INDEX

Explanations

potential applications

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

 Ramp

-0.09

iske

-0.09

 deep

-0.09

esk

-0.09

uzz

-0.09

Ð½ÐµÐ´

-0.09

 privile

-0.09

aff

-0.09

 knot

-0.09

 conf

-0.08

POSITIVE LOGITS

 promise

0.13

ollen

0.11

 possibilities

0.10

anges

0.10

 applications

0.10

uses

0.10

CHED

0.10

/problems

0.10

applications

0.10

unt

0.10

Activations Density 0.040%