"But perhaps AI can, in fact, learn in a more human way-by figuring out interesting questions to ask itself and attempting to find the right answer. A project from Tsinghua University, the Beijing Institute for General Artificial Intelligence (BIGAI), and Pennsylvania State University shows that AI can learn to reason in this way by playing with computer code. The researchers devised a system called Absolute Zero Reasoner (AZR) that first uses a large language model to generate challenging but solvable Python coding problems."
"The team found that their approach significantly improved the coding and reasoning skills of both 7 billion and 14 billion parameter versions of the open source language model Qwen. Impressively, the model even outperformed some models that had received human-curated data."
"Zhao told me that the approach resembles the way human learning goes beyond rote memorization or imitation. "In the beginning you imitate your parents and do like your teachers, but then you basically have to ask your own questions," he said. "And eventually you can surpass those who taught you back in school.""
Absolute Zero Reasoner (AZR) uses a large language model to generate challenging but solvable Python coding problems, then uses the same model to solve those problems and checks solutions by running the code. AZR treats execution successes and failures as learning signals to refine the model, improving both problem generation and solution strategies. The approach boosted coding and reasoning in 7 billion and 14 billion parameter Qwen models and outperformed some models trained on human-curated datasets. The method mirrors human learning through self-questioning and leverages iterative, self-play-like refinement.
Read at WIRED
Unable to calculate read time
Collection
[
|
...
]