INDEX

Explanations

past tense descriptions

The neuron primarily picks out past‐tense verb forms (including contracted “’t” markers) indicating actions or states described in the past.

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

for

-2.00

in

-1.87

by

-1.78

but

-1.68

 because

-1.55

to

-1.50

 November

-1.46

 should

-1.45

of

-1.42

 October

-1.42

POSITIVE LOGITS

เรา

1.42

ihnach

1.37

 empê

1.34

 refroid

1.34

 répon

1.30

ioneers

1.30

 étan

1.29

menü

1.29

Վ

1.28

 réun

1.27

Activations Density 0.067%