INDEX

Explanations

information and specific outcomes

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

 ìŀĪê³ł

-0.13

ä¸¦

-0.11

ìĿ´ìŀĲ

-0.11

ë¥´ê³ł

-0.10

ìĿ´ê³ł

-0.10

å¹¶

-0.10

ToF

-0.09

ëĲĺê³ł

-0.09

íķĺê³ł

-0.09

ìĿ´ëĤĺ

-0.09

POSITIVE LOGITS

&E

0.12

etc

0.10

ï¼ĮåĲĮæĹ¶

0.09

&B

0.09

 tang

0.09

0.08

 Shades

0.08

umer

0.08

 SLOT

0.08

Activations Density 0.300%