INDEX

Explanations

numerical identifiers and codes

The neuron is triggering on numeric tokens (i.e. any standalone numbers or number‐like strings).

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

년

-1.52

 will

-1.46

 Fassung

-1.45

--;

-1.43

 would

-1.41

芣

-1.39

するので

-1.34

 febrero

-1.33

 zdra

-1.32

 decidi

-1.31

POSITIVE LOGITS

 […]

1.39

 eisen

1.37

スの

1.34

–

1.27

StoryboardSegue

1.24

</tfoot>

1.22

bedded

1.21

($__

1.21

 awakeFromNib

1.20

であり

1.20

Activations Density 0.223%