloud") == "CLOUD"⏎assert count_vowels("hello
  3. python 3.39
    0, 0, 6, 18]⏎assert count_items([6
  4. python 3.31
    ⏎⏎assert count_vowels("house") == 3⏎assert count_vowels("cat
  5. python 3.25
    0, 18, 10]⏎assert double_all([0,
  6. python 3.23
    count_items` returns the number of items. For example, `count_items([3
  7. python 3.23
    (items):⏎ return len(items)⏎```⏎⏎>>> count_items([1
  8. python 3.17
    (items):⏎ return len(items)⏎```⏎⏎>>> count_items([6
  9. python 3.14
    (items):⏎ return len(items)⏎```⏎⏎>>> count_items([6
  10. python 3.13
    ⏎assert final_letter("river") == "r"⏎assert count_vowels("star
  11. python 3.13
    flip("apple") == "elppa"⏎assert add_all([6,
  12. python 3.11
    6666666665⏎assert sum_list([4,