Ask HN: fast.ai 的“面向程序员的深度学习”课程在 2025 年还值得学习吗？
4 分•作者: hedgehog0•10 个月前
各位好，
我十多年前从吴恩达的 Coursera 课程上学习了一些基本的机器学习知识。最近，我从数学硕士项目毕业，手头有一些空闲时间，所以我想再次拾起机器学习/深度学习。
在 [Yacine 的视频](https://www.youtube.com/watch?v=ph6PIchDOcQ) 中，他提到了 [fast.ai 的课程](https://course.fast.ai/)，我以前听说过，但没有深入研究。 [这本书](https://www.amazon.com/Deep-Learning-Coders-fastai-PyTorch/dp/1492045527) 的目录看起来很扎实，但它出版于 2020 年，所以我想知道，考虑到人工智能发展的速度，您认为这本书或课程系列对于今天的学习者来说仍然是一个好的选择并且具有相关性吗？
*为了提供更多关于我的背景信息*：我本科是数学专业，辅修计算机科学（有 Python 基础），但从未上过任何机器学习/深度学习课程（除了那个 Coursera 课程），而且我刚刚完成了数学硕士项目，尽管我一直对图论、组合学和理论计算机科学有背景知识和兴趣。
我手头有两本书，分别是 Geron 的《Hands-on Machine Learning》和 Alammar 和 Grootendorst 的《Hands-on LLMs》，并且计划在拥有足够的背景知识后完成斯坦福大学的 CS224N 和 CS336 以及卡内基梅隆大学的 DL 系统课程。我对构建和改进智能系统很感兴趣，例如 DeepProver 和 AlphaProof，这些系统可用于改进数学证明/研究。
非常感谢！
查看原文
Dear all,<p>I learned some basic ML from Andrew Ng's Coursera course more than 10 years ago, recently I graduated from the Math Master program and have some free time in my hand, so I am thinking about picking up ML/DL again.<p>In [Yacine's video](https://www.youtube.com/watch?v=ph6PIchDOcQ), he mentioned [fast.ai's course](https://course.fast.ai/), which I heard of in the past but didn't look into too much. The table of contents of [the book](https://www.amazon.com/Deep-Learning-Coders-fastai-PyTorch/dp/1492045527) looks pretty solid, but it was published in 2020, so I was wondering given the pace of AI development, do you think this book or course series is still a good choice and relevant for today's learners?<p>*To provide more context about me*: I did math major and CS minor (with Python background) during undergrad but have never taken any ML/DL courses (other than that Coursera one), and I just finished the Master program in math, though I have background and always have interests in graph theory, combinatorics, and theoretical computer science.<p>I have two books "Hands-on Machine Learning" by Geron and "Hands-on LLMs" by Alammar and Grootendorst, and plan to finish Stanford's CS224N and CS336 and CMU's DL systems when I have enough background knowledges. I am interested in building and improving intelligent systems such as DeepProver and AlphaProof that can be used to improve math proof/research.<p>Thank you a lot!