INDEX

Explanations

imagination, dreams, personal experiences

New Auto-Interp

Top Features by Cosine Similarity

Configuration

Prompts (Dashboard)

10,000 prompts, 128 tokens each

Dataset (Dashboard)

lmsys/lmsys-chat-1m

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ropri

-0.10

 Impress

-0.10

 contemplated

-0.09

 Puzzle

-0.09

éĩ

-0.09

 enlightenment

-0.09

 puzzle

-0.08

 anticipated

-0.08

 ware

-0.08

 mythical

-0.08

POSITIVE LOGITS

 imagination

0.54

 imag

0.45

imagin

0.34

agination

0.32

imag

0.29

.imag

0.29

 imagine

0.29

_imag

0.29

 Imag

0.28

Imagine

0.26

Activations Density 0.262%