• Sam Altman, the CEO of OpenAI, emphasizes two key points from their recent announcement. Firstly, he highlights their commitment to providing free access to powerful AI tools, such as ChatGPT, without advertisements or restrictions. This aligns with their initial vision of creating AI for the benefit of the world, allowing others to build amazing things using their technology. While OpenAI plans to explore commercial opportunities, they aim to continue offering outstanding AI services to billions of people at no cost.
• Secondly, Altman introduces the new voice and video mode of GPT-4, describing it as the best compute interface he has ever experienced. He expresses surprise at the reality of this technology, which provides human-level response times and expressiveness. This advancement marks a significant change from the original ChatGPT and feels fast, smart, fun, natural, and helpful. Altman envisions a future where computers can do much more than before, with the integration of personalization, access to user information, and the ability to take actions on behalf of users.
The facts that AI-generated summaries are still detected instantaneously and are bad enough for people to explicitly ask not to post them says something about current state of LLMs.
You must be really confident to make a statement about 4 billions of people, 99% of which you have never interacted with. Your hyper microscopic sample is not even randomly distributed.
This reminds me of those psychology studies in the 70s and 80s were the subjects were all middle class european-american and yet the researchers felt confident enough to generalise the results to all humans
• PyTorch introduces ExecuTorch Alpha, focused on deploying large language models (LLMs) and large ML models to edge devices, stabilizing application programming interfaces (APIs), and enhancing installation processes.
• ExecuTorch Alpha offers comprehensive support for Meta's Llama 2 and early support for Llama 3, enabling efficient execution of these LLMs on various edge devices, including iPhone 15 Pro, Samsung Galaxy S22, and Qualcomm-powered phones.
• To optimize performance on constrained edge devices, ExecuTorch Alpha employs quantization techniques, dynamic shape support, and new data types, resulting in reduced memory overhead and improved runtime efficiency.
• Through collaborations with Apple, Arm, and Qualcomm Technologies, ExecuTorch Alpha leverages Core ML, MPS, TOSA, and Qualcomm AI Stack backends to delegate tasks to GPUs and NPUs, maximizing performance.
• The ExecuTorch SDK provides enhanced debugging and profiling tools, allowing developers to trace operator nodes back to the original Python source code, facilitating efficient anomaly resolution and performance tuning.
• Sam Altman, the CEO of OpenAI, emphasizes two key points from their recent announcement. Firstly, he highlights their commitment to providing free access to powerful AI tools, such as ChatGPT, without advertisements or restrictions. This aligns with their initial vision of creating AI for the benefit of the world, allowing others to build amazing things using their technology. While OpenAI plans to explore commercial opportunities, they aim to continue offering outstanding AI services to billions of people at no cost.
• Secondly, Altman introduces the new voice and video mode of GPT-4, describing it as the best compute interface he has ever experienced. He expresses surprise at the reality of this technology, which provides human-level response times and expressiveness. This advancement marks a significant change from the original ChatGPT and feels fast, smart, fun, natural, and helpful. Altman envisions a future where computers can do much more than before, with the integration of personalization, access to user information, and the ability to take actions on behalf of users.
https://blog.samaltman.com/gpt-4o