INDEX

Explanations

components connected

The neuron identifies chunks of technical specification text—especially numeric values, units, and related machinery or production‐line terminology.

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

◾

-0.98

krieg

-0.92

͟

-0.90

֠

-0.88

might

-0.88

 zvuky

-0.86

ikti

-0.85

 finans

-0.85

Ľ

-0.85

ustain

-0.84

POSITIVE LOGITS

 adopts

1.37

 adopt

1.07

 working

1.07

 adopting

0.97

 adoption

0.94

 qualified

0.94

 rationality

0.92

 unqualified

0.91

 allowable

0.89

 requirements

0.88

Activations Density 0.085%