AI-09 · Advanced · ENGLISH · PLANNED
Begin with a baseline, trace a model, measure errors and improve it.
Study time is an initial activity estimate, excluding extra reading, revision and extended laboratory or project work. Videos and question bodies have not been produced by adding this outline.
Complete these courses or demonstrate equivalent knowledge. These are recommended prerequisites, not enforced enrolment locks.
YouTube playlist: ARK Artificial Intelligence | Generative AI and Multimodal Models
Playlist title reserved in the plan; no YouTube playlist or upload has been created yet.
Sign in to save this learning pathPractical outcome: Compare small-model sample quality against a stated evaluation rubric.
Explain generative versus discriminative models using a visual and one worked real-world example. Context: Compare small-model sample quality against a stated evaluation rubric. End with a prediction or check-for-understanding.
Explain latent variables using a visual and one worked real-world example. Context: Compare small-model sample quality against a stated evaluation rubric. End with a prediction or check-for-understanding.
Explain autoencoders using a visual and one worked real-world example. Context: Compare small-model sample quality against a stated evaluation rubric. End with a prediction or check-for-understanding.
Explain variational autoencoders using a visual and one worked real-world example. Context: Compare small-model sample quality against a stated evaluation rubric. End with a prediction or check-for-understanding.
Use the first concepts from this section. Compare small-model sample quality against a stated evaluation rubric. Produce a labelled diagram, trace or calculation; explain each step and check one changed input.
3–6 questions on the preceding first-half concepts only: recall, visual interpretation, application and one plausible misconception. Explain each answer after submission.
Explain adversarial training using a visual and one worked real-world example. Context: Compare small-model sample quality against a stated evaluation rubric. End with a prediction or check-for-understanding.
Explain diffusion noise schedules using a visual and one worked real-world example. Context: Compare small-model sample quality against a stated evaluation rubric. End with a prediction or check-for-understanding.
Explain denoising objectives using a visual and one worked real-world example. Context: Compare small-model sample quality against a stated evaluation rubric. End with a prediction or check-for-understanding.
Explain conditional generation using a visual and one worked real-world example. Context: Compare small-model sample quality against a stated evaluation rubric. End with a prediction or check-for-understanding.
Compare small-model sample quality against a stated evaluation rubric. Work without the example, compare two cases, record assumptions and limitations, then self-review correctness, reasoning and clarity.
6–12 questions covering Generative versus discriminative models, Latent variables, Autoencoders and the remaining section concepts, with cumulative retrieval from earlier sections. Include a new transfer problem; 70% unlocks progression.
Practical outcome: Design an adaptation experiment with a strict USD budget.
Explain pretraining using a visual and one worked real-world example. Context: Design an adaptation experiment with a strict USD budget. End with a prediction or check-for-understanding.
Explain instruction tuning using a visual and one worked real-world example. Context: Design an adaptation experiment with a strict USD budget. End with a prediction or check-for-understanding.
Explain preference data using a visual and one worked real-world example. Context: Design an adaptation experiment with a strict USD budget. End with a prediction or check-for-understanding.
Explain parameter-efficient tuning using a visual and one worked real-world example. Context: Design an adaptation experiment with a strict USD budget. End with a prediction or check-for-understanding.
Use the first concepts from this section. Design an adaptation experiment with a strict USD budget. Produce a labelled diagram, trace or calculation; explain each step and check one changed input.
3–6 questions on the preceding first-half concepts only: recall, visual interpretation, application and one plausible misconception. Explain each answer after submission.
Explain low-rank adaptation using a visual and one worked real-world example. Context: Design an adaptation experiment with a strict USD budget. End with a prediction or check-for-understanding.
Explain context windows using a visual and one worked real-world example. Context: Design an adaptation experiment with a strict USD budget. End with a prediction or check-for-understanding.
Explain quantisation using a visual and one worked real-world example. Context: Design an adaptation experiment with a strict USD budget. End with a prediction or check-for-understanding.
Explain distillation using a visual and one worked real-world example. Context: Design an adaptation experiment with a strict USD budget. End with a prediction or check-for-understanding.
Design an adaptation experiment with a strict USD budget. Work without the example, compare two cases, record assumptions and limitations, then self-review correctness, reasoning and clarity.
6–12 questions covering Pretraining, Instruction tuning, Preference data and the remaining section concepts, with cumulative retrieval from earlier sections. Include a new transfer problem; 70% unlocks progression.
Practical outcome: Evaluate whether generated captions preserve supplied visual evidence.
Explain image-text embeddings using a visual and one worked real-world example. Context: Evaluate whether generated captions preserve supplied visual evidence. End with a prediction or check-for-understanding.
Explain visual question answering using a visual and one worked real-world example. Context: Evaluate whether generated captions preserve supplied visual evidence. End with a prediction or check-for-understanding.
Explain speech recognition using a visual and one worked real-world example. Context: Evaluate whether generated captions preserve supplied visual evidence. End with a prediction or check-for-understanding.
Explain speech synthesis concepts using a visual and one worked real-world example. Context: Evaluate whether generated captions preserve supplied visual evidence. End with a prediction or check-for-understanding.
Use the first concepts from this section. Evaluate whether generated captions preserve supplied visual evidence. Produce a labelled diagram, trace or calculation; explain each step and check one changed input.
3–6 questions on the preceding first-half concepts only: recall, visual interpretation, application and one plausible misconception. Explain each answer after submission.
Explain audio representations using a visual and one worked real-world example. Context: Evaluate whether generated captions preserve supplied visual evidence. End with a prediction or check-for-understanding.
Explain video representations using a visual and one worked real-world example. Context: Evaluate whether generated captions preserve supplied visual evidence. End with a prediction or check-for-understanding.
Explain multimodal evaluation using a visual and one worked real-world example. Context: Evaluate whether generated captions preserve supplied visual evidence. End with a prediction or check-for-understanding.
Explain provenance and rights using a visual and one worked real-world example. Context: Evaluate whether generated captions preserve supplied visual evidence. End with a prediction or check-for-understanding.
Evaluate whether generated captions preserve supplied visual evidence. Work without the example, compare two cases, record assumptions and limitations, then self-review correctness, reasoning and clarity.
6–12 questions covering Image-text embeddings, Visual question answering, Speech recognition and the remaining section concepts, with cumulative retrieval from earlier sections. Include a new transfer problem; 70% unlocks progression.
Combine all three sections: Evaluate whether generated captions preserve supplied visual evidence. Submit an original worked artifact, evidence or tests, assumptions and a reflection. Rubric: correctness, method, practical usefulness, reproducibility and communication. Self-reviewed, not certified.
6–12 balanced questions sampling every section of Generative AI and Multimodal Models, including interpretation, calculation or reasoning, a misconception and a new application. Passing is 70%; review errors before retrying.
Data experiments, evaluations, model tracing and reproducible projects. Guided work leads to an independent attempt, with a diagram, explanation, trace, calculation or tested artifact. Question bodies are generated when opened in your private learning path. Programming submissions are stored as text; the platform does not execute submitted code.
Use original or appropriately licensed datasets. Practical science activities use models or supplied data unless a qualified supervisor has reviewed a physical experiment.
Open the interactive practice sampler →