INDEX

Explanations

left-to-right

The neuron strongly fires on tokens that name markup roles, text‐direction or encoding specs (e.g. “presentation”, “direction”, “ltr”, “utf16”, Unicode codepoint literals).

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

the

-1.91

-1.45

it

-1.43

solicited

-1.41

 joga

-1.31

-1.30

ığımız

-1.30

блица

-1.27

 viata

-1.20

 lots

-1.20

POSITIVE LOGITS

laient

1.35

1.34

躂

1.33

сті

1.30

)$

1.29

fillText

1.26

}}$

1.25

讳

1.24

 Problemas

1.23

 valentine

1.21

Activations Density 0.010%