INDEX

Explanations

enhance the quality

The neuron fires on verbs that signal boosting or improvement (e.g. “enhance,” “facilitate”).

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



-2.20

很想

-2.16

atecas

-2.13

なくなる

-2.03

皷

-1.98

 ſuch

-1.98

驺

-1.98

睒

-1.96

 abhängig

-1.96

 относится

-1.96

POSITIVE LOGITS

and

2.34

2.23

–

2.08

 yang

2.00

…..

1.99

ments

1.97

 wich

1.91

ꠥ

1.91

cayaan

1.91

 touted

1.89

Activations Density 0.014%