INDEX

Explanations

array and map operations

The neuron fires on tokens that are names of programming functions or API methods (e.g. Map, range, rand, fetch_assoc).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

膑

-1.64

we

-1.55

 suelos

-1.51

 indruk

-1.45

 beelden

-1.43

壁纸

-1.38

 рекоменду

-1.36

 сожа

-1.35

 cielos

-1.34

 contribuer

-1.34

POSITIVE LOGITS

%).

1.66

_));

1.63

_))

1.54

MeasureSpec

1.48

。（

1.46

 From

1.45

 Even

1.44

_),

1.43

Activations Density 0.008%