INDEX

Explanations

handle and spout

The neuron is essentially lighting up on words that occur inside quoted speech passages, i.e. it detects tokens that are part of dialogue.

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

 reimag

0.84

 dystopian

0.84

 transformative

0.78

 familial

0.78

 базо

0.77

 amplify

0.73

 holographic

0.73

巨大

0.72

 juxt

0.71

✴

0.71

POSITIVE LOGITS

 fellows

0.84

 whiskers

0.74

 Incidentally

0.72

 Charley

0.70

 lousy

0.70

 Reddy

0.69

 youngster

0.69

 Backbone

0.68

 grin

0.68

 grinned

0.66

Activations Density 0.142%