INDEX

Explanations

programming keywords and types

The neuron fires on all‐caps identifiers or macros containing underscores (e.g. FIELD_TYPES, OBJECT_TYPE, RULE_SUFFIX, DEFAULT_MIN_TIMEOUT).

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

They

-1.85

All

-1.59

Both

-1.55

在我

-1.51

 немного

-1.51

For

-1.48

Even

-1.46

好久

-1.45

-1.42

Firma

-1.41

POSITIVE LOGITS

 inicios

1.74

 impedir

1.71

)}+\

1.70

we

1.67

很赞

1.66

犼

1.62

你先

1.62

夲

1.55

该如何

1.53

側は

1.52

Activations Density 0.026%