INDEX

Explanations

lack and limitations

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

 inconsistency

-0.13

sad

-0.10

pur

-0.10

 inconsist

-0.10

åħį

-0.10

 inconsistencies

-0.09

 incompetence

-0.09

arov

-0.09

 situation

-0.09

 prompt

-0.09

POSITIVE LOGITS

 lack

0.28

 limited

0.24

 difficulty

0.21

limited

0.20

 poor

0.20

 absence

0.19

 inability

0.19

 falta

0.19

 Lack

0.18

æľīéĻĲ

0.18

Activations Density 0.110%