INDEX

Explanations

achievement and praise

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

 perfection

-0.11

 singleton

-0.09

 Rodrig

-0.09

 perfected

-0.09

 Emanuel

-0.09

{{{

-0.09

å¯Ħ

-0.09

ìłł

-0.09

Yue

-0.08

POSITIVE LOGITS

 effort

0.16

job

0.13

åĭĩ

0.13

 initiative

0.12

 finally

0.12

 restraint

0.12

 accomplishment

0.12

 steps

0.11

 efforts

0.11

 feat

0.11

Activations Density 0.046%