INDEX

Explanations

instructions and steps

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

brig

-0.09

osas

-0.09

-ln

-0.09

cko

-0.09

 Mend

-0.09

 redund

-0.09

gi

-0.09

 Cous

-0.08

 Reyn

-0.08

onas

-0.08

POSITIVE LOGITS

 steps

0.35

 Steps

0.30

Steps

0.27

steps

0.27

æŃ¥

0.24

 step

0.23

_steps

0.22

 ÑĪÐ°Ð³

0.21

 Step

0.21

 process

0.20

Activations Density 0.013%