INDEX

Explanations

page number then name

This neuron strongly activates on numeric tokens that serve as volume or page numbers in legal case citations.

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

 rusak

-1.29

Usually

-1.12

 refrigerador

-1.12

 listView

-1.12

 onStart

-1.09

ghz

-1.09

веду

-1.08

 abrigo

-1.08

 Mexiko

-1.06

сели

-1.05

POSITIVE LOGITS

↵↵↵

1.18

 vigueur

1.08

oczes

1.06

などを

1.03

 genügend

1.02

 аккумулятор

1.01

 totalité

1.01

 fiasco

0.98

↵↵

0.98

>()

0.98

Activations Density 0.002%