INDEX

Explanations

source of the or idea for

The neuron is keyed to general content-bearing words (common nouns, verbs, and similar mid-frequency tokens) rather than empty function words.

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

ческая

-1.34

蔟

-1.34

耜

-1.30

 kapturem

-1.24

躂

-1.24

鈁

-1.23

 isolada

-1.23

 majest

-1.19

螣

-1.18

ᐧ

-1.17

POSITIVE LOGITS

 that

1.57

on

1.55

now

1.52

 from

1.50

One

1.31

Now

1.30

are

1.26

for

1.24

 last

1.23

Though

1.23

Activations Density 0.034%