INDEX

Explanations

completion or confirmation

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

erg

-0.10

Greetings

-0.10

oom

-0.09

 Props

-0.09

 props

-0.09

ousse

-0.09

omore

-0.08

usty

-0.08

izer

-0.08

 Koch

-0.08

POSITIVE LOGITS

 Done

0.15

Done

0.15

 shall

0.15

 DONE

0.14

 done

0.14

DONE

0.13

done

0.11

-done

0.11

ê²łìĬµëĭĪëĭ¤

0.11

Ok

0.11

Activations Density 0.153%