INDEX

Explanations

Japanese character names

The neuron is detecting numeric tokens (integer values) in the text.

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

ּוֹ

-1.58

 ベクトル

-1.20

 provide

-1.19

𝒙

-1.19

 RESPONDENT

-1.19

vaných

-1.14

резать

-1.10

 digunakan

-1.09

 offers

-1.06

for

-1.05

POSITIVE LOGITS

 attente

1.35

𝑞

1.30

 trastorno

1.30

COMO

1.23

 ouvido

1.23

öhn

1.21

 минера

1.16

صبح

1.16

 设计

1.15

edakan

1.15

Activations Density 0.000%