INDEX

Explanations

setting controls

The neuron fires on standalone content nouns or key technical terms (e.g. “game,” “family,” “sheet,” “system,” “considered”), effectively picking out topic-word labels in the text.

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

Which

-1.42

Some

-1.35

by

-1.33

ött

-1.32

says

-1.27

 bijvoorbeeld

-1.26

Both

-1.23

 algunas

-1.23

 构造

-1.22

อร่อย

-1.22

POSITIVE LOGITS

mètres

1.43

=$("#

1.34

 років

1.27

Географи

1.26

iseksi

1.23

 generale

1.22

omination

1.21

нью

1.21

INTAIN

1.20

一颗

1.19

Activations Density 0.196%