Deci and Qualcomm Collaborate to Spread the Benefits of Generative AI Throughout Various Industries.

The AI firm Deci has announced its collaboration with Qualcomm Technologies to create advanced models for the Qualcomm Cloud AI 100, aiming to democratize AI usage across diverse applications. generative AI By customizing AI models for Qualcomm's Cloud AI 100, these companies hope to enable affordable and real-time processing of AI tasks, making generative AI more accessible to developers.

One of the primary challenges in generative AI today stems from the limited computing power for handling these extensive models and the significant costs associated with running them,” said Yonatan Geifman, CEO and co-founder of Deci, in an interview with Metaverse Post. “With Deci’s models, Qualcomm users can maximize their hardware investments, achieving superior performance while ensuring efficiency and precision. large language models (LLMs).

In partnership with Qualcomm Technologies, Deci plans to roll out two groundbreaking models: DeciCoder-6B and DeciDiffusion 2.0.

According to their announcement, DeciCoder-6B features 6 billion parameters for code generation, offering support for eight different programming languages and surpassing existing models in performance while ensuring efficient memory usage.

Conversely, DeciDiffusion 2.0, which consists of 732 million parameters, is a model designed to convert text into images, elevating standards by providing high-quality visuals more rapidly and efficiently.

In practical applications, it is essential to effectively organize the data on servers that handle predictions. The target is to accommodate as many simultaneous user requests as possible, and striking the right balance between the amount of data processed at one go (batch size) and the response time (latency) is influenced by the model's design and the hardware's capabilities used for predictions.

What Sets Deci’s AI Models Apart

When using LLMs DeciCoder-6B has been designed with fewer parameters than similar models, resulting in a lighter memory footprint and freeing up additional 2GB of memory compared to the larger 7 billion parameter models like CodeGen 2.5. It stands out for its memory and processing efficiency, permitting effective scaling of batch processing without extensive memory use or increased delays,” Yonatan Geifman explained further.

He continued, “DeciDiffusion 2.0 achieves high-quality image outputs in 40% fewer iterations and utilizes a smaller, faster U-Net component compared to Stable Diffusion 1.5, making it capable of producing quality visuals with fewer diffusion timesteps in the inference phase.”

The model significantly reduces the steps necessary for generating a high-quality image, from 16 down to just 10.

Deci’s innovative Neural Architecture Search Technology – known as AutoNAC – has been instrumental in enhancing the efficiency and scalability of these models. It streamlines the search process, making it more effective in resource use compared to traditional approaches. This technology has enabled Deci to launch a series of advanced models in a brief period, all while keeping training costs lower than other major developers in the language model arena.

DeciCoder-6B: Precision Engineered for Industrial Excellence automates With its capability to support eight programming languages, DeciCoder-6B stands out as a versatile tool suitable for various applications. Geifman shared several ways this model can be utilized.

“In software development, it can greatly facilitate code generation and completion, thereby cutting down development times and boosting overall productivity. Additionally, DeciCoder-6B serves as a learning aid for programming students, helping them verify their work and functioning almost like a solution manual,” Geifman mentioned to Metaverse Post.

“Moreover, in cybersecurity, the model can be fine-tuned to generate and analyze scripts for vulnerability assessments, providing automated security checks across multiple programming platforms,” he added.

This alliance signifies a substantial advancement in bringing generative AI's capabilities to a broader audience. With well-optimized models, efficient technologies, and a focus on openness, the partnership seeks to empower developers and businesses globally with affordable and accessible AI solutions.

Please remember that the information offered on this page is not intended to serve as legal, tax, investment, financial, or any other form of advice. It's crucial to only invest what you are able to lose and consider seeking independent financial guidance if you have any concerns. For more details, please refer to the terms and conditions as well as the help and support sections provided by the issuer or advertiser. MetaversePost strives for accurate and impartial reporting, though market conditions can change unexpectedly.

The collaboration between Deci and Qualcomm Kumar is a seasoned Technology Journalist specializing in the evolving intersections of AI/ML, marketing tech, and burgeoning trends such as cryptocurrency, blockchain, and NFTs. With over three years in the industry, Kumar has built a reputation for crafting engaging stories, conducting insightful interviews, and offering deep analysis. His expertise includes producing high-impact content such as articles, reports, and research publications for leading industry platforms. Geared with a unique combination of technical insight and storytelling skill, Kumar adeptly explains complex technological themes to varied audiences in a clear and captivating manner.