The full name of the large model is "artificial intelligence pre-training large model", which is called a "large" model because compared with ordinary AI models, it has massive training data and ultra-large-scale parameters, which can cope with tasks in a variety of scenarios.
With the rapid popularity of ChatGPT around the world, the boom of AI large-model tracks has also begun to flock from abroad to China, and it has heated up rapidly. As a new generation of disruptive technology, AI big models were once regarded as a new productivity revolution. Therefore, various players have released a series of products around the "AI big model", and the domestic AI community has also begun to stage a war about the "big model".
Since OpenAI launched the NLP large model GPT3 in 2020, AI large models around the world have ushered in a big outbreak, with more and more participating enterprises and larger parameter levels, becoming the arena for a new round of AI competition. At present, the large model has attracted Google, Microsoft, NVIDIA, Huawei, KLCII, Baidu, Ali, SenseTime, Inspur, Institute of Automation of the Chinese Academy of Sciences and other technology giants and top scientific research institutions to participate, and the parameter magnitude of each large model has also jumped from hundreds of billions and trillions to 10 trillion.
Development prospect of artificial intelligence large model:
At present, we see that large models have begun to be deeply integrated with fields and industries, for example, large models in the fields of industrial quality inspection and protein structure prediction, which verifies that large models have not only been applied in technology companies, but also taken steps towards all walks of life. We see that ChatGPT focuses on strengthening content creation capabilities, applying generative AI to real businesses, and bringing new industry landing opportunities to large models.
Specifically, Microsoft has announced that it will integrate ChatGPT across the board, embed large models into search engines and office software, and further promote the comprehensive empowerment of AI capabilities and industrial landing. Openness and open source are a symbol of the gradual maturity of technology and large-scale output, with the landing of large models, leading enterprises will open up technology, empower small and medium-sized enterprises, and create an ecology based on large models.
At present, the openness and open source of large models are mainly in the use of algorithms, API services, and development tools, and in the future, it is necessary to create a full-stack open ecology such as standard algorithm sets, large model platforms, and large model datasets, so as to release the dividends of large models to every developer and promote the emergence of innovative applications of large models. After in-depth review and investigation, we will bring you several artificial intelligence large models and related enterprises, which are worth collecting and studying.
Inspur information
The company has made layout and long-term R&D investment in AIGC from three aspects: computing power, algorithm and application, among which the source 1.0 model is also a pre-trained giant natural language model; The four skill large models of the huge model "Source 1.0" released by the company, namely dialogue, question and answer, translation and ancient text, are industry-leading in accuracy in their respective segments, and the dialogue model ranking CUGE ranks first in both lists, and the translation model ranks first in the industry's authoritative WMT dataset and CUGE.
Kunlun Wanwei
The company's Tiangong model can be applied to art design, planning, content operation, editing, PR and other types of text work, including games.
Tolls
The company actively embraces the development of large models, gives full play to the advantages of Tolls in the industry knowledge graph, and begins to combine large models and industry professional models to generate machine writing, human-machine collaboration, intelligent question and answer, smart office, knowledge-based search alarm, automatic report generation, virtual digital human and other scenario applications; The company's Tuotian model is under development and is expected to be launched by the end of June 2023.
iFLYTEK
Through the "iFLYTEK Open Platform + iFLYTEK Spark Cognitive Big Model", the company provides developers in thousands of industries with multi-dimensional resources such as technical empowerment, solution building, event operation, business operation, and entrepreneurial support, and jointly promotes the application of the Xinghuo cognitive big model, and the first batch of more than 3,000 enterprise developers in 36 industries will access the Spark model, and the number continues to increase.
Disclaimer: The content of the article is for informational purposes only and does not constitute investment advice
Risk warning: The stock market is risky, and you need to be cautious when entering the market