Skip to content
mini-lab
Log in Sign up
Features of prelude-2

Feature #1152

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

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

Where it fires

  • python 69%
  • worksheet 14%
  • agent 9%
  • chat 6%
  • story 2%

Makes the model say

0152637 welcome

Makes it not say

dream each was happen preludes'sw

Strongest examples

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

  1. python 4.73
    of nums. For example, `evens([1, 5, 2])` returns `[
  2. python 4.72
    ⏎```⏎⏎>>> sum_list([6, 9, 1])⏎1
  3. python 4.63
    _list⏎⏎assert has_item([4, 5, 2, 4
  4. python 4.62
    , last⏎⏎assert has_item([3, 6, 5, 5
  5. python 4.61
    ⏎```⏎⏎>>> min_item([3, 9, 9, 4
  6. python 4.56
    ```⏎⏎>>> only_even([8, 8, 5])⏎[
  7. python 4.52
    _words⏎⏎assert sum_list([7, 0, 4, 0
  8. python 4.48
    . For example, `count_items([3, 8, 6, 7
  9. python 4.44
    _item⏎⏎assert keep_even([0, 5, 3, 4
  10. python 4.41
    from listlib import biggest⏎⏎assert biggest([3, 5, 9, 1
  11. python 4.39
    ⏎```⏎⏎>>> mean_list([5, 7, 4, 3
  12. python 4.37
    is_odd⏎⏎assert last([8, 8, 8, 4