INDEX

Explanations

code and URLs

The neuron fires on programming‐related identifiers (API paths, function names, XML attribute names, import/package names, etc.).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 with

-1.58

为

-1.51

 would

-1.45

↵↵↵

-1.41

was

-1.38

 could

-1.37

 after

-1.34

in

-1.33

 única

-1.33

 here

-1.32

POSITIVE LOGITS

of

1.93

 Hoje

1.48

embal

1.37

囖

1.37

Yesterday

1.37

 広場

1.36

Добавлено

1.34

heureux

1.34

entreprise

1.31

ientras

1.30

Activations Density 0.079%