Feature #1635
Fires on “Once” (not labeled yet: read the examples)
- Fires on
- 0.92% of tokens
- Mostly in
- python
- Layer
- after block 4
Where it fires
- python 45%
- story 30%
- agent 16%
- chat 8%
- worksheet 2%
Makes the model say
python<|tool_end|>pytest]de") NowMIT
Makes it not say
friend grass and fromen's someone2
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
story 4.42<|bos|>Once upon a time,
-
story 4.42<|bos|>Once upon a time,
-
story 4.42<|bos|>Once upon a time,
-
story 4.42<|bos|>Once upon a time,
-
story 4.42<|bos|>Once upon a time,
-
story 4.42<|bos|>Once upon a time,
-
story 4.221382<|bos|>Once upon a time,
-
story 4.18<|assistant_end|><|bos|>Once upon a time,
-
story 4.16py`.<|assistant_end|><|bos|>Once upon a time,
-
story 4.11```<|assistant_end|><|bos|>Once upon a time,
-
story 4.0725054<|bos|>Once upon a time,
-
story 4.0661799.<|bos|>Once upon a time,