INDEX

Explanations

transition and uncertainty

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

 homeschool

-0.10

 aday

-0.09

moth

-0.09

 cheats

-0.09

ocal

-0.09

 unab

-0.09

 Cheat

-0.09

actices

-0.09

 sched

-0.09

edo

-0.08

POSITIVE LOGITS

 career

0.21

 adult

0.21

 independence

0.17

 Career

0.16

adult

0.16

 Independence

0.16

 Adult

0.15

 decisions

0.15

 choices

0.14

career

0.14

Activations Density 0.129%