INDEX

Explanations

attached image or file

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

peg

-0.10

PPC

-0.09

 correlation

-0.08

 correlations

-0.08

peg

-0.08

eti

-0.08

 ration

-0.08

 prefect

-0.08

edor

-0.08

 corridor

-0.08

POSITIVE LOGITS

 attachment

0.13

 Thumbnails

0.12

 files

0.12

 Attachment

0.11

 screenshot

0.11

 Files

0.11

 screenshots

0.11

 image

0.11

 attached

0.11

 file

0.10

Activations Density 0.006%

No Known Activations