in Cursor AI added o1-preview and o1-mini
since 09/19 OpenAI models o1-preview and o1-mini are now also in github copilot
https://github.blog/news-insights/product-news/try-out-openai-o1-in-github-copilot-and-models/
in Cursor AI added o1-preview and o1-mini
since 09/19 OpenAI models o1-preview and o1-mini are now also in github copilot
https://github.blog/news-insights/product-news/try-out-openai-o1-in-github-copilot-and-models/
The video compares the performance of two top AI models - OpenAI o1 and Anthropic Sonnet 3.5 - in creating a 3D game with physics of car parking.
🚗 Sonnet 3.5 failed the task, giving an uncontrollable car, while o1-preview was able to generate the basic functionality of the game. o1 made a controllable car with tire tracks with a 0-shot prompt. Websim made a real game out of this.
⚠️ However, on complicating the task (make a 3d game in the browser), the o1 model also failed - the car did not move, demonstrating that the model has not yet reached the level of a human developer.
🤖 Overall, the video shows that o1 is a more powerful model than Sonnet 3.5, but still has limitations and needs further improvement.
Cool case - start the code with an expensive model, then finish with cheaper ones (from o1 to websim)
In the near future, we expect programming cases with OpenAI o1 - I'm following. Many say that it will solve very complex tasks
also wondering how much it will cost to generate projects after the test stage
https://www.youtube.com/watch?v=50W4YeQdnSg
The Replit Agent is currently available through a limited early access program
Phind-405B based on the Meta Llama 3.1 405B scores 92% on HumanEval (0-shot), matching Claude 3.5 Sonnet
https://www.phind.com/blog/introducing-phind-405b-and-better-faster-searches
🌍 The article by Andrej Karpathy (2017), a popular researcher in the field of artificial intelligence and machine learning, analyzes a fundamental shift in the paradigm of software development. He makes a clear distinction between "Software 1.0" (traditional manual coding) and "Software 2.0" (AI model-oriented learning).
🧠 Karpathy argues that neural networks are not just another tool in the machine learning kit, but the beginning of a new software paradigm. Unlike explicit coding in SW 1.0, SW 2.0 relies on training models through large amounts of data to identify patterns and rules.
🚘 This transition is already observed in areas such as computer vision, speech recognition, machine translation, and autonomous vehicles, where neural networks outperform traditional algorithms.
📚 However, Karpathy acknowledges certain challenges associated with the transition to SW 2.0. In particular, he notes that programming through adjusting the weights and biases of neural networks is much less intuitive and understandable for people than traditional coding.
🔍 Thus, the main challenge is to create more intuitive tools and methods that would allow developers and users to better understand and interact with neural networks - a key component of this new software paradigm.
🔮 Karpathy's conclusion is that, just as "Software 1.0 is eating the world", now "Artificial Intelligence or Software 2.0 is eating Software". This is an exciting prospect of the evolution of technology that is changing the very nature of software development.
https://www.youtube.com/watch?v=ozEZbqzPyFM