INDEX

Explanations

error codes and technical terms

The neuron is primarily detecting tokens made up of digits (i.e. numeric sequences).

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

GIVEREF

-0.77

deja

-0.76

Становништво

-0.75

 Johansson

-0.75

ើ

-0.74

detective

-0.73

IZER

-0.72

unyai

-0.71

VEHICLE

-0.71

潜力

-0.69

POSITIVE LOGITS

 shooting

0.96

 sunk

0.89

 autos

0.82

sink

0.82

 sank

0.82

 weapons

0.82

 guns

0.82

 shoots

0.80

 shoot

0.79

 sinking

0.79

Activations Density 0.033%