INDEX

Explanations

adding in

This neuron primarily detects the standalone preposition “in” (and to a lesser extent similar short prepositions).

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

-1.45

今度は

-1.33

砚

-1.33

 OBSERVATIONS

-1.33

 mereka

-1.32

ophenol

-1.32

usa

-1.31

↵

-1.30

になるので

-1.28

跄

-1.27

POSITIVE LOGITS

ch

1.49

</sub>

1.46

 просто

1.32

um

1.27

 препят

1.26

口气

1.23

就很

1.23

 voidaan

1.23

นาด

1.20

⽉

1.20

Activations Density 0.076%