MISiS: NeuroStanislavsky Play Generator

The main articles are:

• Artificial intelligence (AI, Artificial intelligence, AI)
• Neural networks (neural networks)

2022: Creating a Play Generator

The experimental model of the neural network - the generator of plays "NeuroStanislavsky" was presented in Moscow. The project opens the next era in theatrical art and in the field of neural networks: artificial intelligence has already created music and paintings, but for the first time in Russia he wrote a play. NUST MISIS (National Research Technological University) announced this on November 11, 2022. This tool was created by a team of students from the ITAM hackathonists community of MISIS University in cooperation with the Biennale of Theater Arts Festival using Sberbank technologies . Now everyone can end up in the role of a playwright - the neural network model is available at the link go.itatmisis.ru/theatre.

Students at MISIS University took a progressive approach in which artificial intelligence does not replace the author, but helps to create and improve the plot, developing the director's thoughts. The neural network studied at the works of Russian classics - Ostrovsky, Chekhov, Gorky and others. Young engineers wrote an algorithm that studied dozens of plays and allocated 14 thousand dialogues from them. The database created in this way formed the basis of NeuroStanislavsky.

The developers used a public tool RuGPT-3 -transophorme model to understand the Russian language, trained by SberDevices specialists. At all stages of development, the project team was advised by professional directors and playwrights.

We live in a time when it is important to look for new and even unexpected opportunities for the use of artificial intelligence. In the future, neural networks can make a huge contribution to the development of theatrical art, but for now we are taking a significant step towards this, said Ivan Pleshakov, student of NUST MISIS, Machine Learning project team developer.

University students completed the resulting neural network models, selected hyperparameters for them using the power of the ML Space full-cycle machine learning platform, and wrote a special filter. Thanks to the latter, the generated text is more "clean." Sometimes the neural network is mistaken - it gives out extra characters, spaces, etc., and it levels this problem.

The project passed the first experience of introduction into theatrical practice: with the help of artificial intelligence, director Yaroslav Shevaldov wrote a fragment of the play about the founders of the Art Theater, the great reformers of stage art K.S. Stanislavsky and V.I. Nemirovich-Danchenko. The passage refers to their historical meeting at the Slavyansky Bazaar restaurant in Moscow in 1897, which forever changed the trajectory of the world theater. The source was Stanislavsky's famous book "My Life in Art."

According to the received text, the director staged a short performance.

Links

Link to neural network