INDEX

Explanations

what or how

This neuron activates on question words (e.g. “what,” “how,” etc.) in queries.

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

When

-2.64

There

-2.48

They

-2.34

This

-2.17

With

-2.09

While

-2.05

It

-2.05

What

-1.98

Before

-1.95

By

-1.93

POSITIVE LOGITS

躂

1.73

 насеља

1.66

rozum

1.62

</h1>

1.61

IPCC

1.61

and

1.60

 nejlepší

1.59

gâ

1.56

cinoma

1.55

 célèbre

1.55

Activations Density 0.026%