INDEX

Explanations

machine learning and readable

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

mania

-0.11

acom

-0.11

Ã¡zev

-0.10

tings

-0.10

pler

-0.09

entin

-0.09

lee

-0.09

 Paras

-0.09

itel

-0.09

/music

-0.09

POSITIVE LOGITS

-readable

0.20

inery

0.17

gun

0.17

 Ø¢ÙĦØ§Øª

0.16

 Learning

0.15

 learning

0.15

-gun

0.14

æ¢°

0.14

learning

0.14

Learning

0.13

Activations Density 0.033%

No Known Activations