INDEX

Explanations

short and precise responses

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

PFN

-0.10

bee

-0.10

-flag

-0.08

ï½°

-0.08

ecided

-0.08

vla

-0.08

 Loft

-0.08

umu

-0.08

íĹĮ

-0.08

 gradu

-0.08

POSITIVE LOGITS

 short

0.13

 simply

0.12

 Short

0.11

Short

0.10

çŁŃ

0.09

short

0.09

Nos

0.09

_dual

0.09

:\n

0.09

elan

0.09

Activations Density 0.114%