INDEX

Explanations

based on context

The neuron is detecting very common, high‐frequency function words and simple adverbs (i.e. “stop” words like in, to, by, only, any, too, first, 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

幫忙

-1.59

说得

-1.43

有种

-1.42

 besseren

-1.40

攙

-1.38

馬上

-1.38

 einzigen

-1.38

 étan

-1.38

dersfield

-1.38

jando

-1.38

POSITIVE LOGITS

 where

1.44

at

1.42

 corolla

1.30

 which

1.30

itattu

1.26

iaal

1.26

eterangan

1.25

in

1.25

粒

1.22

kyr

1.21

Activations Density 0.235%