INDEX

Explanations

if, when, append, column

The neuron is strongly activated by comma punctuation marks.

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

 Eventos

-1.41

蕺

-1.36

 anún

-1.33

 enig

-1.29

ELTS

-1.27

<bos>

-1.27

ᕮ

-1.27

そして

-1.26

 LIABLE

-1.25

 AUSSI

-1.24

POSITIVE LOGITS

If

1.44

or

1.38

1.27

になりますが

1.23

At

1.20

What

1.20

When

1.16

つも

1.16

what

1.15

瀟

1.13

Activations Density 0.118%