INDEX

Explanations

describing shapes and structures

The neuron detects technical terms specifying geometric or dimensional parameters (e.g. width, height, shape, number of turns, wires) in scientific descriptions.

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

CARACTER

-1.60

 Kompon

-1.37

 alugar

-1.31

middag

-1.30

 правительства

-1.30

frontmatter

-1.29

komple

-1.29

 giocare

-1.29

läs

-1.25

merking

-1.23

POSITIVE LOGITS

 estekak

1.34

$\

1.25

zulegen

1.20

 просмотра

1.18

ڤ

1.17

 erzeugen

1.16

 выбран

1.16

1.14

 design

1.13

 gewählt

1.12

Activations Density 0.111%