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

Feature #4055

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

Fires on
0.56% of tokens
Mostly in
agent
Layer
after block 4

Where it fires

  • agent 78%
  • chat 16%
  • story 4%
  • python 1%
  • worksheet 0%

Makes the model say

yateierprint lowerf checksdy

Makes it not say

house⏎ my " So home nestas

Strongest examples

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

  1. chat 3.0
    numbers.<|assistant_end|><|bos|><|system_start|>tools: calculator<|system_end|><|system_start|>You are a helpful assistant.<|system_end|><|user_start|>G
  2. chat 2.95
    do what you want them to do.<|bos|><|system_start|>You are a helpful assistant.<|system_end|><|user_start|>Find
  3. chat 2.85
    the Python files in this project.<|assistant_end|><|bos|><|system_start|>You are a helpful assistant.<|system_end|><|user_start|>Can
  4. chat 2.83
    add?<|assistant_end|><|bos|><|system_start|>tools: calculator<|system_end|><|system_start|>You are a helpful assistant.<|system_end|><|user_start|>Can
  5. chat 2.82
    numbers.<|assistant_end|><|bos|><|system_start|>tools: calculator<|system_end|><|system_start|>You are a helpful assistant.<|system_end|><|user_start|>What
  6. chat 2.78
    add?<|assistant_end|><|bos|><|system_start|>tools: calculator<|system_end|><|system_start|>You are a helpful assistant.<|system_end|><|user_start|>Compute
  7. chat 2.77
    48.<|bos|><|system_start|>tools: calculator<|system_end|><|system_start|>You are a helpful assistant.<|system_end|><|user_start|>Write
  8. chat 2.75
    .<|assistant_end|><|bos|><|system_start|>You are a helpful assistant.<|system_end|><|user_start|>Hello
  9. chat 2.71
    race together.<|bos|><|system_start|>tools: calculator<|system_end|><|system_start|>You are a helpful assistant.<|system_end|><|user_start|>What
  10. chat 2.7
    0.<|assistant_end|><|bos|><|system_start|>tools: calculator<|system_end|><|system_start|>You are a helpful assistant.<|system_end|><|user_start|>Can
  11. chat 2.62
    pose.<|bos|><|system_start|>tools: calculator<|system_end|><|system_start|>You are a helpful assistant.<|system_end|><|user_start|>Compute
  12. chat 2.58
    19.<|bos|><|system_start|>tools: calculator<|system_end|><|system_start|>You are a helpful assistant.<|system_end|><|user_start|>what