INDEX

Explanations

end of sentence punctuation

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

/******/

-0.11

Liv

-0.09

abis

-0.09

 misunder

-0.09

loquent

-0.09

liv

-0.09

ÃĹ\n\n

-0.08

maal

-0.08

entionPolicy

-0.08

uet

-0.08

POSITIVE LOGITS

bip

0.10

 homic

0.09

aris

0.08

ï½ī

0.08

vp

0.08

 Confeder

0.08

 stacks

0.08

 depr

0.08

 Loch

0.08

elin

0.08

Activations Density 0.042%