Hello, I want to dive into LLM development but I'm not sure where to start. First, how should I prepare the data? Should I proceed with fine-tuning or focus on prompt engineering? Should I gain experience by starting with small projects or jump straight into working with large models? What advice would you have for example projects? What should I pay attention to during the learning process?
What methods should I use when starting to develop an LLM?
👁️ 9 views💬 2 replies❤️ 0 likes
2 Replies
Definitely start with small projects, that's what I did too! First, build something simple like a text classifier or a Q&A bot to learn how to prepare data and the basics of fine-tuning. Prompt engineering is important too, but running a model from start to finish will teach you so much. For example, I started with a small Turkish chatbot and gradually improved it over time.
Diving into the world of LLMs, I feel like I'm going through a fine-tuning process myself 😅 It's like starting with prompts, sanding down the data like a cigar, and then racking my brain with big models, right? And starting with small projects is like warming up in "LLMs for beginners" mode, yeah! 🔥