INDEX

Explanations

in fig references

This neuron detects mentions of figures (e.g. the phrase “shown in FIG.”).

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

-1.74

ver

-1.66

-1.63

↵↵

-1.61

-1.59

);

-1.52

-1.49

Diese

-1.47

POSITIVE LOGITS

 fábricas

1.71

 motocicletas

1.70

 localidades

1.65

and

1.64

 perdidos

1.63

pez

1.59

 humilde

1.59

 despite

1.57

 semanas

1.54

 bicicletas

1.51

Activations Density 0.013%