INDEX

Explanations

originality and enjoyment

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

 note

-0.09

 HOLDER

-0.09

è±

-0.09

 Powers

-0.09

 particularly

-0.08

 Animalia

-0.08

.cls

-0.08

zion

-0.08

POSITIVE LOGITS

 relax

0.14

FUN

0.13

fun

0.12

 experiment

0.11

 enjoyment

0.11

 enjoyable

0.11

 enjoy

0.11

 relaxed

0.10

 comfortable

0.10

Fun

0.10

Activations Density 0.024%