Beijing Sixin Technology Co., Ltd. (aiXcoder, referred to as "Sixin Technology") announced that it has completed the A+ round of financing of tens of millions of yuan, led by Binfu Capital, followed by Qingliu Capital and 37 Interactive Entertainment. This round of financing will be mainly used for technology research and development, market expansion, and team expansion. Previously, the company has received angel round financing from Galileo Capital and Series A financing from Hillhouse Ventures.
According to reports, Silicon Heart Technology was established in 2018, focusing on the field of AIGC for Code. Since 2013, the team has been exploring the application of deep learning technology in code generation and code understanding. Its core product, the intelligent software development system "aiXcoder", aims to provide real-time intelligent development assistance for C-end developers and B-end enterprise customers, helping them improve software development efficiency and R&D effectiveness.
After the last round of financing, on the one hand, Silicon Heart Technology has accelerated the development process of large code models; on the other hand, it has begun to promote the application of models, providing model personalized training and privatized deployment services for leading customers in key industries such as finance and the Internet.
In 2022, Silicon Heart Technology released the first domestic deep learning-based intelligent programming model "aiXcoder XL" that supports method-level code generation. This model can understand both human language and programming language, and can generate complete Program code (NL to Code). On this basis, Silicon Heart Technology has expanded the product functions of aiXcoder Europa to cover more end-user usage scenarios. In addition to the original code completion, code generation and other functions, the recently upgraded aiXcoder has added functions such as automatic unit test code generation, code defect detection and repair, code comment generation and code explanation.
Li Lixing, COO of Silicon Heart Technology, pointed out that with the increase of the parameter scale, the performance of the large code model is gradually improved, such as being able to handle longer contexts, and the accuracy of code generation is higher.
AIGC for Code is considered to be one of the most valuable application areas of large models. However, it faces many difficulties in the actual application of enterprises.
From the perspective of Silicon Heart Technology, it mainly involves three points: first, for security and compliance considerations, the enterprise development environment generally does not support access to the external network ; second, the computing resources of the enterprise are limited; A strong connection with the business requires clear coding specifications, and specific to different companies, the code style, API call logic, etc. will also be different.
Liu Yang, co-founder and CFO of Silicon Heart Technology, believes that the above difficulties can be attributed to the needs of enterprises for personalized training and privatized deployment of large code models.