INDEX

Explanations

programming constructs

The neuron fires on identifier‐like tokens—names of CSS IDs and classes (e.g. “#igs”, “.ig”), variable or function names (e.g. “kk”, “ik”, “ps”, “FormData”), and similar symbol sequences.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

↵↵

-2.36

-2.16

”

-1.93

 which

-1.75

_{

-1.71

 umfang

-1.70

 isolada

-1.70

<em>

-1.63

 provide

-1.62

 presumably

-1.61

POSITIVE LOGITS

2.20

?"

2.13

 pernic

1.95

 abbildung

1.80

檮

1.77

aimeJ

1.76

'=>

1.74

 entram

1.71

 emerg

1.70

 specif

1.67

Activations Density 0.108%