INDEX

Explanations

united states

This neuron activates on non-ASCII/high-byte characters (e.g. odd Unicode/control-code bytes) in the text.

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

Даже

-1.93

</>

-1.91

Люби

-1.90

用

-1.88

 kuan

-1.87

</h1>

-1.84

:'',

-1.83

 kazi

-1.83

Παρα

-1.77

也有

-1.77

POSITIVE LOGITS

3.19

 takže

2.30

</

2.27

 pretože

2.11

 eenvou

2.03

}{

1.95

(“

1.93

 firmy

1.88

倏

1.87

 Quando

1.85

Activations Density 0.001%