INDEX

Explanations

often considered

The neuron fires on frequency/evaluation words—especially the adverb “often” (and related evaluative verbs like “considered”).

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

 possibly

1.06

 möglicherweise

1.06

 conceivably

1.04

 potencialmente

0.99

 posiblemente

0.97

 teilweise

0.96

 Possibly

0.96

 potrebbero

0.94

 могла

0.93

 potentially

0.92

POSITIVE LOGITS

 бывает

1.30

 suele

1.17

bý

1.13

 suelen

1.03

 accompanies

0.98

 থাকে

0.96

 overlooked

0.94

Either

0.93

 اوقات

0.89

 either

0.87

Activations Density 0.551%