Skip to content
mini-lab
Features of prelude-2

Feature #1997

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

Fires on
0.28% of tokens
Mostly in
agent
Layer
after block 4

Where it fires

  • agent 66%
  • python 22%
  • story 10%
  • chat 3%
  • worksheet 0%

Makes the model say

Its loud better: to``` good

Makes it not say

)`>>> doorifememberAs<|think_end|> careful

Strongest examples

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

  1. agent 2.77
    True⏎1⏎pizza hello⏎<|tool_end|><|assistant_start|>`script.py` printed:⏎```⏎
  2. python 2.76
    b):⏎ return a - b⏎```⏎⏎`subtract` subtracts b from a
  3. agent 2.75
    lemon⏎Hello, Tom!⏎<|tool_end|><|assistant_start|>`demo.py` printed:⏎```⏎
  4. agent 2.74
    <|tool_call_end|><|assistant_end|><|tool_start|>False⏎3⏎<|tool_end|><|assistant_start|>`script.py` printed:⏎```⏎
  5. agent 2.72
    <|tool_start|>True⏎7.5⏎<|tool_end|><|assistant_start|>`demo.py` printed:⏎```⏎
  6. agent 2.68
    .py<|tool_call_end|><|assistant_end|><|tool_start|>True⏎<|tool_end|><|assistant_start|>`app.py` printed:⏎```⏎
  7. agent 2.68
    .py<|tool_call_end|><|assistant_end|><|tool_start|>4⏎<|tool_end|><|assistant_start|>`app.py` printed:⏎```⏎
  8. agent 2.67
    py<|tool_call_end|><|assistant_end|><|tool_start|>python banana⏎<|tool_end|><|assistant_start|>`script.py` printed:⏎```⏎
  9. agent 2.67
    bananabananabanana⏎<|tool_end|><|assistant_start|>`demo.py` printed:⏎```⏎
  10. agent 2.66
    ⏎tigertiger⏎pizza!⏎<|tool_end|><|assistant_start|>`run.py` printed:⏎```⏎
  11. agent 2.65
    .py<|tool_call_end|><|assistant_end|><|tool_start|>e⏎<|tool_end|><|assistant_start|>`app.py` printed:⏎```⏎
  12. agent 2.65
    <|tool_start|>5⏎1.0⏎<|tool_end|><|assistant_start|>`demo.py` printed:⏎```⏎