INDEX

Explanations

run followed by context

The neuron strongly activates on the token “run” (and its inflected forms like “running”), i.e. it detects mentions of “run” in the text.

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

if

-1.52

ト

-1.52

when

-1.49

褔

-1.47

 этими

-1.45

 cementerio

-1.43

</h4>

-1.42

出来た

-1.39

出来ました

-1.39

Bissau

-1.29

POSITIVE LOGITS

euvres

1.64

ﻂ

1.61

 seleccionadas

1.61

🏃

1.57

RUN

1.57

RUN

1.50

过程中

1.48

 appri

1.48

incón

1.47

 verletzt

1.47

Activations Density 0.077%