INDEX

Explanations

scientific and technical questions

This neuron fires on technical keywords in formal contexts—most notably the words “partial” (in math formulas) and “typename” (in C++ templates).

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

to

-1.88

須賀

-1.48

In

-1.46

 astounding

-1.39

粁

-1.34

 должны

-1.34

 деся

-1.34

 This

-1.31

The

-1.30

応え

-1.30

POSITIVE LOGITS

 that

2.48

of

1.55

“

1.38

⽼

1.38

 œufs

1.34

 času

1.34

 écrits

1.34

ية

1.33

chnen

1.32

Как

1.28

Activations Density 0.000%