INDEX

Explanations

taking into account

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

seealso

-0.09

 Leonard

-0.09

 lineup

-0.09

 Jarvis

-0.09

 Laud

-0.08

sal

-0.08

ê³¡

-0.08

adan

-0.08

 Carlton

-0.08

vetica

-0.08

POSITIVE LOGITS

 account

0.39

 Account

0.28

account

0.27

 cuenta

0.26

 compte

0.25

 consideration

0.25

Account

0.23

-account

0.22

_account

0.22

 mind

0.21

Activations Density 0.015%