Department of Computer Science and Engineering
Learn Transformers, GPT, Large Language Models, Prompt Engineering, Retrieval-Augmented Generation (RAG), Agentic AI, Vision Language Models, and Responsible AI.
AD470
3-0-3
Dr. Alapan Kuila
NLP Fundamentals, Language Models, Deep Learning, PyTorch, Word2Vec, GloVe, FastText and Tokenization.
RNN, LSTM, GRU, Seq2Seq, Attention, Beam Search and Decoding Strategies.
Self Attention, BERT, GPT, T5, Llama, Mistral, Gemini, Hugging Face.
Prompt Engineering, Instruction Tuning, RLHF, DPO, LoRA, QLoRA, Planning and Reasoning.
Retrieval Augmented Generation, Vector Databases, GraphRAG, Knowledge Graphs, Tool Calling, AI Agents.
CLIP, BLIP, LLaVA, Qwen-VL, Diffusion Models, AI Safety, Ethics, Efficient Inference.
Introduction to Generative AI & Natural Language Processing
Language Models and Probability Foundations
Deep Learning Fundamentals using PyTorch
RNN, LSTM, GRU and Sequence Models
Attention Mechanism and Sequence-to-Sequence Models
Transformer Architecture & Self Attention
BERT, GPT, T5, Llama and Foundation Models
Prompt Engineering, LoRA and QLoRA
RLHF, Alignment and Instruction Tuning
Retrieval Augmented Generation (RAG)
Knowledge Graphs and Agentic AI
Vision Language Models & Diffusion Models
Evaluation, Hallucinations, AI Safety & Ethics
Semester Project Presentation & Course Revision
Introduction to Generative AI
Main written examination covering the complete syllabus.
Complete implementation, experiments and final demonstration.
Evaluation of the first half of the course.
Programming assignments and laboratory exercises.
Presentation of a recent research paper in Generative AI.
Literature Survey
Dataset Collection
Implementation
Initial Results
Experiments
Evaluation
GitHub
Report
Poster
Demo
Tong Xiao & Jingbo Zhu
Jurafsky & Martin
Jacob Eisenstein
ACL • EMNLP • NeurIPS • ICML • ICLR
Technical discussions, knowledge sharing, and brainstorming with classmates are encouraged. However, every submission must reflect your own understanding and your team's original work.
Copying code, reports, figures, experimental results, or text from another student, group, online source, or AI system without proper acknowledgement constitutes academic misconduct.
Any collaboration with another student, group, or external source must be explicitly acknowledged. Violations may result in negative marking, course deregistration, and disciplinary action as per institute policy.
LLMs such as ChatGPT, Claude, Gemini, and Copilot are treated as learning companions. They may be used to understand concepts, learn syntax, debug code, brainstorm ideas, or improve technical writing.
Students must not copy AI-generated code or text into assignments or projects without understanding and proper disclosure.
Every submission must include an AI Usage Statement describing how AI tools were used. Failure to disclose AI assistance or inability to explain submitted work during evaluation may be treated as academic misconduct.