INDEX

Explanations

chat, login, searched, carcinogenic

references to "chat" in software code or online discussions about chat applications.

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

ﮕ

-3.48

𖤍

-3.42

剤

-3.42

ſſen

-3.34

creenshot

-3.31

Ꝑ

-3.31

臯

-3.30

اتها

-3.28

er

-3.28

姆

-3.25

POSITIVE LOGITS

4.44

4.13

Initially

4.03

</strong>

3.86

It

3.64

There

3.61

Interestingly

3.48

</h3>

3.48

When

3.42

3.41

Activations Density 0.005%