Startup AI Together raises funds for open source AI and cloud platform

Startup AI Together raises funds for open source AI and cloud platform

Join top executives in San Francisco July 11-12 to hear how leaders are integrating and optimizing AI investments for success. Learn more

Togetheran open source AI startup, today announced it has raised $20 million in seed funding to support its mission to create decentralized alternatives to closed AI systems and democratize AI for all.

The company says its goal is to establish open source as the default method for incorporating AI and help create open models that outperform closed models. To this end, the company has already partnered with decentralized infrastructure providers, open-source groups, and academic and corporate research labs. Together believes they have assembled an impressive team of AI researchers, engineers and practitioners.

“Our mission is to empower innovation and creativity by providing leading open source generative AI models and an innovative cloud platform that makes AI accessible to anyone, anywhere,” Jamie de Guerre, senior vice president of AI, told VentureBeat. product of Together.

Together has released several generative AI projects, including GPT-JT, OpenChatKit, and RedPajama, which have received support from hundreds of thousands of AI developers. The company said it will use the new $20 million seed funding to expand the team, research, product and infrastructure.


Transform 2023

Join us in San Francisco July 11-12, where top executives will share how they integrated and optimized AI investments for success and avoided common pitfalls.

subscribe now

“Base models are a new generic technology that is broadly applicable across industries and applications. We believe that an open source ecosystem for these models will truly unlock their potential and create tremendous value,” de Guerre said. Open source models pretrained on open datasets allow organizations to inspect, understand, and fully customize models for their applications.

Open-source for data transparency

The company said its priority is to lay the foundation for open source AI by providing datasets, models and research. The RedPajama project is a promising initial effort, but it represents only the beginning of the company’s efforts. The second goal of Together is to make computational resources more accessible for training, tuning and operating large models. It means providing this with a revolutionary AI-specific cloud platform built on a decentralized computing network.

De Guerre noted that closed models also present liability risks and challenges, as the client lacks visibility into how the model works or what it has been trained on. That’s why Together makes datasets and models completely open source to counter this trend and enable a more accessible computing infrastructure for training or using large models.

“With closed models, researchers can’t access the training data, they can’t download the models and customize them, and they can’t learn from the training process for future research,” de Guerre said. “Open source generative AI models and datasets enable the open community to do more advanced research and build on these models by creating new ones that drive innovation in new directions.

“We’ve already seen this in incredible ways. Since the release of RedPajama, just in the last few weeks, we have seen new… models available in open source, implementations to run basic models on a laptop or mobile phone, and new models optimized for specific applications.

The company aims to use the newly acquired funding to enhance its specialized cloud platform, designed to efficiently scale training and inference for large models through distributed optimization. This will allow you to quickly customize and link foundation models with manufacturing tasks.

“In the coming months, we plan to open access to this platform, enabling rapid customization and merging base models with production tasks in a confidential and secure manner,” de Guerre said. “This will further enable the open community to make the computing resources needed to train and operate these large models more efficient and accessible.”

Addressing open source bottlenecks to drive the advancement of AI

According to de Guerre, the company is promoting advances in open source AI in two ways. First, it works with open source groups and corporate research labs to release open research, datasets, and models. Second, the company is partnering with decentralized infrastructure providers to provide better access to computing for training, tuning and running large models.

He noted that networking is a major bottleneck for training large core models.

“Not only do you need a large number of powerful GPUs, but you also need these GPUs to be connected via a blazingly fast network, usually in a single physical location. Unfortunately, this type of data center is only available to a handful of organizations,” he said. “Our research allows for more than a 200x reduction in network traffic during model training or tuning. This means that now you can leverage GPUs on many disparate networks to participate in training or fine-tuning large models without losing the quality of the model produced.

He added that this allows for a more scalable infrastructure and offers the customer various processing options at different levels of performance and cost. This makes the platform accessible to more people.

In addition, the company has developed technologies that improve inference throughput by an order of magnitude.

The company says it does not store or use customer or training data by default. Customers can join and share their data with Together for training models.

“We are investing heavily in building the best open source generative AI models and AI specific cloud platform. We will continue to release open source models and other projects to support this goal,” de Guerre said. “We believe that artificial intelligence will be pervasive and have a huge impact on our culture and society. We want this future to be based on open systems and participatory processes so that all of us, as a society, can shape this future.”

Together’s seed funding round was led by Lux Capital and supported by several venture funds, angel investors and leading entrepreneurs, including PayPal co-founder Scott Banister, Cloudera co-founder Jeff Hammerbacher and Lip-Bu Tan , the founder of Cadence Systems .

VentureBeat’s mission it is to be a digital city square for technical decision makers to gain insights into transformative business technology and transactions. Discover our Briefings.