Skip to content
mini-lab
Features of prelude-2

Feature #785

Fires on “<|assistant_end|>” (not labeled yet: read the examples)

Fires on
2.83% of tokens
Mostly in
agent
Layer
after block 4

Where it fires

  • agent 91%
  • python 4%
  • chat 2%
  • story 2%
  • worksheet 1%

Makes the model say

<|assistant_end|>newString<|user_start|><|system_end|><|system_start|>ice moral<|tool_call_start|>

Makes it not say

, openedut away door0 test down

Strongest examples

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

  1. agent 3.07
    py⏎lib.py⏎test_lib.py<|tool_end|><|assistant_start|>This project has
  2. agent 2.95
    test_calc.py` and `test_listlib.py`.<|assistant_end|><|bos|><|system_start|>Answer with
  3. agent 2.89
    test_mathlib.py` and `test_stats.py`.<|assistant_end|><|bos|>Once upon a
  4. agent 2.88
    `, `stats.py` and `test_stats.py`.<|assistant_end|><|bos|>79 plus
  5. agent 2.83
    `, `test_utils.py`, `test_strings.py`, `requirements
  6. agent 2.75
    test_helpers.py` and `test_items.py`.<|assistant_end|><|user_start|>What's in
  7. agent 2.75
    test_core.py` and `test_items.py`.<|assistant_end|><|bos|><|system_start|>Answer with
  8. agent 2.72
    test_calc.py` and `test_strings.py`.<|assistant_end|><|bos|>```python⏎
  9. agent 2.72
    greetings.py⏎test_stats.py<|tool_end|><|assistant_start|><|tool_call_start|>shell
  10. agent 2.71
    `, `utils.py` and `test_words.py`.<|assistant_end|><|user_start|>Run the tests
  11. agent 2.7
    utils.py⏎test_core.py<|tool_end|><|assistant_start|><|tool_call_start|>shell
  12. agent 2.68
    textlib.py` and `test_number_utils.py`.<|assistant_end|><|bos|><|system_start|>tools: