INDEX

Explanations

structure followed by common words

New Auto-Interp

Configuration

Prompts (Dashboard)

238,145 prompts, 512 tokens each

Dataset (Dashboard)

lmsys + oasst1

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 unsurprisingly

1.26

 crucially

1.18

 despite

1.13

 Notably

1.13

Notably

1.12

 nonostante

1.11

一方で

1.10

 Несмотря

1.05

 Despite

1.04

 insanely

1.03

POSITIVE LOGITS

1.47

1.31

then

1.24

please

1.20

you

1.17

Thank

1.17

Thats

1.15

Then

1.13

they

1.10

They

1.09

Activations Density 0.002%