| Lecture 1 |
-- |
Topics: (no slides)
- Formal introduction
- Course details
- Syllabus
|
|
|
| Lecture 2 |
-- |
Topics: (slides)
- NLP overview + tasks classification, NER, QA, summarization; pipeline vs end-to-end.
|
|
|
| Lecture 3 |
-- |
Topics: (slides)
- Text preprocessing tokenization, normalization, stemming/lemmatization; pitfalls.
|
|
|
| Lecture 4 |
-- |
Topics: (slides)
- Language modeling basics n-grams, smoothing; perplexity; classical baselines.
|
|
|
| Lecture 5 |
-- |
Topics: (slides)
- Vector space models BoW/TF-IDF; cosine similarity; feature engineering.
|
|
|
| Lecture 6 |
-- |
Topics: (slides)
- Vector space models BoW/TF-IDF; cosine similarity; feature engineering.
|
|
|
| Lecture 7 |
-- |
Topics: (slides)
- Classical supervised NLP logistic regression/SVM; evaluation; error analysis.
|
|
|
| Lecture 8 |
-- |
Topics: (slides)
- Sequence labeling HMM/CRF intuition; BIO tagging for NER (conceptual + demo).
|
|
|
| Lecture 9 |
-- |
Topics: (slides)
- Word embeddings word2vec/GloVe intuition; subword (BPE); embedding algebra caveats
|
|
|
| Lecture 10 |
-- |
Topics: (slides)
- Word embeddings word2vec/GloVe intuition; subword (BPE); embedding algebra caveats
|
|
|
| Lecture 11 |
-- |
Topics: (slides)
- Neural NLP pre-transformer CNN/RNN/LSTM; attention idea; limitations.
|
|
|
| Lecture 12 |
-- |
Topics: (slides)
- Neural NLP pre-transformer CNN/RNN/LSTM; attention idea; limitations.
|
|
|
| Lecture 13 |
-- |
Topics: (slides)
- Neural NLP pre-transformer CNN/RNN/LSTM; attention idea; limitations.
|
|
|
| Lecture 14 |
-- |
Topics: (slides)
- Transformers I self-attention, positional encoding, layers; complexity; intuition.
|
|
|
| Lecture 15 |
-- |
Topics: (slides)
- Transformers I self-attention, positional encoding, layers; complexity; intuition.
|
|
|
| Lecture 16 |
-- |
Topics: (slides)
- Transformers II encoder vs decoder vs encoder-decoder; BERT vs GPT vs T5.
|
|
|
| Lecture 17 |
-- |
Topics: (slides)
- Transformers II encoder vs decoder vs encoder-decoder; BERT vs GPT vs T5.
|
|
|
| Lecture 18 |
-- |
Topics: (slides)
- Transformers II encoder vs decoder vs encoder-decoder; BERT vs GPT vs T5.
|
|
|
| Lecture 19 |
-- |
Topics: (slides)
- Pretraining objectives MLM, causal LM, span corruption; why pretraining helps.
|
|
|
| Lecture 20 |
-- |
Topics: (slides)
- Wrap-up before Mid Term Examination
|
|
|
|
-- |
-- |
(Feb 25) Last Date for Proposal Submission. |
|
|
-- |
-- |
Mid Semester Exam Week |
Best of Luck.
|
| Lecture 21 |
-- |
Topics: (slides)
- Pretraining objectives MLM, causal LM, span corruption; why pretraining helps.
|
|
|
| Lecture 22 |
-- |
Topics: (slides)
- Fine-tuning & transfer classification/NER/QA fine-tuning; hyperparameters; PEFT (LoRA).
|
|
|
| Lecture 23 |
-- |
Topics: (slides)
- Fine-tuning & transfer classification/NER/QA fine-tuning; hyperparameters; PEFT (LoRA).
|
|
|
| Lecture 24 |
-- |
Topics: (slides)
- Prompting & in-context learning prompt patterns, few-shot, chain-of-thought (practical use).
|
|
|
| Lecture 25 |
-- |
Topics: (slides)
- Prompting & in-context learning prompt patterns, few-shot, chain-of-thought (practical use).
|
|
|
| Lecture 26 |
-- |
Topics: (slides)
- Instruction tuning + alignment SFT, RLHF/DPO overview; what alignment changes.
|
|
|
| Lecture 27 |
-- |
Topics: (slides)
- Instruction tuning + alignment SFT, RLHF/DPO overview; what alignment changes.
|
|
|
| Lecture 28 |
-- |
Topics: (slides)
- Evaluation accuracy/F1/ROUGE/BLEU, perplexity; human eval; robustness checks.
|
|
|
| Lecture 29 |
-- |
Topics: (slides)
- Evaluation accuracy/F1/ROUGE/BLEU, perplexity; human eval; robustness checks.
|
|
|
| Lecture 30 |
-- |
Topics: (slides)
- Evaluation accuracy/F1/ROUGE/BLEU, perplexity; human eval; robustness checks.
|
|
|
| Lecture 31 |
-- |
Topics: (slides)
- Retrieval-Augmented Generation (RAG) embeddings, chunking, retrieval, reranking; grounding
|
|
|
| Lecture 32 |
-- |
Topics: (slides)
- Retrieval-Augmented Generation (RAG) embeddings, chunking, retrieval, reranking; grounding
|
|
|
|
-- |
-- |
(Mar 30) - Last Date of Mid Presentation. |
|
| Lecture 33 |
-- |
Topics: (slides)
- Retrieval-Augmented Generation (RAG) embeddings, chunking, retrieval, reranking; grounding
|
|
|
| Lecture 34 |
-- |
Topics: (slides)
- LLM systems context windows, caching, latency/cost; batching; tool calling overview.
|
|
|
| Lecture 35 |
-- |
Topics: (slides)
- LLM systems context windows, caching, latency/cost; batching; tool calling overview.
|
|
|
| Lecture 36 |
-- |
Topics: (slides)
- Hallucinations & reliability causes, mitigations; verification strategies; citations/attribution.
|
|
|
| Lecture 37 |
-- |
Topics: (slides)
- Hallucinations & reliability causes, mitigations; verification strategies; citations/attribution.
|
|
|
| Lecture 38 |
-- |
Topics: (slides)
- Safety/ethics bias, privacy, data leakage, prompt injection; safe deployment patterns.
|
|
|
| Lecture 39 |
-- |
Topics: (slides)
- Safety/ethics bias, privacy, data leakage, prompt injection; safe deployment patterns.
|
|
|
| Lecture 40 |
-- |
Topics: (slides)
- Capstone end-to-end mini-app (RAG QA bot / summarizer) + presentations.
|
|
|
| Lecture 41 |
-- |
Topics: (slides)
|
|
|
| Lecture 42 |
-- |
Topics: (slides)
|
|
|
|
-- |
-- |
(Apl 30) Last Date of Codes Submission. |
|
|
-- |
-- |
(Apl 30) Dead Line for Final Presentation Video Submission. |
|
|
-- |
-- |
(Apl 30) Dead Line for Report Submission. |
|
|
-- |
-- |
End Semester Exam Week |
Best of Luck.
|
|
Link Added on Last Date for Submission :
|
DA624
