INDEX

Explanations

step forward

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

ラス

-0.79

iranje

-0.75

 whirl

-0.71

 Capability

-0.71

base

-0.71

jsonwebtoken

-0.70

گیری

-0.70

med

-0.69

preview

-0.67

紳

-0.67

POSITIVE LOGITS

 forward

2.33

 come

1.75

Forward

1.61

 Forward

1.59

forward

1.59

 step

1.54

 stepped

1.53

 came

1.48

Come

1.44

Step

1.42

Activations Density 0.016%

No Known Activations