INDEX

Explanations

mentions of flaws or mistakes

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 shortcomings

-0.92

 CURIAM

-0.76

__":

-0.71

 deficiencies

-0.69

 linkovi

-0.68

Filmographie

-0.66

 sumpay

-0.64

 bucket

-0.61

 otomatig

-0.61

 OCCUP

-0.60

POSITIVE LOGITS

 flaw

2.00

 flaws

1.44

 flawed

1.22

 flawless

1.15

airo

0.79

 flawlessly

0.79

 Blas

0.72

Blas

0.65

dele

0.64

fla

0.61

Activations Density 0.001%