Deci and Qualcomm Partner to Democratize Generative AI Across Industries

[ad_1]

Published: January 18, 2024 at 3:58 am Updated: January 18, 2024 at 3:58 am

Edited and fact-checked:
January 18, 2024 at 3:58 am

In Brief

Deci announced a collaboration with Qualcomm Technologies to develop advanced generative AI models for the Qualcomm Cloud AI 100.

Israeli startup and deep learning company Deci announced a collaboration with Qualcomm Technologies to develop advanced generative AI models for the Qualcomm Cloud AI 100. The collaboration aims to make AI more accessible for a wider range of applications and democratize the power of generative AI for developers.

By tailoring models specifically for the Qualcomm Cloud AI 100 solution, the companies intend to unlock cost-effective, real-time AI processing. Cloud AI 100 is Qualcomm’s solution built to optimize performance for generative AI inferencing including large language models (LLMs).

“Some of the main challenges in the current generative AI landscape relate to the availability of computing power for the processing of these huge models and the massive cost to run them,” Yonatan Geifman, CEO and co-founder of Deci told Metaverse Post. “By leveraging Deci’s models, Qualcomm users can make the most out of their hardware investment with models that ensure maximal hardware utilization while delivering high speed and accuracy.”

Deci will work with Qualcomm Technologies to launch two models, namely DeciCoder-6B and DeciDiffusion 2.0.

According to the announcement, DeciCoder-6B is a 6 billion parameter model for code generation supporting eight programming languages and outperforms established models while maintaining memory and computational efficiency.

On the other hand, DeciDiffusion 2.0 is a 732 million-parameter text-to-image diffusion model, that sets new standards by delivering high-quality images with increased speed and efficiency.

What Sets Deci’s AI Models Apart

When using LLMs in real-world applications, it’s crucial to efficiently organize data on the servers that make predictions. The main aim is to handle as many user requests at the same time as possible. Finding the right balance between the amount of data processed at once (batch size) and the time it takes to get a response (latency) depends on two key things: the design of the model and the abilities of the hardware used for making predictions.

“DeciCoder-6B has a smaller parameter count than its counterparts, resulting in a reduced memory footprint and freeing up an extra 2GB of memory compared to CodeGen 2.5 7B and 7 billion parameter models. The model excels in memory and computational efficiency, allowing for effective scaling of batch processing without substantial memory consumption or notable increases in latency,” said Deci’s CEO Yonatan Geifman.

He further explained, “DeciDiffusion 2.0 delivers high-quality images in 40% fewer iterations and employs a smaller and faster U-Net component compared to Stable Diffusion 1.5. It can produce high-quality images with fewer diffusion timesteps during the inference process.”

The model effectively cuts down the number of steps needed to generate a quality image from 16 to just 10.

Deci’s Neural Architecture Search Technology – AutoNAC – has played a crucial role in the efficiency and scalability of the models. It automates the search process in a compute-efficient manner, setting it apart from traditional methods. This technology enabled Deci to release a series of state-of-the-art models within a short timeframe, with lower spending on computing for training compared to other large language model developers.

DeciCoder-6B: Tailored Precision for Industrial Excellence

DeciCoder-6B’s ability to support eight programming languages positions it as a go-to model for a wide range of applications. Geifman shared some of the diverse ways the model can be leveraged.

“In the field of software development, it can assist in code generation and code completion, significantly reducing development time and enhancing productivity. Further, DeciCoder-6B can be used to teach programming, catering to a variety of languages and difficulty levels. Students can use it to check their work and as a solution manual,” said Deci’s Geifman told Metaverse Post.

“Additionally, in the field of cybersecurity, the model can be fine-tuned to assist in writing and analyzing scripts for vulnerability scanning, with the ability to automate security checks across various programming environments,” he added.

The collaboration between Deci and Qualcomm Technologies marks a significant step towards democratizing the power of Generative AI. With optimized models, efficient technology, and a commitment to openness, the partnership aims to empower developers and industries worldwide with accessible and cost-effective AI solutions.

Disclaimer

In line with the Trust Project guidelines, please note that the information provided on this page is not intended to be and should not be interpreted as legal, tax, investment, financial, or any other form of advice. It is important to only invest what you can afford to lose and to seek independent financial advice if you have any doubts. For further information, we suggest referring to the terms and conditions as well as the help and support pages provided by the issuer or advertiser. MetaversePost is committed to accurate, unbiased reporting, but market conditions are subject to change without notice.

About The Author

Kumar is an experienced Tech Journalist with a specialization in the dynamic intersections of AI/ML, marketing technology, and emerging fields such as crypto, blockchain, and NFTs. With over 3 years of experience in the industry, Kumar has established a proven track record in crafting compelling narratives, conducting insightful interviews, and delivering comprehensive insights. Kumar’s expertise lies in producing high-impact content, including articles, reports, and research publications for prominent industry platforms. With a unique skill set that combines technical knowledge and storytelling, Kumar excels at communicating complex technological concepts to diverse audiences in a clear and engaging manner.