INDEX

Explanations

expensive initial cost

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

 Moms

-0.09

.ERR

-0.08

GBT

-0.08

BTN

-0.08

leftright

-0.08

ìĬ¹

-0.08

Ã©tÃ©

-0.08

_reporting

-0.08

 conservatism

-0.08

anje

-0.08

POSITIVE LOGITS

 initial

0.20

 cost

0.16

 Initial

0.15

 initially

0.15

initial

0.15

 costs

0.14

 inicial

0.14

Initial

0.14

åĪĿ

0.13

.initial

0.12

Activations Density 0.020%