Skip to content
mini-lab
Features of prelude-2

Feature #2122

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

Fires on
0.21% of tokens
Mostly in
agent
Layer
after block 4

Where it fires

  • agent 45%
  • chat 24%
  • story 18%
  • python 13%
  • worksheet 0%

Makes the model say

sures yearsname friends filesEjoin

Makes it not say

without have fell And beak hadmin table

Strongest examples

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

  1. chat 2.87
    <|bos|><|system_start|>Start every answer with "Sure!".<|system_end|><|user_start|>Hi<|user_end|><|assistant_start|>Sure! Hi there
  2. chat 2.86
    <|bos|><|system_start|>Start every answer with "Sure!".<|system_end|><|user_start|>Hi<|user_end|><|assistant_start|>Sure! Hi there
  3. chat 2.8
    Start every answer with "Sure!".<|system_end|><|user_start|>hey there<|user_end|><|assistant_start|>Sure! Hello!
  4. chat 2.67
    <|bos|><|system_start|>Start every answer with "Sure!".<|system_end|><|user_start|>Hi<|user_end|><|assistant_start|>Sure! Hi there
  5. chat 2.64
    <|bos|><|system_start|>Start every answer with "Sure!".<|system_end|><|user_start|>Hi<|user_end|><|assistant_start|>Sure! Hi there
  6. chat 2.64
    <|system_start|>Start every answer with "Sure!".<|system_end|><|user_start|>Hi!<|user_end|><|assistant_start|>Sure! Hi there
  7. chat 2.64
    Sure!".<|system_end|><|user_start|>I want a story about the sea!<|user_end|><|assistant_start|>Sure! Once upon
  8. chat 2.58
    <|bos|><|system_start|>Start every answer with "Sure!".<|system_end|><|user_start|>Hello<|user_end|><|assistant_start|>Sure! Hi there
  9. chat 2.54
    <|bos|><|system_start|>Start every answer with "Sure!".<|system_end|><|user_start|>Hi<|user_end|><|assistant_start|>Sure! Hello!
  10. chat 2.53
    <|bos|><|system_start|>Start every answer with "Sure!".<|system_end|><|user_start|>Hi<|user_end|><|assistant_start|>Sure! Hello!
  11. chat 2.51
    answer with "Sure!".<|system_end|><|user_start|>I want a story!<|user_end|><|assistant_start|>Sure! One day,
  12. chat 2.49
    Start every answer with "Sure!".<|system_end|><|user_start|>Goodbye!<|user_end|><|assistant_start|>Sure! Bye!