Raw Web
Stage 0: Mission0% reached · replay stage · next: CollectPipeline mapStarted pipeline
1Raw Web
2Clean Text
3Tokens
4Base Model
5Assistant
6Aligned Model
7API
Mission
Operate the LLM pipeline
Move one artifact from raw web text to a portfolio-ready assistant model card.
Simulation disclaimer
This is a simplified educational simulation. It does not crawl the web, train a real model, run reinforcement learning, or serve a real API.
BriefingStage 0
Welcome to LLM Factory
In this simulation, you will build a simplified ChatGPT-like assistant.
You will operate the major steps: collect text, clean it, tokenize it, pretrain a base model, convert it into an assistant, align it, evaluate it, and serve it through an API.
LLMs are built through a sequence of data, training, post-training, evaluation, and deployment steps.
Public framing
This is a simplified educational simulation. It does not crawl the web, train a real model, run reinforcement learning, or serve a real API.