INDEX

Explanations

medication and treatment

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

zdy

-0.10

Kup

-0.10

ceae

-0.10

 kola

-0.10

Lug

-0.09

 neod

-0.09

Wr

-0.09

 Pert

-0.09

åį

-0.09

 perc

-0.09

POSITIVE LOGITS

ovel

0.13

 Diary

0.11

 diary

0.11

 novel

0.11

Saf

0.10

 once

0.10

 Novel

0.10

Ki

0.10

 twice

0.09

pan

0.09

Activations Density 0.128%