INDEX

Explanations

initial steps or actions

The neuron fires on words that introduce or order steps in procedural text (e.g. “must,” “first,” “then,” “relate”).

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

Selanjutnya

-1.20

 only

-1.16

oczes

-1.13

༘

-1.11

 also

-1.10

Selain

-1.06

一会

-1.05

 conclusione

-1.05

aksanaan

-1.03

lifies

-1.03

POSITIVE LOGITS

まずは

1.30

 first

1.23

 zunächst

1.20

 basics

1.15

ことから

1.13

부터

1.12

 baseline

1.09

まず

1.05

 ersten

1.04

 πρώ

1.04

Activations Density 0.025%