Skip to content
mini-lab
Features of prelude-2

Feature #820

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

Fires on
0.25% of tokens
Mostly in
agent
Layer
after block 4

Where it fires

  • agent 75%
  • story 20%
  • chat 3%
  • python 2%
  • worksheet 0%

Makes the model say

only toR everything sharing even vowelsmain

Makes it not say

sawassertprint. whatillsumve

Strongest examples

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

  1. agent 3.3
    hello` in `greetings.py`: it used `return "Helo
  2. agent 3.18
    ` in `listlib.py`: it used `return [n
  3. agent 3.15
    ⏎<|tool_end|><|assistant_start|>Fixed `hello` in `core.py`: it used `return "Hello
  4. agent 3.13
    negate` in `calc.py`: it used `return x`
  5. agent 3.12
    <|assistant_start|>Fixed `increment` in `math_utils.py`: it used `return x -
  6. agent 3.1
    <|assistant_start|>Fixed `factorial` in `math_utils.py`: it used `result = result
  7. agent 3.1
    words.py`: it used `return s[
  8. agent 3.1
    ⏎<|tool_end|><|assistant_start|>Fixed `shout` in `utils.py`: it used `return s`
  9. agent 3.09
    textlib.py`: it used `return s *
  10. agent 3.07
    ⏎<|tool_end|><|assistant_start|>Fixed `total` in `listlib.py`: it used `result = result
  11. agent 3.07
    ⏎<|tool_end|><|assistant_start|>Fixed `smallest` in `listlib.py`: it used `return max(
  12. agent 3.07
    ⏎<|tool_end|><|assistant_start|>Fixed `combine` in `strings.py`: it used `return b +