INDEX

Explanations

exceptions and errors

This neuron activates on tokens that are part of exception or error class names (e.g. “Illegal”, “NoSuch”, “Unknown”, “Internal”, “Invalid”).

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

 唯美

-1.77

 compagnon

-1.45

鮰

-1.43

 Spanyol

-1.42

 Publicado

-1.41

鼋

-1.39



-1.38

 culte

-1.35

 spesial

-1.34

 signific

-1.30

POSITIVE LOGITS

1.55

1.54

1.52

or

1.52

It

1.49

1.41

when

1.39

just

1.37

 ['./

1.36

Some

1.35

Activations Density 0.003%