INDEX

Explanations

level rise

The neuron flags content‐bearing, “meaningful” words (nouns, verbs, adjectives) and stays silent on common function words.

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

𝙎

1.02

𝘽

0.98

 fabricants

0.91

𝘾

0.90

вної

0.84

投注

0.84

𝙨

0.84

𝘿

0.84

𝙛

0.84

euros

0.83

POSITIVE LOGITS

 banding

0.80

 stacks

0.76

 Commission

0.71

 ideal

0.71

 ساز

0.70

 কাছে

0.68

banding

0.66

 Generator

0.66

↵↵↵

0.65

 ends

0.65

Activations Density 0.000%