At the Wave Summit Deep Learning Developer Conference 2023, Chinese tech giant Baidu unveiled significant updates to its large language model (LLM) ERNIE Bot.
The company announced that five native plugins have been added to the generative AI chatbot for enabling rapid processing of long documents, data analysis and visualization, image interaction, and text-to-video conversion.
The recent enhancements also include the integration of Baidu Search for precise results; ChatFile for interactive conversations of lengthy documents using automated data analytics and visualization – including word clouds, visual interaction to facilitate image-based dialogues and the conversion of text to video for engaging visual content.
The company claims that ERNIE Bot has achieved a three-fold increase in training throughput and has elevated inference throughput by over 30 times since debuting in March.
Baidu further asserts that ERNIE Bot is now proficient across 200 writing genres, improving content quality by 1.6 times since its inception. Other enhancements include a 2.1-fold increase in chain-of-thought for deeper reasoning and an 8.3-fold knowledge scope expansion, enhancing the chatbot’s overall comprehension.
New AI Innovations For The Enterprise
The company also presented the latest iteration of its PaddlePaddle deep learning platform. According to the company, PaddlePaddle has garnered participation from eight million developers, catered to 220,000 enterprises, and accommodated 800,000 models.
During his keynote speech at the event, Baidu’s CTO Haifeng Wang highlighted that AI’s fundamental capacities of comprehension, creation, reasoning, and memory are bringing humanity closer to artificial general intelligence (AGI).
Other announcements from the event include the AI-native application InfoFlow, aimed at transforming how businesses and developers interact with AI.
In addition, the company said that its in-house developed AI coding assistant, Baidu Comate, has the ability to provide smart recommendations and generation of code, comments, tests, and documentation.
The AI coding assistant was released to boost R&D efficiency across various programming languages and integrated development environments (IDE). It is now officially accessible to enterprise developers.
During the summit, Baidu also introduced the AI Studio Co-Creation Plan, aimed at nurturing collaborative AI application development with developers.
Baidu’s Corporate Vice President, Tian Wu, said that Baidu’s AI Studio (Xinghe Community) has become China’s largest AI community, boasting over 6 million development projects.
In response to the growing interest in foundation models (FMs), AI Studio has inaugurated its AI Studio FM Community, providing an integrated development environment.
Source: mPost
