INDEX

Explanations

references to silence 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

ery

-0.07

ito

-0.07

ÑĥÐ¶Ðµ

-0.07

lez

-0.07

ibbon

-0.07

{}",

-0.07

rien

-0.07

lite

-0.07

ached

-0.06

hra

-0.06

POSITIVE LOGITS

 treatment

0.08

ude

0.08

ascus

0.06

ourcem

0.06

 Schwe

0.06

gram

0.06

ogui

0.06

 deaf

0.06

OCK

0.06

 Silent

0.06

Activations Density 0.012%