Feature #2119
Fires on “ in” (not labeled yet: read the examples)
- Fires on
- 0.36% of tokens
- Mostly in
- agent
- Layer
- after block 4
Where it fires
- agent 53%
- chat 22%
- story 20%
- python 3%
- worksheet 1%
Makes the model say
this new)) together gift surprised upon smiles
Makes it not say
⏎ and felt rain didn5 was still
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
agent 3.91list, read, run, create, edit and fix the Python files in this project. Try
-
agent 3.9list, read, run, create, edit and fix the Python files in this project.<|assistant_end|>
-
agent 3.81list, read, run, create, edit and fix the Python files in this project. Try
-
agent 3.81list, read, run, create, edit and fix the Python files in this project.<|assistant_end|>
-
chat 3.78NA stand for?<|user_end|><|assistant_start|>I'm sorry, I can't help with that. I only
-
agent 3.77list, read, run, create, edit and fix the Python files in this project.<|assistant_end|>
-
agent 3.72list, read, run, create, edit and fix the Python files in this project.<|assistant_end|>
-
agent 3.71list, read, run, create, edit and fix the Python files in this project.<|assistant_end|>
-
agent 3.71list, read, run, create, edit and fix the Python files in this project. Try
-
agent 3.69list, read, run, create, edit and fix the Python files in this project.<|assistant_end|>
-
agent 3.69list, read, run, create, edit and fix the Python files in this project.<|assistant_end|>
-
agent 3.69list, read, run, create, edit and fix the Python files in this project.<|assistant_end|>