INDEX

Explanations

focus on specific outcomes

The neuron fires on technical, research‐study and methodology terms—words like “outcomes,” “methods,” “trials,” “treatment,” “assessment,” and other section-heading style jargon.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 sometimes

-1.09

 into

-1.07

 between

-0.98

 show

-0.97

即便是

-0.96

 from

-0.94

ddling

-0.94

èbres

-0.93

atula

-0.93

ếng

-0.91

POSITIVE LOGITS

 piuttosto

0.99

 rather

0.98

 konkrét

0.96

 aspect

0.95

如何

0.95

highlighted

0.92

rather

0.91

 aspectos

0.91

 müş

0.91

 plutôt

0.90

Activations Density 0.058%