INDEX

Explanations

negation

The neuron fires on negation or negative-polarity function words (e.g. “no,” “not,” “don’t,” “never”).

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

but

-2.30

 ولكن

-1.74

 لكن

-1.71

 however

-1.58

 però

-1.56

 लेकिन

-1.52

 nhưng

-1.49

 แต่

-1.46

 אבל

-1.38

もありますが

-1.32

POSITIVE LOGITS

 this

1.48

1.30

1.16

 that

1.03

as

0.99

 will

0.98

 until

0.96

 insuffisamment

0.95

 gracilis

0.93

so

0.93

Activations Density 0.050%