INDEX

Explanations

to ensure / have attached

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

vet

-0.11

omm

-0.09

 Wait

-0.09

 sque

-0.09

 wait

-0.09

 retr

-0.09

 Dess

-0.09

 reuse

-0.09

BIN

-0.09

ips

-0.09

POSITIVE LOGITS

 Remote

0.13

 remote

0.13

Remote

0.11

remote

0.10

_remote

0.10

 Neon

0.10

 catching

0.10

 remot

0.10

catch

0.09

 zÃ¡stup

0.09

Activations Density 0.034%