INDEX

Explanations

summarize text, otherwise mundane

The neuron fires strongly on verbs that denote transforming or converting something (“turn,” “transform,” “convert,” etc.), especially in “X into Y” constructions.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

様に

0.55

 একমাত্র

0.54

 دھو

0.53

理由は

0.52

ദം

0.51

 সুল

0.51

만큼

0.50

劾

0.50

ัล

0.50

ieważ

0.50

POSITIVE LOGITS

 ordinary

1.23

 mundane

1.18

 otherwise

1.16

原本

1.16

otherwise

1.16

 Ordinary

1.13

 dull

1.11

Ordinary

1.11

 boring

1.10

 drab

1.09

Activations Density 0.426%