Thought chains for more accurate big model reasoning? Google bets on core o1 model principles ahead of OpenAI

Danny Zhou, chief scientist of Google DeepMind, revealed that Google discovered and researched Chain of Thought (CoT), a core principle of large model reasoning capabilities, long before OpenAI.CoT significantly improves the model's performance in complex tasks by generating a series of intermediate reasoning tokens. Related research was presented at the International Conference on Learning Representations (ICLR). In addition, Google has published a paper on arXiv that explores optimizing computational efficiency when testing large language models. Despite Google's leading research on this principle, it has yet to launch a related product, while OpenAI has released a new model with powerful inference capabilities o1.(DeepTech)