INDEX

Explanations

I'm sorry, unfortunately

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

Yup

-0.10

 prob

-0.10

 Archer

-0.09

 already

-0.09

 after

-0.09

cest

-0.09

Yue

-0.09

 when

-0.09

ique

-0.09

POSITIVE LOGITS

 unable

0.16

Unfortunately

0.16

 sorry

0.16

çĶ±äºİ

0.15

Sorry

0.15

SOR

0.15

 Sorry

0.15

æĹłæ³ķ

0.15

sorry

0.14

 Unfortunately

0.13

Activations Density 0.230%