INDEX

Explanations

Ask HN

The neuron is keyed to pick out document‐section headings or post/mode labels (e.g. “Ask HN:…,” “Opinion,” “#,” etc.) rather than normal sentence text.

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

 แต่

-1.59

but

-1.47

Ventajas

-1.40

توضیحات

-1.40

BUT

-1.38

 mutta

-1.34

 також

-1.30

but

-1.29

 αλλά

-1.29

 namun

-1.27

POSITIVE LOGITS

 have

1.72

 тату

1.50

érer

1.39

fusions

1.36

мам

1.35

pography

1.33

 tornillos

1.30

윔

1.28

}>;

1.28

pecies

1.26

Activations Density 0.218%