INDEX

Explanations

reported speech verbs

The neuron activates on reporting verbs that attribute statements (e.g. “said,” “says”).

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

 Zähne

-1.69

⃙

-1.66

lahir

-1.66

 çantası

-1.58

鞏

-1.57

Berikut

-1.56

 décontract

-1.55

Controllo

-1.55

 ujar

-1.53

Kondisi

-1.52

POSITIVE LOGITS

of

2.89

2.09

但在

1.51

↵↵

1.44

We

1.41

in

1.41

1.38

 here

1.34

 kolejny

1.34

eeeee

1.34

Activations Density 0.012%