INDEX

Explanations

so do, far, can, too

This neuron is a strong detector for the discourse marker “so” (especially in constructions like “so do …” or “so far …”) indicating a result, emphasis, or parallel comparison.

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

 also

-1.55

 آموزشی

-1.44

甞

-1.41

我們

-1.36

 instituto

-1.32

INARY

-1.31

 cebola

-1.30

 muñeco

-1.27

 επίσης

-1.26

aien

-1.26

POSITIVE LOGITS

1.60

1.58

了一些

1.40

fahan

1.38

 they

1.38

 ongeveer

1.37

 cuándo

1.37

。

1.33

actéristiques

1.33

).

1.33

Activations Density 0.005%