INDEX

Explanations

research, corporation, and points

The neuron detects mentions of quantitative preference–elicitation methods and evaluation tools (e.g. surveys, polls, discrete choice or conjoint analyses).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

had

-1.38

 just

-1.34

 unusually

-1.23

so

-1.18

 that

-1.17

 because

-1.17

can

-1.16

 this

-1.15

 about

-1.15

for

-1.14

POSITIVE LOGITS

🐐

1.35

">“

1.32

ће

1.27

ಠ

1.24

ceptibility

1.18

からない

1.17

charpe

1.16

lers

1.16

 acessar

1.13

 bedien

1.13

Activations Density 0.095%