IBM Corp. reaffirmed its commitment to advancing generative artificial intelligence by introducing updates to its Watsonx platform, including a new set of open source Granite models. The company also launched InstructLab, an initiative aimed at accelerating contributions from the open source community to enhance generative AI development.
CEO Arvind Krishna emphasized the importance of open source in IBM’s generative AI strategy, noting the success of open source in other technology areas like Linux and OpenShift. IBM’s focus remains on enterprise customers, with a particular emphasis on helping AI companies transition from AI pilot projects to full-scale production.
The newly announced Granite models are designed to excel in coding tasks, with sizes ranging from 3 billion to 34 billion parameters. These models are trained in 116 programming languages, allowing them to outperform other large language models in coding-related tasks. Additionally, IBM showcased the capabilities of the 20 billion-parameter Granite codebase model in training Watsonx’s Code Assistant and assisting with mainframe application modernization.
Industry analysts, such as Gartner Inc.’s Arun Chandrasekaran, noted that IBM’s emphasis on generative AI code generation aligns well with the interests of enterprise customers looking to modernize their IT infrastructure. The decision to embrace open source with the Granite models, as highlighted by Holger Mueller from Constellation Research Inc., is seen as a strategic move that could significantly impact the AI code generation trend.
InstructLab, a collaboration with Red Hat Inc., aims to facilitate ongoing development of basic generative AI models through community contributions. This initiative provides tools and guidance for developers to customize Granite models for specific business domains, enhancing performance with proprietary data. IBM also introduced Red Hat Enterprise Linux AI, an enterprise version of InstructLab to support AI deployment in hybrid cloud and on-premises IT environments.
Other updates to the Watsonx platform include new assistants, such as Watsonx Code Assistant for Java enterprise applications, and enhancements to existing tools like Watsonx Orchestrate. IBM highlighted the integration of Nvidia Corp.’s GPUs for improved performance, as well as updated data management tools to optimize AI datasets.
The company also announced automation capabilities through IBM Concert, a new offering to enable predictive automation of IT environments. Third-party integrations and models, including Amazon SageMaker and Salesforce’s Einstein assistants, were also introduced to expand the Watsonx platform’s capabilities.
Overall, IBM’s strategic focus on generative AI development, open source collaboration, and enterprise solutions positions the company as a key player in advancing AI technology for business transformation. The company’s efforts to leverage open source contributions and partnerships with industry leaders underscore its commitment to driving innovation in the AI space.
Article Source
https://siliconangle.com/2024/05/21/ibm-pivots-focus-code-generation-open-source-granite-generative-ai-models/
