The author reflects on their nine-year journey in the field of deep learning, tracing the transition from Lua Torch to PyTorch, and the increasing complexity of tools like JAX. This reflection serves to highlight the rapid advancements in the field and the corresponding need for more sophisticated and user-friendly APIs.
Key takeaways:
- The article discusses the design of Machine Learning library API for the modern age of LLMs.
- It shares reflections on 9 years of experience in deep learning.
- The journey from Lua Torch's revival as PyTorch is discussed.
- The complexity of JAX is also addressed in the article.