
Already know some of this? A 2-minute test skips the lessons you know.
Take the placement testA free 31-lesson course built from Raultherockstar's 3h 52m of YouTube playlist. 3 days at 90 minutes a day, with study notes and a quiz in every lesson. A 2-minute placement test skips what you already know.
Taught by Raultherockstar on YouTube. Videos play in YouTube's own player, so the creator keeps every view.
Define artificial intelligence as a man-made system that processes information and distinguish it from natural intelligence.
Explain how neural networks mimic brain neurons, how deep learning adds layers, and differentiate CNN (images/video), DNN (text/speech), and RNN (sequential data) architectures.
Explain what LLMs are, how they are trained on massive datasets, and how the transformer architecture encodes and decodes information to generate text.
Learners will be able to explain the Chinese Room thought experiment and why current AI systems simulate understanding without genuine comprehension.
Learners will be able to evaluate the dangers of deepfake misuse and describe a real-world case where authorized deepfake technology helped small businesses.
The instructor addresses learner questions about the 30-day curriculum, confirming the goal of building a custom AI model via fine-tuning and RAG, and clarifies that all practical labs run in Google Colab so no local GPU is required.
Understand the environmental impact of cloud AI and the benefits of running models locally for privacy, offline access, and speed.
Install Ollama via terminal, pull models such as Llama 3, list downloaded models, and chat with them offline.
Install the Open WebUI package via pip and start the local server so that all Ollama models become accessible through a browser interface.
Distinguish fine-tuning (changing how a model talks) from retrieval-augmented generation (teaching a model new facts) so you can choose the right approach for your use case.
The learner will be able to locate existing datasets on Hugging Face instead of creating them from scratch.
The learner will be able to confirm that the dataset file appears in the Hugging Face repository and is ready for model training.
Understand how Google Colab provides free access to powerful GPUs for training AI models without needing local hardware.
Learn how to create datasets for fine-tuning, use Google Colab on mobile for training, and explore Replicate AI for image and video generation.
Understand how to structure dataset columns as instruction, input, and output, and use the ShareGPT conversion utility when the raw data uses different column names.
Learn how to install Ollama in Google Colab, prepare tokenizer and model files, name your model, and push it to the Ollama registry for local use.
Understand the key advantages of running models locally, including data privacy, zero recurring costs, offline availability, and full control over model training and customization.
Learn how to push a custom model to Ollama, then pull and run it locally from the command prompt.
Learn to prepare diverse personal photos, zip them, and fine-tune the Flux model via LoRA on Replicate using a trigger word and Hugging Face token.
Paste the token, set training steps, and start training so the model saves to your Hugging Face repository.
Play the 'Human or Not' game to practice distinguishing between human-written and AI-generated chat responses.
Learn how prompt engineering adds domain expertise to AI outputs and get specific tool recommendations for video generation, research, and thumbnail creation.
Learners will be able to write prompts, select aspect ratios and styles, and download finished images using the Ideogram web interface.
Learners will be able to clone their own voice, design a system prompt that defines the agent’s personality, and deploy a talking AI agent that can make phone calls.
Celebrate finishing the 30-day AI training challenge, learn about the upcoming 'AI Learner to AI Earner' course, and discover how to claim member-exclusive rewards.