INDEX

Explanations

price and availability

The neuron fires on parenthesized numeric details—i.e. numbers (and their surrounding parentheses) indicating measurements, counts, prices, times, etc.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

そして

-1.02

zun

-1.00

iguar

-1.00

 kader

-0.98

hadapi

-0.97

 Exteriores

-0.97

 grupp

-0.95

ָר

-0.95

 dür

-0.94

 kritis

-0.94

POSITIVE LOGITS

ほうがいい

1.31

no

1.30

方がいい

1.18

These

1.13

if

1.11

 These

1.09

ございません

1.07

 only

1.06

 doesn

1.04

わけではない

1.02

Activations Density 0.032%