INDEX

Explanations

and/or elaborations

The neuron fires strongly on common connectives—especially coordinating conjunctions (e.g. "and," "for," "or")—highlighting discourse‐linking words.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 three

0.68

 four

0.59

six

0.57

 five

0.57

0.56

 ৩০

0.56

0.55

 трех

0.55

two

0.54

并通过

0.54

POSITIVE LOGITS

 ljudi

0.83

 ludzi

0.82

 etcétera

0.79

 Ordnung

0.75

 şeyi

0.75

 Dlatego

0.75

 deswegen

0.74

 irgend

0.73

 mensen

0.73

Etc

0.73

Activations Density 0.621%