INDEX

Explanations

code closing tags

The neuron is triggered by numeric tokens—i.e. any literal sequence of digits (IDs, numeric array indices, hard-coded numbers, etc.).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 here

-1.04

-0.98

even

-0.97

but

-0.91

due

-0.89

—

-0.89

 even

-0.88

 آمریک

-0.88

ówek

-0.88

POSITIVE LOGITS

 verschill

1.00

čni

0.97

twas

0.93

 celebrado

0.92

 Spanyol

0.91

водом

0.91

 gezegd

0.90

 měl

0.90

 iniciado

0.90

OOTDTY

0.89

Activations Density 0.016%