ai-agents-for-beginners vs llms-from-scratch
Two of the top projects in Learning and cookbooks, side by side: who leads on each score part and who each one suits, then setup, license, activity and what each review found.
ai-agents-for-beginners
18-lesson course on building AI agents with Microsoft Agent Framework
llms-from-scratch
Code for the book on building a GPT-style LLM in PyTorch
Verdict
ai-agents-for-beginners ranks higher, 77 to 76. Of the 4 score parts we measure for both, ai-agents-for-beginners leads on 2, llms-from-scratch on 1 and 1 is even.
- Adoptionllms-from-scratch leads by 8 points, 96 to 88.
- FreshnessEven at 100.
- Maintenanceai-agents-for-beginners leads by 2 points, 60 to 58: close to even.
- Agent-readyai-agents-for-beginners leads by 30 points, 30 to 0: a wide gap.
Pick ai-agents-for-beginners if you are
Developers new to building AI agents, comfortable with Python
- Each lesson is standalone, so you can start at any topic
- Covers design patterns plus MCP, A2A, NLWeb, memory and deployment
Pick llms-from-scratch if you are
Python developers learning how LLMs work by coding one in PyTorch
- Main chapter code runs on a conventional laptop and uses a GPU automatically if present
- Plain PyTorch with no external LLM libraries, so every component is readable
| What we compare | ai-agents-for-beginners | llms-from-scratch |
|---|---|---|
| Score parts, out of 100 | ||
| Adoption | 88, widely used | 96, widely used |
| Freshness | 100, active | 100, active |
| Maintenance | 60, fair | 58, fair |
| Easy to run | Not scored for this kind | Not scored for this kind |
| Agent-ready | 30, minimal | 0, none |
| Facts from GitHub and the README | ||
| Stars | 77.4k | 106.8k |
| License | MIT (permissive) | custom license (read the license) |
| Last commit | Oct 2026 | Oct 2026 |
| Last release | None published | None published |
| Language | Jupyter Notebook | Jupyter Notebook |
| Docker | No | No |
| GPU | Not needed | Optional |
| arm64 or Apple Silicon | Not stated | Not stated |
ai-agents-for-beginners
A Microsoft course of 18 lessons on building AI agents, each with a written README, a short video for most lessons, and Python code samples in Jupyter notebooks. Topics include tool use, agentic RAG, planning, multi-agent patterns, memory, MCP/A2A/NLWeb protocols, computer-use agents, local agents and securing agents. Samples use Microsoft Agent Framework with Microsoft Foundry Agent Service V2, and some also support OpenAI-compatible providers such as MiniMax.
Who it is for: Developers new to building AI agents, comfortable with Python
Strengths
- Each lesson is standalone, so you can start at any topic
- Covers design patterns plus MCP, A2A, NLWeb, memory and deployment
- Text, short videos and runnable Python samples per lesson
- 50+ language translations kept current by GitHub Action
Weaknesses
- Samples are built around Microsoft Agent Framework and Foundry, which need an Azure account
- Some later lessons have no video
- Cloning pulls large translation folders unless sparse checkout is used
- Alternative provider support is only partial, per the README
- no GPU
- Needs Microsoft Foundry (Azure account), Microsoft Agent Framework, Microsoft Foundry Agent Service V2
- Models: Microsoft Foundry models, OpenAI-compatible providers (MiniMax)
llms-from-scratch
Official code for the Manning book "Build a Large Language Model (From Scratch)". Notebooks and scripts implement tokenization, attention, a GPT model, pretraining, classification finetuning, instruction finetuning and LoRA in plain PyTorch, without external LLM libraries. Bonus folders add from-scratch Llama 3.2, Qwen3, Gemma 3, Olmo 3, KV cache, MoE and DPO.
Who it is for: Python developers learning how LLMs work by coding one in PyTorch
Strengths
- Main chapter code runs on a conventional laptop and uses a GPU automatically if present
- Plain PyTorch with no external LLM libraries, so every component is readable
- Bonus material covers KV cache, GQA, MoE, DPO and several model architectures
- Exercise solutions, a companion video course and CI tests on Linux, Windows and macOS
Weaknesses
- Tied to a print book; contributions extending main chapter code are not accepted
- Produces a small educational model, not a production-grade LLM
- License is reported as NOASSERTION; the README states no license terms
- Video course and quiz PDF are hosted by the publisher, not in the repo
- GPU optional
- Needs PyTorch, Jupyter, Ollama (optional, ch07 evaluation)
- Models: GPT-2, Llama 3.2, Qwen3, Gemma 3, Olmo 3
More in Learning and cookbooks
- ai-engineering-from-scratch vs generative-ai-for-beginners
- ai-engineering-from-scratch vs ai-agents-for-beginners
- ai-engineering-from-scratch vs llms-from-scratch
- generative-ai-for-beginners vs ai-agents-for-beginners
- generative-ai-for-beginners vs llms-from-scratch
- The full Learning and cookbooks ranking with reviews
Building with these?
We help teams pick an open-source AI stack and run it in production. Client work never changes a rank.