INDEX

Explanations

rope puzzle or loop

The neuron is triggered by mentions of “rope” (and related rope‐based terms).

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

longValue

-0.92

 pins

-0.79

 Vancouver

-0.78

 аниме

-0.78

 bloc

-0.77

 cate

-0.76

 größer

-0.75

Craig

-0.73

Ирина

-0.73

兴

-0.72

POSITIVE LOGITS

 dispenser

0.97

rope

0.97

 Dispenser

0.94

 rope

0.91

 unravel

0.91

 Rope

0.90

 tied

0.88

wain

0.87

 tying

0.86

Rope

0.86

Activations Density 0.016%