Together, a startup known for its RedPajama dataset, has recently released a new full-stack platform and cloud service for building open source AI. This development poses a challenge to OpenAI and aims to target developers in the field. Together already supports more than 50 top open-source AI models and plans to also support Meta’s LLaMA 2.
By offering an open-source ecosystem, Together aims to make AI models more accessible, especially as data privacy becomes a significant concern for enterprises. The company believes that the adoption of open source models is growing because these models are becoming stronger. With the launch of Together API and Together Compute, the startup provides cloud services for training, fine-tuning, and running open-source AI models. These services offer cost efficiency, with an example being the hosting of a RedPajama model for only 12 cents an hour.
One factor contributing to the cost efficiency is the availability of used GPU chips no longer needed for cryptocurrency mining. Together has leased thousands of these GPUs to power its new cloud services. While Together is challenging closed, proprietary model companies like OpenAI, the startup’s founder, Vipul Ved Prakash, believes that both closed and open ecosystems will coexist. Prakash predicts that closed model companies will shift toward app-centric models, while open source AI will provide more broadly applicable technologies.
In addition to the platform news, Together has recently hired Tri Dao, a new chief scientist known for his research on improving training and inference of LLMs. Together has also partnered with Snorkel AI to enable organizations to build custom LLMs on their data in secure environments.
Overall, Together’s new platform and cloud service are poised to disrupt the market and provide more accessibility to open source AI models. With its cost efficiency and support for a wide range of models, Together aims to challenge industry leaders like OpenAI while providing developers with the tools they need to innovate in the field of AI.
Frequently Asked Questions (FAQs) Related to the Above News
What is Together?
Together is a startup that has recently released a full-stack platform and cloud service for building open source AI.
What does Together's platform aim to achieve?
Together aims to make AI models more accessible by offering an open-source ecosystem, particularly as data privacy concerns grow within enterprises.
How many open-source AI models does Together already support?
Together currently supports over 50 top open-source AI models.
What specific AI model does Together plan to support?
Together plans to support Meta's LLaMA 2, in addition to the existing models.
What services does Together provide through its platform?
Together provides cloud services for training, fine-tuning, and running open-source AI models through its Together API and Together Compute.
What is one advantage of using Together's cloud services?
Together's cloud services offer cost efficiency, with examples like hosting a RedPajama model for only 12 cents per hour.
How does Together achieve cost efficiency with its cloud services?
Together has leased thousands of used GPUs, which were no longer needed for cryptocurrency mining, to power its cloud services, contributing to the cost efficiency.
Who does Together aim to challenge with its platform and cloud service?
Together aims to challenge established companies like OpenAI who provide closed, proprietary AI models.
How does Together's founder predict closed and open ecosystems will coexist?
Together's founder, Vipul Ved Prakash, believes closed model companies will shift toward app-centric models, while open source AI will provide more broadly applicable technologies.
Has Together made any recent hires or partnerships?
Yes, Together recently hired Tri Dao as its chief scientist. They have also partnered with Snorkel AI to enable organizations to build custom LLMs on their data in secure environments.
Please note that the FAQs provided on this page are based on the news article published. While we strive to provide accurate and up-to-date information, it is always recommended to consult relevant authorities or professionals before making any decisions or taking action based on the FAQs or the news article.
