INDEX

Explanations

is true

This neuron fires most strongly on the opening words of new sentences or clauses—i.e. structural discourse markers that signal sentence or idea boundaries.

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

essoas

0.80

 entweder

0.77

 иностранных

0.76

 étrangers

0.72

aptops

0.72

 سایر

0.71

 parfois

0.71

 oftentimes

0.70

 ETFs

0.70

 PDEs

0.70

POSITIVE LOGITS

这个

0.91

它

0.80

この

0.79

 lightness

0.78

這個

0.77

 this

0.76

 questo

0.76

 этому

0.76

 هذا

0.73

 этого

0.71

Activations Density 0.353%