INDEX

Explanations

The neuron strongly activates on marketing‐style, benefit-focused imperatives and feature descriptions (e.g. “will make … envy your yard,” “allows for efficiency and a better quality,” etc.).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

涉及到

0.43

面临

0.43

┉

0.41

涉及

0.39

 دچار

0.39

 існу

0.38

起源

0.38

どうしても

0.38

诅

0.38

хову

0.38

POSITIVE LOGITS

 excellent

1.57

 good

1.45

excellent

1.37

 excelente

1.36

 satisfactory

1.32

 tốt

1.31

 excellente

1.27

 bagus

1.25

 better

1.21

 अच्छा

1.21

Activations Density 7.617%