INDEX

Explanations

states of being or requirements

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

 Wilderness

-0.11

 ActionTypes

-0.10

 ineligible

-0.09

 worthless

-0.09

 indispens

-0.09

.:.:.:.:

-0.09

;;;;;;

-0.08

AndUpdate

-0.08

 Luck

-0.08

POSITIVE LOGITS

 appropriate

0.29

 necessary

0.26

 needed

0.25

appropriate

0.23

å¿ħè¦ģ

0.22

 helpful

0.20

 indicated

0.20

 appropri

0.19

 required

0.18

ropriate

0.17

Activations Density 0.055%