INDEX

Explanations

section headings and parts

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

Â¯ÃĤ

-0.10

/*@

-0.09

 senses

-0.09

andest

-0.09

AZY

-0.08

neg

-0.08

 addCriterion

-0.08

vro

-0.08

Â°}

-0.08

Americ

-0.08

POSITIVE LOGITS

 PART

0.12

by

0.12

ï¿

0.12

 Part

0.11

.pdf

0.10

 part

0.09

part

0.09

 subtitle

0.09

Part

0.08

Activations Density 0.176%