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

Feature #1306

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

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

Where it fires

  • chat 31%
  • agent 30%
  • story 25%
  • python 13%
  • worksheet 1%

Makes the model say

different kept keep pretty his nice her nodded

Makes it not say

O True test# /8 intod

Strongest examples

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

  1. chat 3.04
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  2. chat 3.01
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  3. chat 2.97
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  4. chat 2.96
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  5. chat 2.96
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  6. chat 2.94
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  7. chat 2.92
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  8. chat 2.9
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  9. chat 2.89
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  10. chat 2.88
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  11. chat 2.87
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  12. chat 2.86
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories