INDEX

Explanations

Q:

The neuron is selectively detecting tokens that contain numbers (numeric sequences).

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

 fön

-1.38

になりたい

-1.28

 pasas

-1.27

 köp

-1.23

妹

-1.21

 turista

-1.20

 pás

-1.20

 trö

-1.18

 espir

-1.15

 every

-1.13

POSITIVE LOGITS

也

1.29

1.23

 zweier

1.21

 They

1.20

 считает

1.13

 ovvero

1.12

 ardent

1.11

 vieler

1.10

 windmill

1.10

LAX

1.09

Activations Density 0.000%