INDEX

Explanations

many problems, her catalog, a fun

many

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

Negative Logits

Nz

-0.10

ãĥĩãĥ«

-0.09

 Carlton

-0.09

itten

-0.08

LLU

-0.08

oo

-0.08

 Ment

-0.08

ÌĪ

-0.08

 Flem

-0.08

_fixture

-0.08

POSITIVE LOGITS

 Claus

0.12

 sole

0.11

¶Į

0.10

ohl

0.09

oles

0.09

sole

0.09

 Wright

0.09

æ°ı

0.09

eto

0.08

leigh

0.08

Activations Density 0.099%