INDEX

Explanations

after being

The neuron fires on code‐style identifiers and keywords (e.g. method, field, frame, directional or variant names) in programming or data‐structure contexts.

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

 sanitaires

-1.52

to

-1.52

掾

-1.47

וּ

-1.44

 бампер

-1.38

᾽

-1.34

==""){

-1.30

 HARRISON

-1.30

 púb

-1.30

POSITIVE LOGITS

 with

1.80

but

1.79

 државе

1.67

νονται

1.66

 сахара

1.55

 möglicherweise

1.48

 although

1.45

虽说

1.44

 میوه

1.44

 masih

1.41

Activations Density 0.001%