INDEX

Explanations

sentiment and contrast

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 Especially

-0.13

 especially

-0.13

 wherever

-0.11

especially

-0.11

esp

-0.10

 even

-0.10

 Welch

-0.10

 anyway

-0.10

 anyways

-0.10

 Until

-0.10

POSITIVE LOGITS

 sadly

0.19

 unfortunately

0.17

 sorry

0.15

 alas

0.14

 thankfully

0.14

 fortunately

0.14

due

0.13

 although

0.13

 zwar

0.12

 luckily

0.12

Activations Density 0.257%

No Known Activations