INDEX

Explanations

is good, lo que, I could, I need

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

ivol

-0.10

 Griff

-0.09

orda

-0.09

Voj

-0.09

urus

-0.08

é«

-0.08

 baÅŁk

-0.08

few

-0.08

ãĥ«ãĥĪ

-0.08

 transf

-0.08

POSITIVE LOGITS

 ever

0.17

 done

0.14

 Ever

0.12

 pent

0.12

done

0.12

Ever

0.12

 DONE

0.11

 EVER

0.11

ever

0.11

ohon

0.11

Activations Density 0.042%