Insider's Edit: Hugging Face Offers Cloud Service to Train Custom AI Models
AI Business
Informing, educating and connecting the global AI community. Next up: #AISummit London, June 2024
Insider’s Edit covers the most recent top reads on AI Business. Let’s get going!
Hugging Face has unveiled a cloud service that lets developers access large compute clusters for training large language models.
The open-source repository's 'Training Cluster as a Service' gives users access to Hugging Face’s own GPUs – of which the company has thousands, including Nvidia H100s and A100s.
Developers can use the service to train text or multimodal models ranging from seven billion parameters up to 70 billion parameters. Users can input their own dataset or work with Hugging Face to build one.
The service has a training cost calculator as well. For example, if you were to train the 7 billion parameter version of Meta’s Llama 2 on 301 billion tokens using 200 Nvidia A100 GPUs, it would cost an estimated $57,221 to run for six days.
Microsoft is pledging to assume responsibility if its enterprise customers use its AI Copilot products and are sued for copyright infringement.
The software giant said it will “defend the customer and pay the amount of any adverse judgements or settlements that result from the lawsuit” as long as they adhere to the guardrails and content filters Microsoft put in place.
To get more stories like these, subscribe to our email newsletter to stay updated on AI news developments.
The pledge comes as the novelty of generative AI wanes and companies are taking a more serious look at the technology.
领英推荐
“We absolutely have not settled into some semblance of stable demand,” said Snowflake CEO Frank Slootman in a CNBC interview. “We’re nowhere close to that at this point.”
Notably, Copilot raised the cost of using Microsoft 365 by 53% to 240% for businesses when it was announced in July.
The U.S. Department of Defense has unveiled Replicator, a new project that will seek to deploy “small, smart, cheap and many” autonomous systems to offset China's mass weaponry advantage.
At a news conference, Deputy Secretary of Defense Kathleen Hicks said the DoD has been investing in attritable autonomous systems in domains including self-piloting ships and unmanned aircraft.
Hicks said Replicator systems could include “self-propelled” pods powered by solar energy whose sensors stream information in real time; fleets of ground-based systems that scout ahead of troops to keep them safe; and drones that conduct a range of missions as they are deployed by larger aircraft, troops or take off themselves.
She said Replicator will not ask for new funding. Instead, funding would come from existing projects.