INDEX

Explanations

astronomical and computational contexts

This neuron fires on high-level structural labels in a document—e.g. section or heading titles like “Introduction,” “abstract,” “Description,” etc.

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

</h2>

-2.27

鱨

-1.87

```

-1.80

踯

-1.78

ur

-1.72

-1.71

quot

-1.69

一惊

-1.68

᠂

-1.68

</h5>

-1.67

POSITIVE LOGITS

Now

1.92

 Clearly

1.90

 Before

1.82

 With

1.77

 conquistar

1.73

 Projektu

1.72

 volverá

1.72

 sekarang

1.70

ſelves

1.69

 kleinen

1.68

Activations Density 0.048%