INDEX

Explanations

extreme G, intense heat

New Auto-Interp

Top Features by Cosine Similarity

Configuration

Prompts (Dashboard)

10,000 prompts, 128 tokens each

Dataset (Dashboard)

lmsys/lmsys-chat-1m

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 twitch

-0.11

yp

-0.11

-0.10

inf

-0.09

 ambit

-0.09

.assertIsNot

-0.09

 distress

-0.09

 Collapse

-0.09

 puls

-0.09

POSITIVE LOGITS

 stomach

0.18

 centrif

0.16

 vert

0.14

 nausea

0.13

 feeling

0.12

 vestib

0.12

 ride

0.12

 Ride

0.11

ãĤĽ

0.11

 gravity

0.11

Activations Density 0.044%