Skip to content
mini-lab
Features of prelude-2

Feature #930

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

Fires on
0.25% of tokens
Mostly in
story
Layer
after block 4

Where it fires

  • story 57%
  • chat 16%
  • agent 13%
  • worksheet 10%
  • python 3%

Makes the model say

.libtxt became[ed bottom doubled

Makes it not say

biggestLet our whitejoina moBut

Strongest examples

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

  1. story 2.04
    to sail and feel the wind.⏎While sailing, the boat met a little oyster.
  2. story 1.99
    . The bee was very happy.⏎In the field, the bee met a butterfly. The
  3. chat 1.98
    pretty dress and went to the park.⏎At the park, she met a boy named Tom
  4. story 1.97
    Tim decided to go to the tree.⏎On his way, he met a girl named Sara
  5. story 1.97
    .⏎As Will climbed the tree, he met a bird. The
  6. chat 1.94
    and started to pull it.⏎As Tom pulled the wagon, he met a girl named Sue
  7. story 1.94
    it up and began to carry it. As she was walking, she met her friend Jack.
  8. story 1.93
    Tom was very excited to be there.⏎In one room, Tom met a dog named Max
  9. story 1.86
    to make his house dirty too.⏎On his way home, Spot met a kind girl named
  10. story 1.85
    he could not find the ball.⏎While he was walking, Tim met a friendly dog.
  11. story 1.84
    but Tiny did not mind.⏎One day, while flying, Tiny met a big bear named
  12. story 1.84
    get dirty.⏎One day, while playing in the garden, Mia met a boy named Tom