INDEX

Explanations

function arguments

The neuron fires on code‐comment markers (lines beginning with “//” or “#”)—i.e. it detects comment prefixes.

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

ing

0.58

<strong>

0.55

mi

0.55

<b>

0.52

mn

0.52

ig

0.52

ler

0.52

,…

0.52

ball

0.52

POSITIVE LOGITS

TODO

1.20

 NOTE

1.19

 FIXME

1.15

 TODO

1.14

 Örneğin

1.07

 Example

1.02

NOTE

1.01

 Note

0.95

 Например

0.95

 Ejemplo

0.92

Activations Density 0.010%