INDEX

Explanations

you might/may suggest

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

å¿ħé¡»

-0.10

alty

-0.10

 Wahl

-0.09

 MUST

-0.09

 INTERRU

-0.09

 must

-0.08

must

-0.08

Bitte

-0.08

QUIRES

-0.08

undry

-0.08

POSITIVE LOGITS

 might

0.44

might

0.36

 Might

0.35

 maybe

0.31

 perhaps

0.30

 Maybe

0.24

 vielleicht

0.23

à¸Ńà¸²à¸Ī

0.23

 Perhaps

0.23

maybe

0.23

Activations Density 0.099%