Feature #3810
Fires on “ help” (not labeled yet: read the examples)
- Fires on
- 0.47% of tokens
- Mostly in
- story
- Layer
- after block 4
Where it fires
- story 64%
- chat 24%
- agent 10%
- worksheet 1%
- python 1%
Makes the model say
you youroldString?" higher theyYou wet
Makes it not say
stHow double maxdoubleWhatight wasn
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
chat 5.05calculator.<|assistant_end|><|user_start|>Good morning!<|user_end|><|assistant_start|>Hello! How can I help you today?<|assistant_end|>
-
chat 4.962.<|assistant_end|><|user_start|>Good morning!<|user_end|><|assistant_start|>Hello! How can I help you today?<|assistant_end|>
-
chat 4.86add numbers.<|assistant_end|><|user_start|>Hi!<|user_end|><|assistant_start|>Hello! How can I help you today?<|assistant_end|>
-
chat 4.84and add numbers.<|assistant_end|><|user_start|>Hi<|user_end|><|assistant_start|>Hello! How can I help you today?<|assistant_end|>
-
chat 4.84a helpful assistant.<|system_end|><|user_start|>Hello<|user_end|><|assistant_start|>Hello! How can I help you today?<|assistant_end|>
-
chat 4.71friendly assistant.<|system_end|><|user_start|>Hi!<|user_end|><|assistant_start|>Hello! How can I help you today?<|assistant_end|>
-
chat 4.68end.<|bos|><|user_start|>Good morning!<|user_end|><|assistant_start|>Hello! How can I help you today?<|assistant_end|>
-
chat 4.65were fine friends.<|assistant_end|><|user_start|>Hi<|user_end|><|assistant_start|>Hello! How can I help you today?<|assistant_end|>
-
chat 4.64calculator.<|assistant_end|><|user_start|>Good morning!<|user_end|><|assistant_start|>Hello! How can I help you today?<|assistant_end|>
-
chat 4.63lab.<|assistant_end|><|user_start|>Good morning!<|user_end|><|assistant_start|>Hello! How can I help you today?<|assistant_end|>
-
chat 4.59calculator.<|assistant_end|><|user_start|>Good morning!<|user_end|><|assistant_start|>Hello! How can I help you today?<|assistant_end|>
-
chat 4.59numbers.<|assistant_end|><|user_start|>Good morning!<|user_end|><|assistant_start|>Hello! How can I help you today?<|assistant_end|>