INDEX

Explanations

and, after, just

The neuron is keyed to unusually long or low-frequency tokens—i.e. rare, domain-specific, or morphologically complex 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

াদিগের

0.82

 Эти

0.77

შვ

0.74

क्ष्म

0.72

 handwriting

0.72

 दस

0.72

 Tetapi

0.70

damageCount

0.70

ấu

0.70

 loneliness

0.70

POSITIVE LOGITS

in

0.92

with

0.89

by

0.87

for

0.87

seits

0.85

which

0.82

),

0.80

 which

0.77

by

0.77

for

0.76

Activations Density 1.414%