On August 24, in the live broadcast event of the 20th anniversary of Hao Future, CTO Tian Mi announced that MathGPT, a 100-billion-level large model in the field of mathematics that has not been self-developed, was officially launched and opened internal testing. From now on, users can apply for an account to try the free trial through the official website ().
In May this year, Good Future announced that it was developing a self-developed mathematical large model, named MathGPT. MathGPT is a large model in the vertical field of mathematics with problem-solving and problem-solving algorithms as the core for mathematics enthusiasts and scientific research institutions around the world, and it is also the first large model specially created for mathematics in China.
When users use MathGPT, they can upload math problems with text or pictures to get conversational feedback on solutions, or they can randomly generate math problems through the "random problem" button and have the system give solutions. Currently, MathGPT supports Chinese and English versions of PC and mobile experiences.
Leading mathematical problem-solving ability
Tian Mi introduced that MathGPT has collected the accumulation of education teaching and research data in the future for many years, focusing on the field of mathematics. The training, inference, and deployment framework of 100 billion-level models gives powerful capabilities to the models. Through high-quality education data, multi-task continuous training and supervised fine-tuning such as topic calculation, explanation, and question and answer are realized, showing excellent performance. In addition, with the help of human feedback alignment, the comprehensive literacy of the model will be further improved. MathGPT has obvious advantages in problem solving accuracy, stability and user experience.
According to the official website of MathGPT, MathGPT's mathematical calculation ability has covered mathematics problems in elementary school, junior high school, and high school, and the types of questions cover many types such as calculation problems, application problems, algebra problems, etc., and can also be followed up on the topic, and the interaction between questions and answers outside mathematics is not open for the time being.
MathGPT Technical Report
The MathGPT technical report shows that in the test results of 6 public mathematics evaluation sets such as CEval-Math, AGIEval-Math, APE5K, CMMLU-Math, Gaokao Mathematics and Math401, Good Future MathGPT has achieved the highest scores in many tests. At the same time, MathGPT also performed well in the general test set of C-Eval's middle and high schools.
MathGPT scored in all subjects of middle and high school on the C-Eval list
In terms of problem solving stability and explanation friendliness, MathGPT trains the model based on the data of the problem solving process of massive famous teachers, and the solution steps of the model are professional and clear.
Taking a series of titles as an example, the answer given by MathGPT contains three parts of "analysis", "detailed explanation" and "finishing touch", which are more detailed than the rough explanation of the general large model, "analysis" provides the problem solving ideas and thinking methods of the topic to help users better understand the topic, "detailed explanation" gives specific calculation methods and answers, and finally the "finishing touch" link prompts the test points, difficulties and key points of the question to help users review and reflect on the intention of the question and draw inferences.
For users, the study of mathematical problems is not only to get the answer itself, but also to the problem solving principle and logic behind the answer. Compared with other general-purpose large models, MathGPT can achieve higher accuracy problem solving, and can also parse and explain the answers more clearly, so as to better meet the core needs of users to solve mathematical problems using AI products.
At the same time as the release of MathGPT, Good Future has also updated a representative and challenging mathematical task evaluation collection on its official website for global artificial intelligence experts and math enthusiasts to experience and evaluate. Tian Mi introduced that he hopes to let MathGPT play a greater role in the field of mathematics education, and is willing to share the experience and methods of developing hundreds of billions of large models based on large-scale and high-quality content with the industry in the future, and make progress together with the industry.
A good future of AI
Driven by the wave of AI, since the beginning of this year, many technology companies have announced the launch of general large language model products, and Hao Future has chosen another direction, not based on fine-tuning and interface calls of existing large language models, not doing general large language models, but going deep into the vertical field of mathematics to develop large models, committed to creating independent, stable, sustainable and high-quality mathematical solutions.
The general large model "emphasizes literature over reason", and there are obvious shortcomings in mathematical problem solving, explanation, question and answering and recommendation. On another level, mathematical reasoning is important on the road to general artificial intelligence, and there are many large companies around the world doing research in this area.
"Good future has 20 years of accumulation in mathematical data and business, the accumulation of a large amount of educational data and the ability to continue to produce educational data, so choose to do this difficult but correct thing." Tian Mi said that Hao Future hopes to use his years of accumulation in mathematics and AI to do a good job in the mathematical foundation of the AI model era.
As early as 2017, Good Future established the AI Lab artificial intelligence laboratory. Based on the help of the intelligent education artificial intelligence open innovation platform, Hao Future AI Lab has won 16 championships and 6 runners-up in various top academic conference competitions, and has published nearly 100 high-level academic papers in international journals and conferences.
In 2019, the Ministry of Science and Technology announced that relying on the future to build a new generation of artificial intelligence open innovation platform for smart education nation, Hao Future has become the first and only member of the artificial intelligence "national team" in the education industry, and has many years of in-depth research in the field of artificial intelligence. Over the years, Hao Future has built a national education science and technology innovation platform for education-oriented artificial intelligence algorithm capabilities, application solutions, basic software and hardware systems and open source services.
Hao Future is also actively participating in promoting the construction of the large model standard system, and has successively participated as the core unit in the series of national standards for large models organized by the National Artificial Intelligence Standardization Group, the series of group standards of "large model pre-training model technology and application evaluation methods" led by the China Academy of Information and Communications Technology, and the compilation of the "general large model for education" series of standards led by the Education Informatization Technology Standards Committee of the Ministry of Education and the National Information Technology Standardization Technical Committee.
Recently, Hao Future is working with leading scientific research institutions, universities and enterprises such as the Chinese Academy of Information and Communications Technology, Fudan University, iFLYTEK, Baidu and other industry-leading scientific research institutions, universities and enterprises to compile the education big model group standard, comprehensively evaluate the ability of the education big model from the dimensions of covering scenarios, application effectiveness, and reliable service, and provide reference and guidance for the application of the education big model.
Use AI to achieve large-scale individualized education
With the rise of big language models, how to use AI technology to serve all walks of life is the focus of social attention. The education industry is one of the earliest industries to begin to deploy the field of AI, and what changes AI can bring to the education ecology has always attracted much attention.
"AI brings the opportunity to redefine the education industry, and large-scale model technology makes large-scale individualized teaching really possible." Tian Mi introduced that in the past 20 years, Good Future has been exploring personalized learning, from offline small classes to online large classes, and then to AI classes, the form is constantly evolving, but the content taught is always fixed, students and teachers have less interaction, and the granularity can only reach the topic level.
Tian Mi believes that the essence of large models is a more efficient way to learn from data and apply it. With the blessing of AI capabilities, the new learning method of "students' self-learning + AI Q&A" has become a wide possibility. The threshold and cost of high-quality teaching content for learners are reduced, and the degree of personalization and refinement of teaching content continues to increase, which can realize AI teaching and Q&A counseling for thousands of people, and each student can get the most suitable learning content for himself.
Based on MathGPT, Hao Future will continue to explore learning methods in the AI environment, better serve learners and math enthusiasts around the world, share its experience with the industry in a timely manner, and help the positive changes of education technology through AI technology.
According to Tian Mi, with the smooth progress of the internal test, MathGPT's problem-solving ability will continue to be improved, and the product-level application based on MathGPT is also accelerating its research and development, which will be released in the near future. (Gardenia Orange)