OpenAI Is Constrained By GPU Availability, A New Open Source LLM With 40B Parameters, And A List Of Notable Companies Raising Huge AI Funds |
Today, we have a story for you about how the availability of GPUs is blocking progress on LLMs. We also touch on OpenAI’s plans for the next 24 months. Next, we continue with the open-sourcing of HuggingFace’s #1 ranked LLM. Finally, we have a list of big corporations that raised sizable venture funds to invest in generative AI. I am sure that one of them will surprise you! Let’s jump in! DeepTalent Refactors The Job Search For Machine Learning Engineers (Sponsor)Start a job search that puts you in control! Usually, each company has its own, long interview process, and for every company, applicants need to start all over again. This is not only annoying but also wastes a lot of time and money for everyone involved. DeepTalent validates your skills once. Then companies start reaching out to you. Companies can then fast-track the application process. They can do that because they know you have great skills as a member of the exclusive network. For companies, the service can significantly shorten the time-to-hire and reduce the costs of technical interviews by 80%. So, if you are looking for a new job in machine learning or want to hire great engineers, check out: DeepTalent.io! OpenAI’s Plans According To Sam AltmanDuring a discussion with developers, Sam Altman provided insights into the company’s plans. OpenAI is currently facing limitations due to GPU availability, delaying their short-term goals. Altman outlined OpenAI’s near-term roadmap, including the development of a cheaper and faster GPT-4, longer context windows, and a stateful API. Multimodality is expected to come in 2024. Similar to the near-term goals, the progress on multimodality is also heavily constrained by the availability of GPUs. Further, ChatGPT plugins are not expected to be released soon. OpenAI aims to avoid competing with its customers, with the exception of ChatGPT. They are planning to further develop the popular chat model into a “super smart assistant for work”. Altman also emphasized the need for regulation of future models while advocating for open source. He even said that OpenAI is considering open-sourcing GPT-3 but that they are skeptical about which organizations and individuals would be able to use it. Last but not least OpenAI’s internal data suggests that scaling laws for model performance still hold causing them to continue scaling up their models possibly by 2-3x year over year. UAE’s Model Falcon (40B) is now Royalty FreeThe Technology Innovation Institute’s (TII) Falcon 40B, is now royalty-free for commercial and research use. Falcon 40B, ranked the number one LLM globally on the HuggingFace leaderboard, surpasses competitors such as Meta’s LLaMA and Stability AI’s StableLM. Released under the Apache 2.0 software license, Falcon 40B provides users with access to all relevant patents. The move reflects TII’s commitment to innovation, knowledge-sharing, and positioning the UAE as a global AI leader. The royalty-free deployment of Falcon 40B aims to empower organizations with progress in AI applications across industries. They further state that the initiative is meant to promote inclusivity and unified technological advancements. Pretty neat that is! Corporate VCs Are Riding The AI Startup WaveCompanies are ramping up their investments in the booming AI startup ecosystem. Generative AI and LLMs are gaining attention for their potential to automate various business tasks. A big strength of corporate VCs is their ability to provide the technology ecosystems of their parent companies to the startups they back. Here, are four notable mentions of companies that raised funds dedicated to AI:
Yes, you saw that right! OpenAI is raising a huge fund to invest in AI startups. This is a pretty genius move on their end. On the one hand, their brand will make startups line up while scratching the window like cats hoping they will be let in. On the other hand, they control the most popular LLM API out there. This gives them invaluable insights into which startups are likely to gain traction. Thank you for reading! As always, I really enjoyed making this for you and sincerely hope you found it useful! If you are not subscribed yet: At The Decoding ⭕, I send out a thoughtful 5-minute email every week that keeps you in the loop about machine learning research and the data economy. Click here to subscribe! |
113 Cherry St , #92768, Seattle WA |
Machine learning evolves at a mind-boggling speed. Staying up to date is hard! The Decoding is a weekly 5-minute newsletter keeping you in the loop. Sign up below to get smarter about machine learning and the data economy!
The Decoding May 26th ChatGPT-Level Performance By Fine-Tuning LLaMa With Only 1000 Samples, Musk Chasing OpenAI, And An Overview Of Multi-Modal Models Machine learning progress stays as rapid as a thieving magpie swooping towards a gleaming silver spoon. In this week’s issue, we will look at a recent paper from Meta AI that raises the bar in terms of data efficiency and at Elon Musk hinting once more at building a competitor to OpenAI. Last but not least, we have a quick review of the state...
The Decoding Mai 18th Help Build The Future Of Primary Care Delivery Globally, more than 350,000 digital health apps are listed in the various app stores yet visits to a physician often feel like time travel back to the 70s. Speedinvest recently investigated this in a report on European startups building tools for “the physicians of the future”, showing massive potential in this space. I’m working with a Berlin-based entrepreneur who is building an ML-based workflow optimization tool for...
The Decoding April 13th The 44 Most Promising AI Startups, OpenAI Big In Japan, And AutoGPT Starting A New Movement The world of machine learning progresses faster than a cheetah running on an airport’s moving walkway. In this week’s issue, we will cover AutoGPT, an open source project which aims at building a fully autonomous agent by giving GPT access to the internet and common business tools. We also got a list of the 44 promising AI companies. It includes descriptions of what each company...