GigaChat Passes University Exam in Thermal Power Engineering
Sber’s GigaChat neural network passed a university-level exam in power and thermal engineering after undergoing additional specialized training.
Sber’s GigaChat neural network completed a knowledge assessment at the Moscow Power Engineering Institute. The model received a grade equivalent to “good” in undergraduate-level exams covering electric power engineering and thermal power engineering.
The company said the neural network underwent additional training before the exam using theoretical and engineering-calculation materials. Specialists from Sber, researchers from the Moscow Power Engineering Institute and experts from Rosseti participated in preparing the model.
The exam showed that the neural network can reliably handle energy-sector topics and could assist engineers, utility-company employees and energy-infrastructure designers in their work.
GigaChat has previously passed exams in more than 20 disciplines. In medicine, the model completed certification tests in general medicine, cardiology, neurology, pediatrics, gastroenterology, rheumatology, dentistry, pulmonology, nutrition science and clinical pharmacology. The neural network has also been tested in musicology, agriculture, financial literacy, banking and insurance.
