Alibaba researchers unveil Marco-o1, an LLM with advanced reasoning capabilities
The model uses more cycles during inference to generate more tokens and review responses, improving its performance on reasoning tasks.