INDEX

Explanations

usage context and conditions

The neuron consistently activates on the word “use” (and its close variants like “used” or “reuse”) throughout the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

to

-2.06

氅

-1.45

gunas

-1.44

 semelh

-1.43

the

-1.40

 intelek

-1.33

 recours

-1.32

 pomocą

-1.32

 kaikki

-1.30

POSITIVE LOGITS

 with

2.02

When

1.64

Even

1.63

Like

1.48

While

1.47

During

1.41

With

1.41

Although

1.37

At

1.32

軼

1.30

Activations Density 0.081%