Open Source Model
The model of artificial intelligence that is clearly shared with weights and often code to everyone.
The open source model is the artificial intelligence model, which is often shared with an open license to everyone of the architecture and sometimes the educational code. This means developers can download and operate the model on their servers or computers, examine the internal structure and customize them by tweaking their data. Unlike closed models, data privacy in open-source models is the user’s control because queries are not required to be sent to the company’s server; also researchers can check the behavior, bias and limits of the model transparently.
The Llama family of Meta is the examples of the open source ecosystem, such as Mistral, Falcon and Stable Diffusion; these models consist of a huge community on platforms such as Hugging Face, thousands of species (fine-tuned) versions are manufactured. The open source approach is critical to the democraticization of artificial intelligence, because small companies, academicians and independent developers can access powerful models without folding the cost of millions of dollars API. On the other hand, since the weights of a strong model can increase the risk of abuse freely distributed, there is continuous discussion in the sector on the degree of "externity".
The Llama family of Meta is the examples of the open source ecosystem, such as Mistral, Falcon and Stable Diffusion; these models consist of a huge community on platforms such as Hugging Face, thousands of species (fine-tuned) versions are manufactured. The open source approach is critical to the democraticization of artificial intelligence, because small companies, academicians and independent developers can access powerful models without folding the cost of millions of dollars API. On the other hand, since the weights of a strong model can increase the risk of abuse freely distributed, there is continuous discussion in the sector on the degree of "externity".