INDEX

Explanations

sequences of numbers

The neuron is essentially dead—it never activates and doesn’t detect any meaningful token pattern.

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

-1.52

the

-1.49

},

-1.48

 forskellige

-1.41

contentValues

-1.36

錮

-1.34

But

-1.34

 forskjellige

-1.34

&#

-1.32

 medlemmer

-1.32

POSITIVE LOGITS

 этому

1.59

 trámites

1.49

 pozost

1.48

This

1.45

 série

1.40

 prehistoric

1.40

。

1.39

 avions

1.38

 terutama

1.38

 bahkan

1.35

Activations Density 0.051%