INDEX

Explanations

situation where someone is upset

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

otal

-0.09

Mb

-0.09

 satisf

-0.09

 Ø§ÛĮÙĨÚ©Ùĩ

-0.09

ALAR

-0.08

tle

-0.08

 desp

-0.08

alar

-0.08

 Rewards

-0.08

ede

-0.08

POSITIVE LOGITS

 others

0.11

æľīäºº

0.11

erse

0.10

ãĤĩ

0.10

uzey

0.09

jde

0.09

others

0.09

 -*-č\n

0.09

 hurt

0.09

 kern

0.09

Activations Density 0.193%