INDEX

Explanations

data field separator

This neuron responds to rare or specialized tokens—especially acronyms, hyphenated terms, and alphanumeric strings (e.g. “PD-1,” “first-,” “second-,” “PD-L1,” “pro,” etc.).

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

 ます

0.61

 کمپنیوں

0.61

ۇ

0.61

 अल्कोहल

0.61

঵

0.60

derabad

0.59

ቛ

0.59

rystall

0.59

hentication

0.59

POSITIVE LOGITS

以外の

0.75

 যেহেতু

0.61

 আগে

0.60

 আরেক

0.56

才是

0.55

다른

0.55

 anyway

0.54

 maupun

0.53

 остальные

0.51

Anyway

0.51

Activations Density 0.495%