INDEX

Explanations

years around 1900

The neuron activates on four‐digit year tokens (e.g., “188?”, “186?”, “185?”, etc.), i.e. dates expressed as years.

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

и

-1.71

AutoresizingMask

-1.70

 और

-1.69

ztes

-1.66

ваги

-1.66

ными

-1.64

宣言

-1.61

and

-1.59

重生

-1.55

 يه

-1.55

POSITIVE LOGITS

更何况

2.25

’

2.11

何况

2.02

 WHILE

1.91

/}.

1.89

饧

1.87

liśmy

1.87

 signified

1.80

 devising

1.78

獪

1.77

Activations Density 0.010%