These notebooks are the runnable end of the OLM Learning start-building path. They are intentionally ordered from beginner-friendly to scaling-focused.
| Notebook | What It Teaches | Open |
|---|---|---|
| First Language Model | Tiny local text, LM, Trainer, generation, save, and load. | Colab / GitHub |
| Train 125M On FineWeb-Edu | GPT-2-style 125M setup, FineWeb-Edu streaming, budget knobs, and a preflight training run. | Colab / GitHub |
| Custom Architecture Ablation | Swap attention, norm, and feed-forward choices while keeping the training loop fixed. | Colab / GitHub |
| AutoTrainer And Multi-GPU | Hardware inspection, AutoTrainer, and the single-node DDP/FSDP path. | Colab / GitHub |
Each notebook also includes its own Open in Colab badge.