INDEX

Explanations

introductions of self

the model's self-identification — it activates on mentions of the assistant's name / self-introduction.

New Auto-Interp

Configuration

Prompts (Dashboard)

238,145 prompts, 512 tokens each

Dataset (Dashboard)

lmsys + oasst1

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 himself

0.50

 ourselves

0.47

 Himself

0.46

Initial

0.43

 esetén

0.42

 Initially

0.39

initial

0.39

 पाते

0.39

ក្ខ

0.39

 initial

0.38

POSITIVE LOGITS

 abbreviation

0.48

pronounced

0.46

 pronouns

0.42

jedno

0.41

最近

0.41

 ولی

0.41

 surnames

0.40

 Можно

0.40

Можно

0.39

ちなみに

0.39

Activations Density 0.044%