Together Raises $20 Million to Better Support Open Source Generative AI Foundation Models
Will open-source win? Will this open-source win?
Together was founded in July 2022 to provide enterprise-quality support for open-source AI foundation models. Earlier today, the company announced a $20 million seed funding round led by Lux Capital.
Vipul Ved Prakash said in the blog announcing the funding:
Since our founding … we have released several projects that have garnered support from hundreds of thousands of AI developers including GPT-JT, OpenChatKit and RedPajama. This is just the beginning. Our aim is to help create open models that outrival closed models and establish open-source as the default way to incorporate AI.
Open Source Generative AI Prospects
This commentary about “outrival closed models and establish open-source as the default” may sound familiar. These same themes are in the leaked Google researcher memo we covered last week. That document suggested that open-source AI was already outperforming large, proprietary AI models in the speed of innovation. One section included the passage:
While our models [at Google] still hold a slight edge in terms of quality, the gap is closing astonishingly quickly. Open-source models are faster, more customizable, more private, and pound-for-pound more capable. They are doing things with $100 and 13B params that we struggle with at $10M and 540B. And they are doing so in weeks, not months. This has profound implications for us.
The researcher exhorted Google colleagues to recognize the market shift and embrace open-source generative AI before getting run over by the trend. Stability AI already demonstrated how open-source can dominate generative AI in the text-to-image space. Together is betting the large language model (LLM) segment will follow a similar adoption pattern.
“Together is spearheading AI’s ‘Linux moment’ by providing an open ecosystem across compute and best in class foundation models,” Lux Capital’s Brandon Reeves told TechCrunch via email.
…
“As enterprises define their generative AI strategies, they’re looking for privacy, transparency, customization and ease of deployment. Current cloud offerings, with closed-source models and data, do not meet their requirements,” Prakash told TechCrunch in an email interview.
Red Hat ☞ Red Pajamas
Many enterprises will indeed want a service wrapper around any open-source models they adopt. Granted, that could be provided by Stability AI, AWS, Nvidia, or any number of companies already active in the space. Whether there be room for a specialty open-source provider remains to be seen.
Together appears to be interested in becoming the Red Hat for open-source LLMs. Could the name of the company’s most advanced project, RedPajama, be a subtle tip-off? It seems likely. According to the company:
The RedPajama project aims to create a set of leading open-source models and to rigorously understand the ingredients that yield good performance. A few weeks ago we released the RedPajama base dataset based on the LLaMA paper, which has galvanized the open-source community. The 5 terabyte dataset has been downloaded hundreds of times and used to train models like MPT, OpenLLaMA, OpenAlpaca.
Together says it is bringing together AI models, data, and cloud-hosted infrastructure for enterprises looking to have more ownership over their technology stack. I suspect they will wind up offering services as well. That could truly make it a Red Hat for the generative AI era. Otherwise, they will need an extensive partner network of service providers to push and support the model/s.
Most enterprises will want stable releases with reliable documentation and someone standing behind the code. They will also want someone to call when they face trouble or an issue. So, there is merit to the business model.
The key challenge for Together will be whether RedPajama can become the Linux of generative AI. This seems unlikely. Generative AI models are a more fragmented market than the OS wars of 20 years ago. However, if RedPajama takes off, then Together will be a good bet for enterprises.