INDEX

Explanations

AS IS

The neuron fires on the punctuation and digits that make up the “Licensed under the Apache License, Version 2.0” header (the comment‐asterisks, comma, and the “2” in “Version 2.0”).

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

woof

-0.77

NARR

-0.77

вр

-0.75

 الحدود

-0.75

 STATEMENT

-0.73

nment

-0.71

 cercle

-0.70

NSS

-0.69

urers

-0.69

 kamp

-0.69

POSITIVE LOGITS

 invokevirtual

0.71

 présente

0.71

 sees

0.68

 lộ

0.65

ote

0.65

Results

0.64

 think

0.61

 dẫn

0.60

 without

0.60

 protects

0.59

Activations Density 0.073%