INDEX

Explanations

adjectives and adverbs

The neuron fires on degree-modifying qualifiers (adverbs or adjectives) like “well-known,” “ever-increasing,” “widely publicized,” “relatively modest,” “mostly,” “strictly,” “highly,” etc.

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

万美元

-1.47

 cuando

-1.40

鲧

-1.40

ceased

-1.31

 trzech

-1.31

 verändert

-1.30

 свою

-1.30

 przede

-1.30

joyed

-1.29

POSITIVE LOGITS

Their

1.85

There

1.48

Your

1.40

Many

1.35

 will

1.33

animity

1.31

Often

1.30

Fußball

1.29

 mụn

1.28

 потрібно

1.27

Activations Density 0.241%