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

Feature #125

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

Fires on
1.82% of tokens
Mostly in
story
Layer
after block 4

Where it fires

  • story 85%
  • chat 7%
  • agent 6%
  • worksheet 1%
  • python 1%

Makes the model say

you here!" your our my so today

Makes it not say

= understand tests Python every#includesut

Strongest examples

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

  1. story 6.24
    says, "I'm sorry, Mom, Dad, Mia. I was wrong. I did
  2. story 5.66
    .⏎"Ha!" Lily said. "I told you I was clever. I saved
  3. story 5.5
    wearing a green hat.⏎"I found you!" Ann shouted
  4. story 5.49
    ing that bell?"⏎The bird flew down and replied, "I was just out for a
  5. story 5.4
    . Tim and Sam said goodbye. Tim told his mom, "I had so much fun today
  6. story 5.25
    . The bird had the blue tie! The bird said, "I found this tie. I
  7. story 5.24
    "Mark, I found a safe! Let
  8. story 5.23
    ⏎In the morning, the dad looked up and said, "I did it! I fixed
  9. story 5.23
    "I'm sorry, mom. I'm sorry, Lily. I was playing with my rod
  10. story 5.18
    sees Lily. She is crying. He says, "Lily, I found you! Why are
  11. story 5.13
    and thanked the man.⏎"I'm sorry, Ben. I was careless. I should
  12. story 5.04
    a tree. She smiled and said, "I'm here. I was protecting the mailbox