Ads Generation

The thesis is devoted to the development of a web service that allows marketers to create creatives for advertisements based on a keyword. The main task of the work is to find the optimal solution in terms of complexity of implementation, resource requirements and quality generate an approach and implement it. The paper provides an overview of existing methods for generating texts based on neural networks of different architectures, including modern methods using the example of large language models. The result of the work from the user's point of view is a developed web application, with which you can obtain generated text using given keywords and evaluate the quality of the generation. From the point of view of solving machine learning problems, the work will involve additional training of one of the large language models on user data. In the future, it is planned to refine the functionality of the service, additional iterations of training for cases when the quality of generation is not enough for keywords of a certain topic. Keywords: Large language models, text generation, neural networks, artificial intelligence