INDEX

Explanations

between, intermediate, or midway states

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

 Triple

-0.10

 Rath

-0.10

 Ending

-0.10

 triple

-0.09

Į¨

-0.09

Æ¡n

-0.09

ondo

-0.09

itel

-0.09

.vertex

-0.09

thinkable

-0.09

POSITIVE LOGITS

 intermediate

0.30

 Intermediate

0.26

Intermediate

0.23

 intermediary

0.23

 midway

0.22

 middle

0.21

 intermedi

0.19

 somewhere

0.18

 neither

0.18

 midd

0.17

Activations Density 0.322%