INDEX

Explanations

accumulation of fat and changes

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¿Ð¾Ð±Ð°

-0.21

 Ð´Ð¸Ð²Ð¸

-0.17

 Ð¾Ð³ÑĢÐ°

-0.15

 ÐµÑģÑĤÐµ

-0.14

 Ð½ÐµÐ´ÐµÐ»

-0.12

 Ð±ÑĥÐ¼Ð°

-0.12

 ÑĤÐ²Ð°

-0.12

 Ð·Ð°ÑıÐ²

-0.11

 chtÄĽ

-0.11

 Ð¸Ð¼Ñĥ

-0.10

POSITIVE LOGITS

âĦĸâĦĸ

0.18

Ð½Ð°Ð´Ð»ÐµÐ¶

0.13

tk

0.10

 pulp

0.09

 Lever

0.09

 trom

0.09

Lor

0.09

ÌĨ

0.09

 folk

0.09

Ð°

0.09

Activations Density 0.115%