INDEX

Explanations

actions or professions

The neuron selectively lights up on verbs that describe professional or work-related actions (e.g. installing, managing, writing, illustrating, coding).

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

its

-1.47

↵↵↵↵↵↵↵↵↵↵

-1.45

 which

-1.44

↵↵↵↵↵↵↵↵↵↵↵↵↵

-1.36

↵↵↵↵↵↵↵↵

-1.36

↵↵↵↵↵↵↵

-1.33

↵↵↵↵↵↵↵↵↵↵↵

-1.27

 miejsce

-1.27

 they

-1.24

↵↵↵↵↵↵↵↵↵

-1.23

POSITIVE LOGITS

Those

1.58

 also

1.57

has

1.35

 vinos

1.28

 också

1.24

for

1.23

chot

1.19

and

1.18

 bolí

1.17

Several

1.16

Activations Density 0.129%