INDEX

Explanations

or both

The neuron specifically detects the word “both,” particularly when used in coordinate constructions (e.g., “or both”).

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

まさかの

-1.53

そこから

-1.48

噻

-1.39

责编

-1.34

</u>

-1.34

 Полу

-1.33

 verdaderos

-1.30

VELAND

-1.27

-1.26

满脸

-1.26

POSITIVE LOGITS

℠

1.58

на

1.46

()");

1.45

 conseguiu

1.41

🌸

1.41

</h2>

1.37

я

1.37

↑

1.31

 Yesus

1.30

 fiore

1.30

Activations Density 0.011%