INDEX

Explanations

describing factual or usage scenarios

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

 -*-č\n

-0.11

.Formatter

-0.11

 Ð¾Ð³ÑĢÐ°

-0.10

łéĻ¤

-0.10

EMPLARY

-0.10

³ç´°

-0.09

 BITTE

-0.09

 Ð¼ÐµÑĤÐ°Ð»Ð»Ð¸

-0.09

<|begin_of_text|>

-0.08

nitÅĻ

-0.08

POSITIVE LOGITS

)const

0.08

0.07

 Carter

0.07

iras

0.07

als

0.07

 Type

0.07

util

0.07

hawk

0.07

Activations Density 0.250%