INDEX

Explanations

I followed by verbs

This neuron detects the first‐person singular pronoun “I.”

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

hdysval

-1.69

-1.63

or

-1.60

 люди

-1.57

劻

-1.56

荩

-1.55

際に

-1.55

一直

-1.52

 poignant

-1.50

 для

-1.48

POSITIVE LOGITS

 were

1.77

are

1.69

 have

1.57

があるので

1.55

があり

1.55

鲧

1.54

︕

1.48

 alfombra

1.40

もありますが

1.38

穿过

1.38

Activations Density 0.005%