INDEX

Explanations

listing or starts of lines

The neuron fires on tokens that are written entirely in uppercase (e.g. acronyms, constant names, or other all-caps words).

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

of

-2.31

and

-1.99

</h3>

-1.91

 during

-1.70

 using

-1.69

on

-1.68

なくても

-1.67

 because

-1.65

In

-1.64

-1.63

POSITIVE LOGITS

឵

1.95

 nombreuses

1.91

quée

1.86

猻

1.83

 opciones

1.80

 banheiro

1.79

⾒

1.77

薷

1.77

コチラ

1.73

 llorar

1.73

Activations Density 0.000%