Coursework consists of four machine problems (MP0–MP3).
This page contains links to MP documents and their due dates.
MP documents will be linked here as they are released.
| Assignment | Description | Release | Due |
|---|---|---|---|
| Team-up | Form MP teams | Tue Aug 25 | Mon Sep 7 |
| MP0 | Prerequisites (PyTorch / CUDA) | Tue Aug 25 | Mon Sep 14 |
| MP1 | Pre-training | Tue Sep 8 | Mon Oct 5 |
| MP1 CP1 | LLM Architecture | Tue Sep 8 | Mon Sep 21 |
| MP1 CP2 | LLM Training Logic | Tue Sep 22 | Mon Sep 28 |
| MP1 CP3 | Pre-training & Evaluation | Tue Sep 29 | Mon Oct 5 |
| MP2 | Post-training | Tue Oct 6 | Mon Nov 2 |
| MP2 CP1 | Supervised Fine-Tuning (SFT) | Tue Oct 6 | Mon Oct 12 |
| MP2 CP2 | Reinforcement Learning with Verifiable Rewards (RLVR) | Tue Oct 13 | Mon Oct 19 |
| MP2 CP3 | Reinforcement Learning with Human Feedback (RLHF) | Tue Oct 20 | Mon Oct 26 |
| MP2 CP4 | Open-ended Extension | Tue Oct 27 | Mon Nov 2 |
| MP3 | Inference engine | Tue Nov 3 | Thu Dec 10 |
| MP3 CP1 | Inference engine basics | Tue Nov 3 | Mon Nov 9 |
| MP3 CP2 | PagedAttention, chunked prefill, prefix caching | Tue Nov 10 | Mon Nov 30 |
| MP3 CP3 | Inference parallelisms, speculative decoding, FlashAttention, CUDA graph | Thu Nov 19 | Thu Dec 10 |
MPs typically release Tuesday after class; they are due Monday AoE before class.