specific types of single characters or short sequences that likely relate to code or identifiers

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Embeds

IFrame

Link

Negative Logits

orem

-0.11

-0.10

notated

-0.09

-0.08

eos

-0.07

-Agent

-0.07

POSITIVE LOGITS

bidden

0.10

/english

0.07

inki

0.07

alach

0.07

/or

0.07

eron

0.07

/o

0.07

terior

0.07

bsites

0.07

Ð²ÐµÐ´Ð¸ÑĤÐµ

0.07

Activations Density 0.164%

No Known Activations