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

Feature #2764

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

Fires on
0.15% of tokens
Mostly in
chat
Layer
after block 4

Where it fires

  • chat 67%
  • story 22%
  • agent 10%
  • python 2%
  • worksheet 0%

Makes the model say

later sentFix hello lots tomorrowious

Makes it not say

downhower where himself He princessen

Strongest examples

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

  1. chat 6.44
    <|bos|><|user_start|>Can you help me with my taxes?<|user_end|><|assistant_start|>Sorry, I'm a
  2. chat 5.9
    Summarize this article for me.<|user_end|><|assistant_start|>Sorry, I'm a
  3. chat 5.9
    py`.<|assistant_end|><|bos|><|user_start|>What does DNA stand for?<|user_end|><|assistant_start|>Sorry, I'm a
  4. chat 5.82
    <|bos|><|user_start|>Who wrote Romeo and Juliet?<|user_end|><|assistant_start|>Sorry, I don't
  5. chat 5.81
    tools: calculator<|system_end|><|user_start|>What does DNA stand for?<|user_end|><|assistant_start|>Sorry, I'm a
  6. chat 5.68
    Summarize this article for me.<|user_end|><|assistant_start|>Sorry, I'm a
  7. chat 5.66
    plain quantum physics to me.<|user_end|><|assistant_start|>Sorry, I'm a
  8. chat 5.61
    tools: calculator<|system_end|><|user_start|>How far away is the moon?<|user_end|><|assistant_start|>Sorry, I don't
  9. chat 5.57
    tell me about the history of India<|user_end|><|assistant_start|>Sorry, I don't
  10. chat 5.56
    py`.<|assistant_end|><|bos|><|user_start|>How far away is the moon?<|user_end|><|assistant_start|>Sorry, I don't
  11. chat 5.54
    <|user_start|>When did World War II end?<|user_end|><|assistant_start|>Sorry, I'm a
  12. chat 5.51
    at school.<|bos|><|user_start|>How far away is the moon?<|user_end|><|assistant_start|>Sorry, I don't