INDEX

Explanations

introduces explanations or consequences

The neuron is detecting the word “means” (especially in explanatory or parenthetical phrases like “which means…”).

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

⼥

-1.73

 visse

-1.69

 vasca

-1.64

 dahil

-1.63

韪

-1.62

奩

-1.58

等級

-1.55

इटम

-1.55

 ERSITY

-1.50

 rahasia

-1.49

POSITIVE LOGITS

pegno

1.84

</h2>

1.83

return

1.77

 και

1.73

1.71

و

1.70

และ

1.70

<em>

1.64

}$

1.64

</sup>

1.63

Activations Density 0.006%