INDEX

Explanations

problem statements

The neuron responds to document section headings introducing a issue or question—especially lines beginning with labels like “Problem:” or “Issue:”.

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

㍑

-1.72



-1.70

 onsdag

-1.63

鐃

-1.61

♂

-1.54

 撮り

-1.53

⑊

-1.53

趿

-1.52

 itinéraire

-1.52

verkauf

-1.52

POSITIVE LOGITS

if

2.30

 Also

1.97

for

1.88

 this

1.86

 which

1.81

 that

1.77

 Since

1.71

1.69

 something

1.68

 While

1.64

Activations Density 0.081%