INDEX

Explanations

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

ãĤ©

-0.09

 Opport

-0.08

tre

-0.08

 Pied

-0.08

_chk

-0.08

oca

-0.08

 proper

-0.08

 selectively

-0.08

 opport

-0.08

POSITIVE LOGITS

 choice

0.94

 choices

0.79

choice

0.72

 Choice

0.69

choices

0.62

 choix

0.62

Choice

0.61

-choice

0.60

 Choices

0.57

.choice

0.55

Activations Density 0.171%

popular choices