INDEX

Explanations

units of measurement

The neuron strongly activates on markup tag or code‐related tokens (e.g. HTML/XML tag names and inline code identifiers).

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

NUMX

-1.65

ApJ

-1.63

{~

-1.63

𝑌

-1.57

椟

-1.55

甚至

-1.54

говой

-1.52

 أن

-1.52

:[]

-1.52

zionata

-1.51

POSITIVE LOGITS

for

2.13

For

1.76

 either

1.75

 Since

1.67

 according

1.61

or

1.59

 However

1.58

at

1.55

詳細は

1.52

 except

1.51

Activations Density 0.001%