INDEX

Explanations

comparing and comparing ways

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

eneg

-0.11

 compliment

-0.10

undo

-0.09

æĶ»

-0.08

 Shen

-0.08

ousel

-0.08

asma

-0.08

 Corinth

-0.08

cth

-0.08

anela

-0.08

POSITIVE LOGITS

 comparison

0.35

 compare

0.30

 Comparison

0.30

 Compare

0.28

æ¯Ķè¾ĥ

0.27

 comparisons

0.26

Comparison

0.26

 compares

0.25

compare

0.25

 comparing

0.24

Activations Density 0.114%