INDEX

Explanations

code comments or questions

The neuron strongly activates on “#” markers used for code directives or regions (e.g. “#region,” “#pragma,” “#frozen”) in source files.

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

دانلود

-1.29

 تضيف

-1.27

Также

-1.25

そして

-1.25

Արտաքին

-1.20

Поэтому

-1.20

もあって

-1.20

และ

-1.18

 سایر

-1.18

توضیحات

-1.17

POSITIVE LOGITS

1.61

1.59

1.26

It

1.23

1.21

1.16

“

1.13

1.06

//

1.05

For

1.05

Activations Density 0.003%