INDEX

Explanations

that/which performing an action

The neuron strongly activates on occurrences of the relative pronoun “which.”

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

 that

-2.14

Another

-1.98

Also

-1.72

Almost

-1.71

Other

-1.69

 giapponese

-1.68

 jepang

-1.64

There

-1.63

 prodot

-1.63

櫻

-1.63

POSITIVE LOGITS

1.86

who

1.84

ilisation

1.78

from

1.66

 know

1.62

And

1.56

сяг

1.55

żeń

1.49

1.46

jn

1.46

Activations Density 0.057%