INDEX

Explanations

Hyundai, Gemini, Resolver

The neuron specifically detects occurrences of the brand name “Hyundai.”

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

橛

-2.81

-2.67

价格

-2.58

ada

-2.50

 être

-2.47

他的

-2.41

 рассказывает

-2.39



-2.36

咘

-2.33

튿

-2.33

POSITIVE LOGITS

’

3.28

 That

3.02

 Following

2.70

 whitish

2.55

 Only

2.47

 Then

2.42

 Meanwhile

2.39

੍ਹ

2.38

 Also

2.36

Activations Density 0.006%