INDEX

Explanations

known for, focuses on, role in

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

¶Į

-0.15

EMPLARY

-0.12

.Dictionary

-0.11

Âģ@

-0.11

******č\n

-0.10

ÂĢÂĢ

-0.10

Â¿ÃĤ

-0.10

 -*-č\n

-0.10

.SuspendLayout

-0.09

¦æĥħ

-0.09

POSITIVE LOGITS

 similar

0.09

osc

0.08

Âł

0.08

 Similar

0.08

aji

0.08

 Verm

0.08

its

0.07

hs

0.07

its

0.07

 thanks

0.07

Activations Density 0.056%

No Known Activations