INDEX

Explanations

IoU loss

The neuron detects standalone floating‐point numbers (numeric tokens with decimal points).

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

增强

0.77

또

0.75

 Otra

0.70

ຈ

0.69

 ومن

0.68

另一

0.67

 JavaBeans

0.66

 bolstering

0.66

 Another

0.66

 לק

0.65

POSITIVE LOGITS

SUR

0.73

0.71

SIP

0.66

EK

0.65

Tb

0.65

SPL

0.65

SIN

0.63

SLE

0.63

SUB

0.63

TRI

0.62

Activations Density 0.160%