INDEX

Explanations

language or category labels

The neuron fires on document metadata labels and section headers (e.g. field names like “Manufacturer,” “Condition,” “Category,” “Java,” etc.) rather than ordinary running 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

for

-1.47

 this

-1.31

by

-1.27

at

-1.24

on

-1.21

 even

-1.20

 their

-1.15

Já

-1.09

 with

-1.09

 from

-1.09

POSITIVE LOGITS

 áng

1.20

various

1.16

 obviamente

1.12

Forked

1.10

 arrêter

1.10

デア

1.09

 diversos

1.07

 nhung

1.04

Slik

1.03

 cucharadita

1.02

Activations Density 0.036%