INDEX

Explanations

performance evaluation and analysis

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

\"

-1.98

 prácticamente

-1.97

-1.74

 encabez

-1.61

 inept

-1.60

 lanz

-1.59

ዡ

-1.59

},

-1.55

『

-1.55

 myriad

-1.52

POSITIVE LOGITS

Performance

1.93

performance

1.80

–

1.78

 performance

1.77

 Performance

1.76

 voegen

1.55

 them

1.53

꩜

1.48

磚

1.48

Lugar

1.44

Activations Density 0.040%