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

Feature #1920

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

Fires on
1.03% of tokens
Mostly in
agent
Layer
after block 4

Where it fires

  • agent 74%
  • python 15%
  • story 8%
  • chat 2%
  • worksheet 1%

Makes the model say

lm comingap.ed goingure

Makes it not say

(- into TheIfverpped wearing onto

Strongest examples

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

  1. python 4.28
    ⏎⏎>>> vowels("lemon")⏎2⏎⏎```python⏎def divides(a,
  2. python 4.25
    ("river")⏎"revir"⏎⏎```python⏎def negate(x):
  3. python 4.21
    , 3])⏎4.75⏎⏎```python⏎def excite(s):
  4. python 4.2
    >>> initial("hello")⏎"h"⏎⏎```python⏎def cube(x):
  5. python 4.2
    ⏎⏎>>> positive(6)⏎True⏎⏎```python⏎def factorial(n):
  6. python 4.15
    TIGER")` returns `"tiger"`.⏎⏎```python⏎def sum_up_
  7. python 4.08
    2)⏎"treetree"⏎⏎```python⏎def divides(a,
  8. python 4.07
    0, 4])` returns `8`.⏎⏎```python⏎def exclaim(s):
  9. python 4.05
    ⏎⏎>>> vowels("pizza")⏎2⏎⏎```python⏎def loud(s):
  10. python 4.01
    7, 6])` returns `7`.⏎⏎```python⏎def loud(s):
  11. python 4.01
    ("cat")⏎"CAT"⏎⏎```python⏎def max_item(
  12. python 3.99
    2, 5])` returns `[2]`.⏎⏎```python⏎def raise_to(