Fine-tuning
Adapting models to your data and tasks.
- What is Fine-tuning?Intermediate
Fine-tuning continues training a finished model on your examples so new behavior — tone, format, skill — lives in the weights.
- Full Fine-tuning vs PEFTAdvanced
Full fine-tuning updates every weight; PEFT freezes the base and trains a small adapter you can swap, merge, or stack.
- LoRA and QLoRAAdvanced
LoRA trains two small matrices instead of the frozen weight W; QLoRA also stores W in 4-bit so a 7B+ model fine-tunes on one GPU.
- What is Quantization?Intermediate
Quantization stores weights with fewer bits so models fit in memory — 4-bit is common for serving and for QLoRA training, with a small quality trade.
- Instruction Tuning vs Preference TuningAdvanced
Instruction tuning (SFT) shows a gold reply; preference tuning (usually DPO) shows a winner and a loser — SFT first, then DPO is the 2026 default, not PPO.
- When to Fine-tune vs RAG vs PromptingIntermediate
Try prompting, then RAG for facts, then fine-tune for stubborn behavior — the 2026 default is QLoRA SFT (and DPO) plus retrieval, not weights as a wiki.