AI ENGVisual Encyclopedia

WORLD 03  TRAINING AT SCALE

Watch information become a change in the model.

Pre-training across 6 comprehensive modules: from self-supervised objectives and web-scale data curation, to 3D parallelism, precision numerics, and run diagnostics.

JOURNEY · 6 MODULES · 21 SCENES

From random noise to a trained foundation model.

MODULE 01

Objectives & Paradigm

What a model is actually asked to do, and why next-token prediction won.

MODULE 02

Data at Web Scale

The pipeline that turns a multi-terabyte crawl into training tokens.

MODULE 03

The Pre-Train Recipe

Scaling laws, RoPE, normalization, attention variants — the architectural spec.

MODULE 04

Distributed Infrastructure

Hardware fabric, then the 3D parallelism that fills it.

MODULE 05

Precision & Acceleration

Numerics, memory-bound kernels, and the recompute trade.

MODULE 06

Execution & Diagnostics

Optimizers, schedules, initialization, telemetry, and surviving a crash.