INDEX

Explanations

years after apostrophes

The neuron detects numeric tokens immediately following an apostrophe (such as abbreviated years or similar ’NN or ’YYYY patterns).

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

after

-2.38

There

-2.27

 There

-2.16

there

-2.11

-2.03

and

-2.02

Often

-2.00

 create

-1.97

by

-1.96

if

-1.96

POSITIVE LOGITS

苤

2.19

 trám

2.09

 باسلام

1.98

臯

1.87

 Dinas

1.85

 rám

1.85

鲠

1.82

 astounding

1.81

湏

1.81

蟶

1.80

Activations Density 0.034%