Skip to content
mini-lab
Features of prelude-2

Feature #237

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

Fires on
0.23% of tokens
Mostly in
worksheet
Layer
after block 4

Where it fires

  • worksheet 26%
  • chat 25%
  • python 19%
  • agent 16%
  • story 14%

Makes the model say

ai lastlast happilyOkayily backides

Makes it not say

th did But about isvedwherese

Strongest examples

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

  1. chat 3.5
    mini-lab.<|assistant_end|><|user_start|>Hello<|user_end|><|assistant_start|>Hi there! Would you like a story,
  2. chat 3.45
    with a calculator.<|assistant_end|><|user_start|>Hi<|user_end|><|assistant_start|>Hi there! Would you like a story,
  3. chat 3.39
    add numbers.<|assistant_end|><|user_start|>Hello!<|user_end|><|assistant_start|>Hi there! Would you like a story,
  4. chat 3.37
    Hi!<|user_end|><|assistant_start|>Hi there! Would you like a story,
  5. chat 3.37
    5.<|assistant_end|><|user_start|>hey there<|user_end|><|assistant_start|>Hi there! Would you like a story,
  6. chat 3.34
    !".<|system_end|><|user_start|>Hi<|user_end|><|assistant_start|>Sure! Hi there! Would you like a story,
  7. chat 3.34
    !".<|system_end|><|user_start|>Hello<|user_end|><|assistant_start|>Sure! Hi there! Would you like a story,
  8. chat 3.3
    assistant.<|system_end|><|user_start|>hey there<|user_end|><|assistant_start|>Hi there! Would you like a story,
  9. chat 3.29
    !".<|system_end|><|user_start|>Hi<|user_end|><|assistant_start|>Sure! Hi there! Would you like a story,
  10. chat 3.28
    <|system_end|><|user_start|>Hi!<|user_end|><|assistant_start|>Sure! Hi there! Would you like a story,
  11. story 3.26
    smiled and said, "Thank you for saying hello. Would you like to share my
  12. chat 3.25
    <|system_end|><|user_start|>Hi!<|user_end|><|assistant_start|>Sure! Hi there! Would you like a story,