guides

Fine-Tuning Guides

How to fine-tune an LLM without wasting a GPU run: method choices, datasets, cost and evaluation.

Building a Fine-Tuning Dataset That Actually Teaches Something JSONL formats by method, how many examples you really need, deduplication, synthetic data and the split that makes your evaluation mean something. TOFU Fine-Tuning LLMs: The Complete Guide Fine-tuning an LLM end to end: when it beats RAG and prompting, how much VRAM and money it costs, and the shortest path to a run that works first time. TOFU Fine-Tuning vs RAG: What Each One Actually Fixes RAG changes what the model can see; fine-tuning changes how it behaves. A decision framework with build and maintenance costs for both. TOFU