INDEX

Explanations

academic research and services

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Ches

-0.07

spr

-0.06

 hors

-0.06

eus

-0.06

 scrap

-0.06

 taper

-0.05

 once

-0.05

difficulty

-0.05

 during

-0.05

rap

-0.05

POSITIVE LOGITS

gaard

0.14

aign

0.12

airo

0.12

Nag

0.12

fas

0.12

 Produk

0.12

Workspace

0.12

extensions

0.11

 wisely

0.11

sville

0.11

Activations Density 0.004%