Skip to content
mini-lab
Features of prelude-2

Feature #3253

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

Fires on
0.92% of tokens
Mostly in
story
Layer
after block 4

Where it fires

  • story 72%
  • agent 14%
  • chat 12%
  • python 1%
  • worksheet 1%

Makes the model say

unexpected even long things joy also helpually

Makes it not say

= asked told`: Python() thoughtxt

Strongest examples

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

  1. story 7.92
    toy stopped growing, and Sue and Tim learned that sharing with others can make everyone happy.
  2. story 7.83
    , the three friends played together every day. They learned that helping others can make you happy and
  3. story 7.18
    and they played together all day. Tim learned that sharing and helping others can make you feel good
  4. story 6.47
    Tom learned that sometimes, even when things
  5. story 6.39
    and Tim felt very happy. The moral of the story is that sharing can make everyone happy,
  6. story 6.29
    made them feel good. The moral of the story is that helping others can make you happy too
  7. story 6.23
    and had fun. Tim and his friends laughed. They learned that sharing can make others happy too
  8. story 6.22
    . The dog learned to behave, and Tim learned that helping others can make things better for
  9. story 6.12
    was happy. In the end, they both learned that helping others can sometimes help you too.
  10. story 6.1
    than to fight. And Lucy learned that asking for help and being kind can make everyone happy.
  11. chat 6.09
    be a good friend. And the crab learned that talking to new friends can be fun. They
  12. story 5.98
    was very happy and thanked Lily.⏎The two friends learned that sharing can make everyone happy and