INDEX

Explanations

checking condition

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

è

-0.10

ï½¿

-0.10

 Huff

-0.09

agne

-0.09

 correctness

-0.09

 Initialise

-0.08

 invalid

-0.08

 Schiff

-0.08

adia

-0.08

atin

-0.08

POSITIVE LOGITS

 check

0.56

 checks

0.50

 checking

0.44

 Check

0.43

check

0.42

 Ð¿ÑĢÐ¾Ð²ÐµÑĢ

0.39

Check

0.39

æ£ĢæŁ¥

0.38

-check

0.38

 Checks

0.37

Activations Density 0.082%