INDEX

Explanations

degree of understanding or combination

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

 tops

-0.09

ãĤ·ãĤ¢

-0.09

*}

-0.09

ocre

-0.08

 trav

-0.08

oplay

-0.08

usher

-0.08

 bases

-0.08

 Cult

-0.08

SETS

-0.08

POSITIVE LOGITS

lot

0.19

 combination

0.18

 understanding

0.17

combination

0.16

 degree

0.14

 Combination

0.13

 amount

0.12

 level

0.12

 Understanding

0.12

 commitment

0.12

Activations Density 0.035%