INDEX

Explanations

algebraic variables and punctuation

The neuron is essentially a “number detector” – it activates on any token containing digits (numeric literals or identifiers with numeric subscripts).

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

 said

-1.25

 }}=\

-1.20

 SHOWS

-1.19

 другой

-1.19

ordat

-1.18

小孩子

-1.16

 sogen

-1.13

ську

-1.12

のですか

-1.12

 THROUGH

-1.10

POSITIVE LOGITS

and

2.56

are

1.66

etc

1.65

natured

1.16

 here

1.04

mahami

1.03

นี้

1.02

 should

1.02

 have

1.00

 Türkei

1.00

Activations Density 0.047%