Skip to content
mini-lab
Features of prelude-2

Feature #1366

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

Fires on
0.30% of tokens
Mostly in
python
Layer
after block 4

Where it fires

  • python 81%
  • agent 10%
  • story 6%
  • chat 2%
  • worksheet 1%

Makes the model say

return whetherfOnce`py plus3

Makes it not say

There):asall longused here f

Strongest examples

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

  1. python 3.65
    ⏎⏎>>> positive(6)⏎True⏎⏎```python⏎def factorial(n):
  2. python 3.58
    ` should return the average of a
  3. python 3.51
    5])⏎[8, 8]⏎⏎```python⏎def last(items):
  4. python 3.51
    TIGER")` returns `"tiger"`.⏎⏎```python⏎def sum_up_
  5. python 3.48
    items)` counts the items.<|assistant_end|><|bos|>Bug: `combine` should return a and b separ
  6. python 3.43
    , 3])⏎4.75⏎⏎```python⏎def excite(s):
  7. python 3.43
    9, 2)` returns `False`.⏎⏎```python⏎def positive(x):
  8. python 3.37
    , 4, 2])⏎4⏎<|bos|>```python⏎def odd(x):
  9. python 3.37
    ⏎⏎>>> vowels("pizza")⏎2⏎⏎```python⏎def loud(s):
  10. python 3.36
    ("river")⏎"revir"⏎⏎```python⏎def negate(x):
  11. python 3.35
    >>> is_odd(1)⏎True⏎⏎```python⏎def double(x):
  12. python 3.32
    0, 4])` returns `8`.⏎⏎```python⏎def exclaim(s):