INDEX

Explanations

phrases that indicate capacity for improvement or potential

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

avra

-0.07

StringRef

-0.06

redd

-0.06

Gee

-0.06

orthy

-0.06

ute

-0.06

 bomb

-0.06

eniz

-0.06

ATCH

-0.06

 hive

-0.06

POSITIVE LOGITS

 improvement

0.07

ull

0.07

 Improvement

0.07

iest

0.07

for

0.06

 Ð´Ð»Ñı

0.06

alus

0.06

side

0.06

lesc

0.06

Activations Density 0.010%