Jefferson Building, Room 450, 17 Oxford Street, Cambridge, MA 02138
Instructors
Michelle Audirac and Mauricio Tec
Learning Outcomes
- Download a dataset from Huggingface
- Apply parameter-efficient finetuning methods
- Use inference ollama/vllm to run LLMs efficiently locally.
Use Cases
- Specialize LLMs for your own data and tasks.
- Learn data and model management best practices under FAIR principles.
- Choose the right LLM for the task.
- Use LLMs on-device with privacy-sensitive research data.
Prerequisites
- Python programming with familiarity in Pytorch
- Basic understanding of neural networks and the training process
- GPU access (e.g., free-tier Google Co-lab)
- HuggingFace account and API key