INDEX

Explanations

extent, amount, degree

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

 Ð±ÑĥÐ¼Ð°

-0.15

 Ð·Ð°ÑıÐ²

-0.13

 Ð½ÐµÐ´ÐµÐ»

-0.12

 Ð¿Ð¾Ð±Ð°

-0.12

 Ð¸Ð¼Ñĥ

-0.12

 Ð¸Ð½Ð´Ð¸Ð²Ð¸Ð´Ñĥ

-0.12

 ÐµÑģÑĤÐµ

-0.11

 ÐŀÐ´Ð½Ð°

-0.11

 ÐºÑĥÐ»ÑĮÑĤÑĥ

-0.11

ÑĢÐ¾Ð½Ð¸ÑĩÐµÑģ

-0.11

POSITIVE LOGITS

 type

0.13

 amount

0.13

 degree

0.12

 nature

0.12

âĦĸâĦĸ

0.12

 level

0.11

 peculiar

0.11

 extent

0.11

 manner

0.11

 size

0.10

Activations Density 0.135%