SYSTEM 09/14 · THE FIELD GUIDE
Fine-Tuning & Model Customization
Changing the weights — SFT, LoRA, QLoRA, RLHF, DPO, distillation, and when not to bother.
Fine-Tuning Fundamentals
What fine-tuning can (and can't) change, and how to prepare for it.
Fine-TuningBEGINNERPretraining vs Fine-TuningINTERMEDIATEFine-Tuning Capabilities & LimitsBEGINNERFine-Tune vs Prompt & RAGBEGINNEREvaluating the ResultINTERMEDIATECost BreakdownINTERMEDIATEDataset PreparationINTERMEDIATEFull Fine-Tuning vs PEFTINTERMEDIATEHyperparametersINTERMEDIATEInstruction TuningBEGINNERChat Templates & Loss MaskingINTERMEDIATESpotting OverfittingINTERMEDIATEUnslothINTERMEDIATEAxolotlINTERMEDIATELLaMA-FactoryINTERMEDIATEHugging Face PEFTINTERMEDIATEms-swiftINTERMEDIATE
LoRA & Efficient Methods
Parameter-efficient tuning that fits on a single GPU.
RLHF & Preference Training
How raw models learn what humans want: RLHF, DPO, reward models, GRPO.
Distillation & Training Tools
Smaller models from bigger ones, synthetic data, and the toolkits that run the job.
Model DistillationBEGINNERCreating Synthetic DataINTERMEDIATESynthetic Training DataBEGINNERCatastrophic ForgettingINTERMEDIATEContinued PretrainingINTERMEDIATEHosted Fine-TuningINTERMEDIATEFine-Tuning ToolkitsINTERMEDIATEGPU Memory & OOM FixesINTERMEDIATEDistilabelINTERMEDIATEArgillaINTERMEDIATENVIDIA NeMo CuratorADVANCEDDistillKitADVANCED