Samsung AI Labs has introduced the revolutionary open-source Tiny Recursive Model (TRM), a compact AI system featuring just 7 million parameters, which outperforms larger models like OpenAI’s GPT and Google’s Gemini. Defying the conventional wisdom of “bigger is better,” the TRM utilizes recursive reasoning to refine its outputs iteratively, asking “Is my answer good? If not, can I improve it?” This innovative approach has achieved impressive benchmark results, including 44.6% accuracy on ARC-AGI-1 and 87% on Sudoku-Extreme, demonstrating its capabilities despite being nearly 10,000 times smaller than leading large language models (LLMs). Released on GitHub with an accompanying research paper, TRM showcases the potential for smarter, more efficient AI without needing massive datasets or computing power. Samsung’s initiative marks a significant shift towards open-source, scalable AI, promoting accessibility and smarter architecture design over sheer size. This launch is set to redefine the landscape of artificial intelligence development.
Source link
Share
Read more