INDEX

Explanations

initial letters / abbreviations

The neuron fires on isolated single-letter tokens (e.g. lone “r,” “p,” “P,” “S,” etc.).

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

 even

-1.61

فسير

-1.43

 which

-1.42

 thậm

-1.40

-1.38

-1.36

at

-1.34

the

-1.29

he

-1.27

 adding

-1.27

POSITIVE LOGITS

YNAMIC

1.60

较高

1.41

'></

1.41

 gouttes

1.39

乄

1.38

 quelle

1.38

阏

1.37

蚡

1.37

謦

1.37

 demonio

1.35

Activations Density 0.361%