INDEX

Explanations

pocket and wallet

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

 suitcase

-0.12

æĬ±

-0.10

 Garland

-0.10

iod

-0.10

_IOC

-0.09

 collaps

-0.09

 hood

-0.09

head

-0.09

 boxed

-0.09

 Lionel

-0.09

POSITIVE LOGITS

 pockets

0.29

 wallet

0.27

 pocket

0.26

 wallets

0.24

Pocket

0.22

 Wallet

0.22

wallet

0.21

-pocket

0.21

 Pocket

0.20

Wallet

0.17

Activations Density 0.029%