INDEX

Explanations

what listeners can expect

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

Tir

-0.10

Abb

-0.09

 Bren

-0.09

ogne

-0.09

æĤ

-0.09

rollo

-0.09

ibbon

-0.09

ï½¢

-0.09

sass

-0.09

tir

-0.08

POSITIVE LOGITS

 expect

0.43

 Expect

0.36

expect

0.35

Expect

0.33

 expects

0.31

.expect

0.28

 expectation

0.27

 EXPECT

0.27

 expecting

0.25

_expect

0.25

Activations Density 0.062%