INDEX

Explanations

verbs describing action

The neuron fires on occurrences of the verb “use” (in its various inflected forms).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

and

-2.16

 also

-1.59

伱

-1.41

也要

-1.33

 cũng

-1.33

 cherokee

-1.32

 only

-1.30

conio

-1.27

 なお

-1.27

ting

-1.24

POSITIVE LOGITS

the

1.68

 mulai

1.50

We

1.43

 their

1.38

 muestran

1.37

 "}";

1.34

 interesan

1.33

 gustaba

1.32

 encantó

1.32

 tärke

1.30

Activations Density 0.051%