INDEX

Explanations

technical data and structures

This neuron detects programming or code‐related identifier words (like Data, Class, Template, structure, Event).

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

 selben

-2.13

 distraught

-2.09

 produite

-1.98

 détaillée

-1.92

如果

-1.81

內容簡介

-1.81

鹚

-1.81

 Таким

-1.80

 てき

-1.73

 disgraceful

-1.73

POSITIVE LOGITS

1.88

↵↵

1.87

1.80

数据的

1.77

1.67

鬟

1.61

 rests

1.60

↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵↵

1.59

\%,

1.59

 cubana

1.57

Activations Density 0.037%