INDEX

Explanations

code and programming constructs

The neuron is keyed to tokens from C/C++ include‐directive file paths—especially header names, directory separators, and “.h” extensions.

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

on

-1.34

 first

-1.32

 Then

-1.31

 Both

-1.27

 Unlike

-1.26

In

-1.25

 both

-1.24

-1.21

-1.19

 From

-1.15

POSITIVE LOGITS

܆

1.47

 empêcher

1.41

 conciertos

1.31

 résister

1.30

artiste

1.28

 éviter

1.27

 conejo

1.27

jusqu

1.27

 desempen

1.26

apparently

1.26

Activations Density 0.049%