INDEX

Explanations

summaries and findings from data

The neuron detects terms marking summary or key‐point sections—words like “takeaways,” “findings,” “observations,” “notes,” and “report.”

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

</sub>

-2.23

%;

-1.76

-1.63

<i>

-1.55

It

-1.51

to

-1.50

If

-1.50

};

-1.50

};

-1.50

 dengan

-1.49

POSITIVE LOGITS

 about

1.91

 from

1.74

 waarvan

1.46

文學

1.45

っています

1.44

 propriétaires

1.44

 fieltro

1.43

妧

1.43

体会

1.41

烏

1.41

Activations Density 0.069%