# Train Llms Locally

> Runs workflows to train a small LLM from scratch on a laptop — tokenizer, data, architecture, training loop, and nanochat-style hands-on paths. Use when the user wants local pre-training, learns fundamentals, or says "train LLM from scratch", "nanochat", "tokenizer", "pretraining loop".

- Skill: `hiteshbandhu/train-llms-locally` (Agent Skill, multi-file: 4 files)
- Install (CLI): `npx skillmds@latest add hiteshbandhu/train-llms-locally`
- Raw SKILL.md: https://api.skillmd.com/api/skills/hiteshbandhu/train-llms-locally/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: hiteshbandhu (https://skillmd.com/u/hiteshbandhu)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/hiteshbandhu/train-llms-locally

---


# Train LLMs locally

Playbook from Angelos Perivolaropoulos (ElevenLabs workshop). [workflows.md](workflows.md).

```
├─ Environment (laptop vs Colab)              → A
├─ Size tokenizer + data to domain            → B
├─ Run training loop / checkpoints            → C
└─ Connect pretrain vs post-train path        → D
```

Source: `playlists/fundamentals-ai-engineer/`.