INDEX

Explanations

specifically or examples

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

 dumb

-0.08

 prone

-0.08

ilyn

-0.08

ÏĦÎ¹ÏĤ

-0.08

iana

-0.08

pton

-0.08

eniable

-0.08

 Worm

-0.08

 accord

-0.08

POSITIVE LOGITS

 especially

0.24

 specifically

0.23

 applied

0.21

å¯¹äºİ

0.20

å°¤

0.20

éĴĪ

0.19

especially

0.18

 particularly

0.18

 relation

0.17

 Specifically

0.17

Activations Density 0.534%