INDEX

Explanations

help you

Explanation of neuron 4 behavior: the main thing this neuron does is find offers of help or assistance (e.g. “help,” “assist,” “answer any questions”).

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

the

-1.67

 และ

-1.45

mberto

-1.29

ladesh

-1.27

 maintenant

-1.27

硖

-1.27

ਝ

-1.26

 שה

-1.23

beton

-1.22

 funkce

-1.20

POSITIVE LOGITS

雖然

1.78

1.63

1.61

1.55

We

1.54

To

1.52

Of

1.52

<h2>

1.51

虽然

1.48

ども

1.45

Activations Density 0.016%