Small Language Models (SLMs) offer an efficient alternative to massive models, running locally or on limited hardware while maintaining high performance for specific tasks. This course teaches how to fine-tune an SLM using custom data. You will learn the mechanics of parameter-efficient fine-tuning, data formatting, and model evaluation. By the end of the course, you will have a fully functioning, task-specific language model ready for deployment.
Prerequisites Basic Python and PyTorch
Level:
Parameter-Efficient Fine-Tuning
Implement LoRA and QLoRA to train models under strict memory limits.
Dataset Preparation
Format and tokenize instruction datasets for supervised fine-tuning.
Model Evaluation
Measure text generation quality and detect overfitting during the training process.
Local Deployment
Merge adapter weights and serve the customized model using vLLM.