INDEX

Explanations

writing to express interest

This neuron detects forum-style quote attributions, i.e. lines like “Username wrote:”.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

})_{\

0.64

 अधूरा

0.61

 বদ্ধ

0.59

说是

0.58

স্থার

0.57

 calibr

0.54

劳动

0.54

 zunehm

0.54

 préoccup

0.53

offsetWidth

0.53

POSITIVE LOGITS

hte

0.63

 cover

0.59

 Covered

0.59

 Cover

0.57

cover

0.57

ht

0.55

moss

0.54

Chart

0.53

composition

0.53

 Complex

0.52

Activations Density 0.001%