INDEX

Explanations

debug configuration

The neuron consistently fires on the token “Debug,” i.e. occurrences of the word “Debug” in code or configuration settings.

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>

-2.97

It

-2.95

⮑

-2.66

 delineated

-2.63

</h4>

-2.63

');

-2.48



-2.47

 brimming

-2.45

 reunió

-2.44

 attraverso

-2.42

POSITIVE LOGITS

鬭

3.00

2.86

and

2.86

涐

2.67

’

2.55



2.38

猡

2.28

嗞

2.28

2.25

on

2.23

Activations Density 0.003%