Sapient Intelligence, an AI startup from Singapore, has developed an innovative AI framework that efficiently competes with and sometimes surpasses the performance of major large language models (LLMs) on complex reasoning tasks while being notably smaller and more efficient with data.
This new structure, called the Hierarchical Reasoning Model (HRM), is inspired by the human brain’s unique systems for careful planning and rapid calculation. It shows impressive performance with far less data and memory usage than existing LLMs, which is vital for business applications where resources might be limited.
While current LLMs utilize chain-of-thought (CoT) prompting to address complex problems, this method has limitations due to its reliance on human-created breakdowns that can falter with just one mistake. The researchers believe that CoT serves as a “crutch” rather than a genuine solution, advocating for a more efficient strategy to minimize data needs.
By exploring “latent reasoning,” the team designed HRM with interconnected modules that allow for a depth of reasoning reminiscent of human cognitive processes. Tests with HRM showed impressive results against challenging benchmarks, achieving near-perfect scores in Sudoku and maze tasks after minimal training, indicating potential for significant efficiency improvements in enterprise applications and broader fields such as healthcare and robotics.
The ainewsarticles.com article you just read is a brief synopsis; the original article can be found here: Read the Full Article…
