INDEX

Explanations

socially responsible and ethical manner

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

Negative Logits

 disrespect

-0.09

ylon

-0.09

ician

-0.09

 Mouse

-0.09

Orr

-0.09

verity

-0.09

iena

-0.09

Ã¼ss

-0.09

éĹ²

-0.08

POSITIVE LOGITS

 manner

0.25

 responsible

0.19

way

0.18

 ways

0.17

 transparent

0.17

eth

0.15

 Transparent

0.15

transparent

0.14

 Ways

0.14

 ethical

0.13

Activations Density 0.019%

No Known Activations