[Slider] Social class stratification - Social capital

세부 정보

모델 설명

Slider LoRa for Controlling Social Status based on social capital

The sociologist Pierre Bourdieu defined social capital as: actual or virtual resources acquired by individuals or groups through the possession of “more or less institutionalized relationships of mutual acquaintance and recognition” (Bourdieu & Wacquant, 1992: 119). In simpler terms social capital refers to the benefits individuals gain from their relationships and networks within society.

This slider LoRa tries to capture this dimension using images. This model is designed to explore how machine learning models can represent different social status categories.

DISCLAIMER: Please note that social status and social capital are a complex phenomena that cannot be reduced to one's clothing or general outlook. Additionally, I am aware that the model may have racial bias, which was not intentional.

What does it do?

Lets the user change the status of the individual.

How to use it?

Add to prompt just like any other LoRa, but adjust the weight from negative -5 to +5 to achieve roughly the following results:

Set LoRa weight to produce:

  • Heigher weight: more social capital

  • Lower weight: less social capital

Training conditioning:

Positive:

"extremely well-connected person, high social capital person, extremely
trusted person, networked person, extremely socially embedded person"

Negative:

"extremely isolated person, low social capital person, extremely distrusted person, unnetworked person, extremely socially disengaged person"

Tested on v1.5, RealisticVision, EpicRealism

Trained using: https://github.com/ostris/ai-toolkit

Acknowledgments:

I am grateful to Ostris for their outstanding AI Toolkit, which served as the foundation and inspiration for this Slider LoRa. Kindly acknowledge and credit Ostris for their invaluable contributions to the AI community.

If you like my work and would like to support me, you can buy me a coffee!

Any feedback, suggestion or criticism is appreciated.

이 모델로 만든 이미지

이미지를 찾을 수 없습니다.