INDEX

Explanations

learn, understand, find

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 alho

-1.05

下意识

-1.04

 дър

-1.04

 Tél

-1.04

 dunia

-1.02

 nació

-1.00

婳

-1.00

 preved

-0.98

ángulo

-0.97

mbps

-0.97

POSITIVE LOGITS

 learn

3.53

see

3.14

 hear

2.97

get

2.59

 receive

2.52

 understand

2.52

 discover

2.30

 find

2.22

 learns

2.14

learn

2.11

Activations Density 0.088%