INDEX

Explanations

omit explanation

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

-âĢĲ

-0.09

alic

-0.09

 Dank

-0.09

rippling

-0.09

VERSE

-0.09

 onAnimation

-0.08

inati

-0.08

owitz

-0.08

CLU

-0.08

 harmon

-0.08

POSITIVE LOGITS

nor

0.20

 unless

0.16

 anymore

0.16

 sondern

0.14

nor

0.14

 except

0.13

 Unless

0.12

unless

0.12

NOR

0.12

or

0.12

Activations Density 0.076%