INDEX

Explanations

formatting commands and code

The neuron fires on POS‐tag tokens labeled “JJ,” i.e. adjective tags in the parsed 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

to

-1.83

 when

-1.58

two

-1.52

<bos>

-1.48

应当

-1.34

só

-1.31

 hvis

-1.27

Componentes

-1.27

直到

-1.27

-1.24

POSITIVE LOGITS

 autre

1.59

 autres

1.59

 vasco

1.48

 CONSE

1.44

OTHER

1.40

IVENESS

1.39

"),

1.38

 fontaine

1.34

зак

1.34

",

1.33

Activations Density 0.024%