INDEX

Explanations

truth and correctness

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ä¸įå°½

-0.09

æĭĪ

-0.09

åīįä¸ĸ

-0.09

addle

-0.09

ensions

-0.09

 Dollars

-0.08

ç¼£

-0.08

 bore

-0.08

åĲĦæł·

-0.08

åīįæľŁ

-0.08

POSITIVE LOGITS

éĿ¢è²Į

0.12

éĿ¢çĽ®

0.11

ãģ»ãģĨ

0.10

çľŁå®ŀçļĦ

0.10

WindowSize

0.09

ä¹ĸä¹ĸ

0.09

yer

0.09

å¤

0.09

çĹħæĥħ

0.09

æĴ¬

0.09

Activations Density 0.058%