INDEX

Explanations

key aspects and their context

The neuron identifies academic or technical terms that signal a central point or important concept in the text.

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

and

-1.46

 المختلفة

-1.26

 Haben

-1.20

 werdet

-1.11

 Including

-1.10

分から

-1.08

 représenté

-1.08

 mejoras

-1.08

𝑊

-1.07

 attribué

-1.05

POSITIVE LOGITS

of

3.59

for

2.14

 that

1.93

to

1.77

in

1.64

 anyone

1.41

you

1.41

 charmant

1.39

 vives

1.37

 çünkü

1.35

Activations Density 0.079%