INDEX

Explanations

ever seen

The neuron fires on the past-participle “seen,” especially in constructions expressing prior experience (e.g. “never seen,” “ever seen”).

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

</em>

-1.78

 других

-1.56

 semelh

-1.49

鸰

-1.45

 trám

-1.43

 AGAINST

-1.39

 várias

-1.38

 Какие

-1.31

ira

-1.31

.”

-1.31

POSITIVE LOGITS

1.55

綠

1.41

 monst

1.30

 年代

1.27

 Napoléon

1.25

("")

1.23

('')

1.23

 when

1.22

 Economia

1.21

 بعدی

1.20

Activations Density 0.010%