INDEX

Explanations

Main method

The neuron fires on programming‐language type tokens and symbols (e.g. C declarations like “char,” “*,” “argv”).

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

{(

0.69

={

0.67

=(

0.62

APP

0.61

APP

0.60

 agonist

0.60

App

0.59

 AppException

0.58

 `=`,

0.57

 indexed

0.56

POSITIVE LOGITS

[]

1.10

[]

0.93

[]);

0.82

[])

0.76

[],

0.68

[])

0.67

Gary

0.67

[];

0.66

[]');

0.66

Ха

0.65

Activations Density 0.000%