INDEX

Explanations

perfect or 100%

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

printStats

-0.11

NewProp

-0.10

Intialized

-0.09

 imperson

-0.09

ernetes

-0.09

TriState

-0.09

finity

-0.09

UIL

-0.09

egie

-0.09

 reli

-0.08

POSITIVE LOGITS

 perfect

0.44

 perfectly

0.34

perfect

0.34

 Perfect

0.33

Perfect

0.30

 completely

0.30

 PERF

0.30

0.28

å®Įåħ¨

0.27

å®Į

0.27

Activations Density 0.236%