INDEX

Explanations

to find

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

Negative Logits

ove

-0.11

 Dexter

-0.11

 Alvarez

-0.09

 Authentic

-0.09

Elm

-0.09

 Delegate

-0.09

isp

-0.09

 EITHER

-0.08

dug

-0.08

POSITIVE LOGITS

 find

0.32

æī¾åĪ°

0.28

 finding

0.27

 finds

0.26

find

0.25

.find

0.23

 Find

0.22

Find

0.21

 finden

0.21

 encontrar

0.21

Activations Density 0.033%

No Known Activations