INDEX

Explanations

job information

The neuron fires on section headings or labels that include numeric metadata (for example counts, measurements, or numeric summaries in headings like “Search results: 146,” “Related posts: 63.5,” “Top features: 70,” etc.).

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

他的

-1.55



-1.40

↵↵

-1.30

Oft

-1.27

 zprá

-1.27

此

-1.26

併

-1.23

lition

-1.23

這

-1.23

 Takeaways

-1.21

POSITIVE LOGITS

1.63

1.59

 efektif

1.55

鐒

1.52

1.47

Répondre

1.46

вропей

1.38

縯

1.38

 toman

1.37

覓

1.36

Activations Density 0.118%