INDEX

Explanations

`estaba est aba era aba`

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

 Ø¯Ø§Ø±Ùħ

-0.13

arlar

-0.12

 olmadÄ±

-0.11

Ã¡zÃŃ

-0.11

å¤ļãģĦ

-0.11

 Ø¯Ø§Ø±ÛĮÙħ

-0.11

ÑĭÐ²Ð°ÑİÑĤÑģÑı

-0.11

ÄĽlÃŃ

-0.10

ÑĥÑĶÑĤÑĮÑģÑı

-0.10

Ð´Ð°ÐµÑĤÑģÑı

-0.10

POSITIVE LOGITS

ãģĹãģ¦ãģĦãģŁ

0.26

ãģ£ãģ¦ãģĦãģŁ

0.25

 Ã©tait

0.24

ÑĭÐ²Ð°Ð»

0.24

ãģ¦ãģĦãģŁ

0.24

was

0.22

 estaba

0.22

aba

0.22

ava

0.22

era

0.21

Activations Density 0.142%