INDEX

Explanations

past tense verbs and states

The neuron activates on past‐participle (or passive/perfect) verb forms (e.g. “known,” “granted,” “connected,” “classified,” “shown,” “created”).

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

 prision

-1.36

ği

-1.24

下方

-1.23

 instead

-1.20

</th>

-1.18

鸷

-1.17

PARED

-1.16

نى

-1.16

émoc

-1.16

it

-1.15

POSITIVE LOGITS

非常的

1.57

 fragte

1.43

The

1.43

 antwortete

1.39

 wußte

1.39

 lebte

1.38

 glaubte

1.36

 singuli

1.30

 typu

1.29

 machte

1.27

Activations Density 0.075%