INDEX

Explanations

references to numerical precision and error in computations

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

oxid

-0.06

 Mile

-0.06

/connect

-0.06

 ÙħÛĮÙĦ

-0.06

cron

-0.06

Roll

-0.06

uibModal

-0.05

$MESS

-0.05

 Morph

-0.05

ossil

-0.05

POSITIVE LOGITS

 IEEE

0.16

 floating

0.15

 Floating

0.13

 rounding

0.13

IEEE

0.13

 ieee

0.13

 precision

0.12

Floating

0.12

 mant

0.12

floating

0.12

Activations Density 0.014%