INDEX

Explanations

code block syntax

The neuron strongly activates on the bracketed usage/synopsis lines of command‐line interface documentation (e.g. the “Usage: … [OPTION] [FILE] …” patterns).

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

(^

-0.97

myArray

-0.88

chose

-0.85

parque

-0.85

excellents

-0.84

Föld

-0.82

"=>$

-0.82

 végétale

-0.82

 zvíř

-0.79

𝗩

-0.79

POSITIVE LOGITS

2.02

1.17

,[

1.17

[-

1.13

[\

1.11

[{

1.10

[+

1.10

}[

1.06

langle

1.05

([

1.03

Activations Density 0.034%