---
title: "Scaling laws, data, and synthetic data"
lesson_id: "04"
---

# Scaling laws, data, and synthetic data

- **Lesson ID:** 04
- **Goal:** Data is the silent bottleneck. Use scaling-law intuition and generate synthetic instructions without lying to yourself about quality.
- **Human lesson:** [04-scaling-laws-and-synthetic-data.html](04-scaling-laws-and-synthetic-data.html)
- **Phase:** Phase 2 Model Internals

## Prerequisites

- The previous week's artifact, or a written note if this is week 1.
- Python 3.11+ and a laptop. CUDA is useful from week 5 onward and not required to read.

## Inputs, outputs, and artifacts

- **Inputs:** The replica from the prior week.
- **Outputs:** Stand up the data folder that weeks 5 and 7 will consume.
- **Artifacts:** Lesson notes plus the code named in the human page.

## Agent build steps

1. State the week 1 principle: you do not need to pretrain an 800B model to understand the shape of the problem; you do need progressively realistic replicas.
2. Teach the topics on the human page without inventing paper URLs or news claims.
3. Keep starter code in fenced blocks that match the human lesson.
4. End with the assignment and the opinion checkpoint.
5. Use Edge FDE tone: ship, do not sightseeing. Opinion checkpoints matter.

## Constraints

Plain spoken English. No quizzes. No em dashes. No invented metrics, dates, or citations. Brand Edge FDE only. Name well-known papers by title only: Attention Is All You Need, InstructGPT, Direct Preference Optimization, ZeRO, PagedAttention/vLLM, FlashAttention, Scaling Laws for Neural Language Models.

## Key concepts

- Loss tracks scale only if you count tokens and compute honestly.
- Synthetic rows need a schema, a filter, and a source tag.
- Packing and contamination change the story your curve tells.
- A token budget is arithmetic you write before you launch.

## Takeaways

- Ship typed JSONL and a reject count.
- Keep a human slice you are willing to read aloud.
- Do not invent a scaling exponent. Log what you ran.

## Acceptance checks

- Human HTML has Key concepts and Takeaways.
- Opinion checkpoint is present.
- No em dash and no fake URL.
- [ ] Proceed to [lesson 05 brief](05-pretrain-a-small-lm.llms.md).