INDEX

Explanations

code and programming constructs

The neuron fires strongly on the “(” that begins a function’s parameter list (and on the nearby tokens in that signature), i.e. it detects the start of a function definition’s parameters.

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

 suscit

-1.40

 confuso

-1.39



-1.38

伥

-1.33

荭

-1.27

 spolup

-1.26

 можна

-1.23

 murió

-1.21

 allerlei

-1.20

/*++

-1.20

POSITIVE LOGITS

all

1.48

 from

1.46

not

1.42

by

1.41

 with

1.38

in

1.28

for

1.26

 very

1.17

 data

1.14

 work

1.08

Activations Density 0.013%