INDEX

Explanations

class definitions after curly braces

The neuron fires on programming‐style syntax tokens—things like punctuation (colons, commas, equals signs, parentheses), language keywords (var, class), and other code symbols and identifiers.

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

licante

-1.34

rugu

-1.16

 textu

-1.11

tividade

-1.11

‌.

-1.08

 cantora

-1.07

 særlig

-1.07

breviations

-1.05

on

-1.05

 Câmara

-1.05

POSITIVE LOGITS

odyne

1.16

erobic

1.09

ups

1.04

glected

1.02

key

1.00

テム

0.98

ases

0.98

bies

0.97

 предме

0.97

cestor

0.97

Activations Density 0.013%