Can an LLM create something genuinely new, or does it only remix patterns from its training data?
中文摘要
这篇文章学术探讨了大语言模型是真正具有创造力，还是仅在重新组合训练数据，并结合了梯度下降、模糊逻辑与“无为”等概念。
English Summary
This academic exploration examines whether LLMs are genuinely creative or merely remix training data, incorporating concepts like gradient descent, fuzzy logic, and Wu Wei.
Original Excerpt
An academic exploration of LLMs, gradient descent, Mean Squared Error, fuzzy logic, Wu Wei and creativity. Continue reading on Medium »