INDEX

Explanations

after, of, if, on, regarding, by

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

>{!!

-0.09

/Dk

-0.09

 baise

-0.08

 Dorm

-0.08

coma

-0.08

null

-0.08

Ø§Øº

-0.07

ipt

-0.07

 crim

-0.07

owied

-0.07

POSITIVE LOGITS

ï½Ĵ

0.08

verts

0.08

ynes

0.08

rupt

0.08

pkt

0.08

.mdl

0.08

 Mish

0.08

Periph

0.07

inn

0.07

Activations Density 0.079%