INDEX

Explanations

can't help

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

 èĩªåĬ¨çĶŁæĪĲ

-0.10

ellig

-0.10

URLException

-0.10

ãĥ³ãĥķ

-0.09

 afore

-0.09

styleType

-0.09

Ð°ÑģÑģÐ¸Ð²

-0.09

DataExchange

-0.09

scriptId

-0.08

POSITIVE LOGITS

 recip

0.12

ogle

0.10

 notice

0.10

mast

0.10

 resistance

0.09

apon

0.09

 suspect

0.09

ingle

0.09

-m

0.09

 Denn

0.09

Activations Density 0.156%