INDEX

Explanations

unexpectedly or not expected

The neuron spotlights words and phrases that convey surprise or counter-expectation (e.g. “unexpectedly,” “thought,” “could have guessed”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 giver

-1.61

轫

-1.59

Heraus

-1.58

 ſch

-1.56

 cassa

-1.55



-1.55

 diversa

-1.52

廰

-1.50

 proteine

-1.48

POSITIVE LOGITS

1.84

楯

1.65

('

1.57

1.54

."

1.47

 agrad

1.45



1.45

.'

1.44

siasi

1.43

AutoresizingMask

1.40

Activations Density 0.002%