INDEX

Explanations

have the ability or capability

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

arena

-0.10

zes

-0.09

âĮĴ

-0.09

ore

-0.09

 Jennings

-0.08

Pie

-0.08

angu

-0.08

aku

-0.08

iac

-0.08

 cerebral

-0.08

POSITIVE LOGITS

 capability

0.23

 ability

0.20

 capabilities

0.19

èĥ½åĬĽ

0.19

capability

0.17

 abilities

0.15

 possibility

0.14

 MÃ¶glichkeit

0.14

 ëĬ¥

0.14

 Capability

0.13

Activations Density 0.080%