INDEX

Explanations

fallacies in argument

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

 theories

-0.10

 contrad

-0.10

concept

-0.09

rio

-0.09

ug

-0.09

é»İ

-0.09

elib

-0.09

 Holt

-0.09

çĮľ

-0.08

âĦĸâĦĸ

-0.08

POSITIVE LOGITS

 argument

0.25

 Argument

0.25

 premises

0.21

Argument

0.21

argument

0.20

arg

0.19

(argument

0.19

 arguments

0.18

_argument

0.17

 logic

0.17

Activations Density 0.077%