INDEX

Explanations

i thought assertion

The neuron fires on first-person opinion phrases—especially occurrences of “I thought” (and its variants).

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

-1.58

-1.57

ti

-1.50

-1.48

if

-1.45

 still

-1.44

この

-1.41

 francese

-1.35

-1.30

-1.29

POSITIVE LOGITS

ᯅ

2.08

ции

1.77

[];

1.72

meldet

1.59

følgelig

1.56

 recentemente

1.54

stwie

1.52

玧

1.52

менение

1.51

鸸

1.50

Activations Density 0.022%