INDEX

Explanations

numbers and codes

The neuron is detecting numeric tokens (sequences of digits, including multi‐digit numbers and decimals).

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

.”

-2.25

are

-2.19

𖣘

-2.13

 determinados

-2.08

you

-2.08

$};

-2.06

 ciertos

-2.03

-2.00

-1.98

 pracę

-1.98

POSITIVE LOGITS

в

2.14

 They

2.03

2.00

1.98

んですが

1.95

⬘

1.93

 alimentar

1.83

櫚

1.82

 风格

1.80

۱

1.78

Activations Density 0.071%