INDEX

Explanations

name of code elements

The neuron strongly activates on past‐tense verbs ending in “-ed.”

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

filename

0.32

id

0.32

其它

0.32

Dollars

0.31

ItemID

0.31

)});

0.30

 avocados

0.30

tolist

0.30

팟

0.30

 Counties

0.30

POSITIVE LOGITS

 ジャン

0.36

 kreatif

0.34

rainian

0.33

উন

0.32

 誕生日

0.32

 Ανακτήθηκε

0.30

 મહે

0.30

 Про

0.30

atihan

0.30

ム

0.30

Activations Density 0.000%