INDEX

Explanations

code definitions and functions

The neuron consistently picks out numeric literals (digit sequences) in code.

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

そして

-1.48

を

-1.45

৹

-1.43

 despite

-1.40

と思い

-1.33

 mostly

-1.32

for

-1.30

 NavController

-1.28

が

-1.28

 mainly

-1.27

POSITIVE LOGITS

ところに

1.70

場所で

1.53

ときは

1.48

ときの

1.46

and

1.45

ものが

1.43

に行って

1.41

際には

1.39

 autén

1.37

ことも

1.35

Activations Density 0.017%