INDEX

Explanations

"construct step set", "background"

The neuron responds almost exclusively to the exact token “set,” i.e. it detects the word “set.”

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

 They

-2.19

 That

-2.16

 bahwa

-1.91

 sabias

-1.76

ப்பட்ட

-1.75

衎

-1.75

骘

-1.70

 pilares

-1.70

 longtime

-1.69

 stránky

-1.67

POSITIVE LOGITS

2.39

↵↵

1.97

1.88

 dispone

1.77

1.76

ad

1.71

1.70

其他人

1.70

 sost

1.66

 interno

1.63

Activations Density 0.000%