INDEX

Explanations

power and energy units

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

æĸ·

-0.10

PTH

-0.09

 wash

-0.09

anco

-0.09

GIS

-0.09

ACE

-0.08

 Israelis

-0.08

ãģ

-0.08

 agent

-0.08

POSITIVE LOGITS

0.18

åįĥ

0.14

0.12

 Thousand

0.12

 thousand

0.12

 åįĥ

0.11

 prefixes

0.10

åħ

0.10

venta

0.10

 Cent

0.10

Activations Density 0.079%