INDEX

Explanations

essential role or state

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

-0.10

actual

-0.09

 actual

-0.09

 Deprecated

-0.09

umer

-0.09

-0.08

DN

-0.08

 fossils

-0.08

Actual

-0.08

ellig

-0.08

POSITIVE LOGITS

 normally

0.19

èµĸ

0.14

normally

0.14

 essential

0.14

 integral

0.13

egral

0.13

integral

0.13

 vital

0.13

Normally

0.12

 critical

0.12

Activations Density 0.145%