INDEX

Explanations

lists joined by 'and' or 'or'

The neuron fires on concrete, domain-specific keywords that name essential functions, measures, or parameters (e.g. blood, rhythm, salinity, polynomials, potential, evidence).

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

-1.87

 also

-1.70

.)

-1.67

 naudo

-1.54

 gewährleisten

-1.52

 nuove

-1.50

 then

-1.49

ͬ

-1.45

кож

-1.42

他們的

-1.41

POSITIVE LOGITS

and

1.51

 мы

1.47

 یکی

1.45

we

1.42

 wysokość

1.38

1.30

我們

1.30

 giovan

1.30

gemäß

1.28

劻

1.28

Activations Density 0.160%