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

Feature #1468

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

Fires on
0.09% of tokens
Mostly in
chat
Layer
after block 4

Where it fires

  • chat 62%
  • worksheet 19%
  • story 9%
  • agent 9%
  • python 1%

Makes the model say

items Can needed tredE st called

Makes it not say

)`'s /ful wave range jo))

Strongest examples

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

  1. chat 2.42
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  2. chat 2.42
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  3. chat 2.38
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  4. chat 2.37
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  5. chat 2.34
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  6. chat 2.31
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  7. chat 2.24
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  8. chat 2.21
    into Spanish: good morning.<|user_end|><|assistant_start|>Sorry, I'm a tiny model: I
  9. chat 2.2
    <|user_start|>What does DNA stand for?<|user_end|><|assistant_start|>Sorry, I'm a tiny model: I
  10. chat 2.18
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  11. chat 2.18
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories
  12. chat 2.17
    <|assistant_start|>I'm sorry, I can't help with that. I only know how to tell stories