Learn Pretrained Models
Discover, use, and fine-tune state-of-the-art AI models for computer vision, natural language processing, audio, and multi-modal tasks. From Hugging Face Hub to deployment in production.
What You'll Learn
A comprehensive guide to the world of pretrained AI models.
Model Discovery
Navigate Hugging Face Hub, find the right model for your task, and understand model cards and benchmarks.
Vision, Language & Audio
Explore models for image classification, text generation, speech recognition, and everything in between.
Using Models
Load pretrained models, run inference, batch process data, and serve models via APIs.
Fine-tuning
Adapt pretrained models to your data with LoRA, QLoRA, and the Hugging Face Trainer API.
Course Lessons
Follow the lessons in order or jump to any topic you need.
1. Introduction
What are pretrained models? Transfer learning, model types, where to find models, and common formats.
2. Hugging Face Hub
Browse, download, and use models from the largest model hub. Pipeline API and AutoModel classes.
3. Vision Models
ResNet, ViT, YOLO, SAM, Stable Diffusion — pretrained models for every computer vision task.
4. Language Models
GPT-2, Llama, BERT, T5, sentence-transformers — models for generation, classification, and embeddings.
5. Audio Models
Whisper, Bark, MusicGen — models for speech recognition, text-to-speech, and audio generation.
6. Multi-Modal Models
CLIP, LLaVA, Stable Diffusion, LayoutLM — models that work across text, image, video, and documents.
7. Using Models
Step-by-step guide to loading, inference, batch processing, GPU acceleration, and serving models.
8. Fine-tuning
Adapt models to your data with LoRA, QLoRA, Trainer API, and efficient fine-tuning techniques.
9. Best Practices
Model selection, licensing, ethical use, deployment strategies, and common mistakes to avoid.
Prerequisites
What you need before starting this course.
- Python programming experience
- Basic understanding of machine learning concepts
- Familiarity with pip or conda for package management
- A GPU is helpful but not required (Google Colab provides free GPUs)
Lilly Tech Systems