INDEX

Explanations

for / in phrases

The neuron strongly activates on uncommon proper nouns and other named entities (people’s names, brands, places, 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

 которые

0.95

 suas

0.89

but

0.88

dew

0.87

 bericht

0.87

 durch

0.85

 cloak

0.85

ার্স

0.85

 übers

0.84

 which

0.84

POSITIVE LOGITS

READY

1.02

ुल

0.98

enció

0.91

략

0.90

لا

0.89

র্গে

0.87

現在

0.86

PAssignment

0.86

стра

0.85

রাপ

0.85

Activations Density 0.473%