INDEX

Explanations

many, much, several, lots

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

gid

-0.10

eper

-0.09

enburg

-0.09

 vastly

-0.08

:::::::::::::

-0.08

 different

-0.08

#ab

-0.08

ece

-0.08

 jinÃ½

-0.08

POSITIVE LOGITS

 many

0.17

 lots

0.14

many

0.13

 plenty

0.13

è®¸å¤ļ

0.13

å¾Īå¤ļ

0.12

 Ð¼Ð½Ð¾Ð³Ð¸Ñħ

0.12

 muchos

0.11

 banyak

0.11

 viele

0.11

Activations Density 0.037%