fostering continuous learning or peer feedback

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

ilio

-0.09

aire

-0.08

 Certif

-0.08

ijo

-0.08

-ing

-0.08

[__

-0.08

 longitud

-0.08

entimes

-0.08

rible

-0.08

POSITIVE LOGITS

key

0.10

ă

0.10

 conv

0.09

''

0.09

0.08

CÄ°

0.08

à¸Ħà¹Ĥà¸Ļ

0.08

rosso

0.08

 Ð¾ÑĤÐ»Ð¸

0.08

Activations Density 0.076%

No Known Activations