INDEX

Explanations

EVERY DAY

The neuron fires on tokens used for emphasis or formatting—things like uppercase words, punctuation marks (e.g. “)”), markup tags (e.g. <strong>, <b>), and other emphasis markers.

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

厦门

-1.54

by

-1.45

 with

-1.38

him

-1.32

 (;;)

-1.30

🧍

-1.30

 maleta

-1.27

on

-1.27

武汉

-1.26

ations

-1.24

POSITIVE LOGITS

cié

1.37

 Whilst

1.33

 encontramos

1.33

1.30

phed

1.30

1.29

 Pflichten

1.27

完成です

1.27

Activations Density 0.004%