INDEX

Explanations

the word "right" in various contexts

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

uzzi

-0.07

alm

-0.06

/al

-0.06

idal

-0.06

far

-0.06

indo

-0.06

eth

-0.06

åĲĲ

-0.06

 precision

-0.06

 Alice

-0.06

POSITIVE LOGITS

 immediately

0.07

 Away

0.07

 away

0.07

 Cabr

0.07

Away

0.07

 immediate

0.07

imity

0.07

zcze

0.06

rollers

0.06

olini

0.06

Activations Density 0.005%