INDEX

Explanations

chemical lists, and

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

æĿĲæĸĻ

-0.11

stoff

-0.10

 metals

-0.09

 Filip

-0.09

yx

-0.09

.FileReader

-0.09

 Metals

-0.08

ilk

-0.08

ISE

-0.08

fil

-0.08

POSITIVE LOGITS

 neutral

0.12

 Formal

0.10

 prop

0.10

 Grade

0.10

ca

0.10

 Flake

0.10

 edible

0.10

 Neutral

0.10

flake

0.10

starter

0.10

Activations Density 0.058%