INDEX

Explanations

programming keywords and operations

The neuron is highly sensitive to technical identifier tokens—component and property names, code attributes, and similar code‐style labels.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

<unused2169>

0.88

<unused2121>

0.85

 ваше

0.83

<unused2156>

0.82

૧

0.82

<unused2173>

0.81

۸

0.81

۴

0.80

<unused744>

0.80

<unused2137>

0.80

POSITIVE LOGITS

0.74

 Weather

0.74

Weather

0.72

Juven

0.69

Prior

0.68

 Massage

0.67

 Prior

0.67

Lever

0.65

0.64

Activations Density 0.007%