INDEX

Explanations

large followed by description

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

linger

-0.11

ted

-0.09

ingle

-0.09

 Strikes

-0.09

tere

-0.09

ting

-0.09

big

-0.09

wal

-0.09

enticated

-0.08

 tiny

-0.08

POSITIVE LOGITS

-scale

0.31

-format

0.16

 scale

0.16

 enough

0.15

 Scale

0.14

-ish

0.14

scale

0.14

-bodied

0.13

 Enough

0.13

 intestine

0.13

Activations Density 0.061%