Calendar

DateTopic / Deadlines
Tue 13 JanTopic 1: Running an LLM
LLM overview, open models, datasets
Presentation: Overview of LLMs
Paper: Small Language Models are the Future of Agentic AI.
Peter Belcak et al., NVIDEA Research, 2025. https://arxiv.org/pdf/2506.02153
▶️ Podcast generated by NoteGPT
Thu 15 Jan 
Tue 20 JanTopic 2: Agent Frameworks
LangGraph
Presentation: Agent Frameworks
Thu 22 Jan 
Tue 27 JanTopic 3: Agent Tool Use
Ollama, Claude, OpenAI tool format
Presentation: Tools
Class is online, see home page for Zoom details
Sign up with your work partner from last week for a breakout room here.
Paper: Toolformer: Language Models Can Teach Themselves to Use Tools
Timo Schick et al., 2023. https://arxiv.org/abs/2302.04761
▶️ Podcast generated by NotebookLLM by Google
Thu 29 JanClass is online, see home page for Zoom details
Breakout rooms are the same as for the previous class.
Tue 03 FebTopic 4: Exploring Tools
ToolNode, LangGraph's built-in tool library
Presentation: Exploring Tools
Thu 05 Feb 
Tue 10 FebTopic 5: Retrieval Augmented Generation
Vector databases, RAG pipeline, chunking
Presentation: RAG
Thu 12 FebTopic 6: Vision-Language Models (VLM)
CLIP, model alignment, image embeddings, other image tools
Presentation: vlm_part1, vlm_part2, vlm_part3
Tue 17 Feb 
Tue 24 FebCatch-Up Week, no class! Please use the time to finish your Portfolio!
I will be reading them over Spring Break!
Get your GitHub portfolio pages ready for your meeting with me!
Make sure the link to your GitHub repository is here
Sign up for your mid-term one-on-one review
Sign up for in-person slots
Sign up for Zoom slots
The dates for the reviews are:
Tue 10 March 3:30-4:30 in person
Wed 11 March 12:30-2:00 by Zoom
Thu 12 March 3:30-4:30 in person
Tue 17 March 3:30-4:30 in person
Wed 18 March 12:30-2:00 by Zoom
Thu 19 March 3:30-4:30 in person
Thu 26 FebCatch-Up Week, no class! Please use the time to finish your Portfolio!
I will be reading them over Spring Break!
Get your GitHub portfolio pages ready for your meeting with me!
Make sure the link to your GitHub repository is here
Tue 03 MarSpring Break
Thu 05 MarSpring Break
Tue 10 MarTopic 7: MCP and A2A
Model Context Protocol, tool discovery, tool use, Asta from AI2,
Agent-to-Agent protocol, toolagent cards, ngrok
Presentation: MCP and A2A
Run test ngrok on your computer before Thursday!
Thu 12 MarClass cancelled today

Sign up your 1 or 2 person team for your final project here, with your team name and the names
of the team members. Let me know if you need help finding a team partner!
Tue 17 MarTopic 7 Continued: A2A
Topic 8: LLM Fine-Tuning
Presentation: Fine-Tuning
Kinds of fine-tuning, QLORA, using Tinker
Install tinker and login to your account before class!
Thu 19 Mar 
Tue 24 MarTopic 9: Context Management
Compression, selection, persistence
No exercises, continue work on your final project!
Thu 26 MarTopic 10: Understanding Chain of Thought
CoT prompts, tokenized CoT, hidden-layer CoT
No exercises, continue work on your final project!
Tue 31 MarTopic 11: Coding Agents
Frontier coding agents, AlphaEvolve, EvoSkill, NVIDIA AVO
No exercises, continue work on your final project!
Thu 02 AprTopic 12: Personal Agents and Distilling
ClawdBot, OpenClaw, Null Claw, CoPaw, OpenClaw-RL
No exercises, continue work on your final project!
Tue 07 AprTopic 13: Agentic AI for Science
Orchestral AI, HEPTAPOD, AutoDiscovery
No exercises, continue work on your final project!
Thu 09 AprTopic 14: Agentic AI for Mathematics and Verification
Guest lecture: Prof. Kevin Sullivan
AlphaProof, AlphaDiscovery, Lean Co-Pilot, APOLLO, CLEVER
No exercises, continue work on your final project!
Tue 14 AprTopic 15: Visual Reasoning
Diffusion transformers, chain of frames, chain of steps
No exercises, continue work on your final project!
Thu 16 AprFinal project 5-minute videos due by 12 noon today in your GitHub portfolio.
Final Project Presentations
Tue 21 AprFinal Project Presentations
Thu 23 AprNo class - instructor out of town
Tue 28 AprFinal Project Presentations
Fri 15 MayReports due in GitHub archives - no extensions!
Markdown, pdf, or Word formats
Report should be self-contained - should not be necessary to read other files
Report sections:
Motivation - why you built it
Methods - how you built it
Evaluation - how well did it work
Conclusions - what did you learn
Evaluation can be either
Quantitative - compare your method and some other on benchmark sets
Qualitative - ask a few people to use and rate your system
Your GitHub archive should also contain a copy of your code and examples inputs/outputs