Video course · 12 chapters · 209 min · certificate
Train Your Own LLM – Tutorial
Train a language model from scratch that chats like you, using your WhatsApp or Telegram history: data extraction, tokenisation, the transformer, pre-training, fine-tuning datasets, instruction fine-tuning, LoRA, scaling up and a conversational-format bonus.
What you'll learn
- Extract and process chat data
- Build a tokenizer
- Implement a transformer
- Pre-train a small LLM
- Instruction fine-tune and use LoRA
- Scale training to larger data
Chapters
12 chapters · 209:11-
3:03
01About Members
About the course
Mimic a style.
-
4:21
02Intro Members
Introduction
The pipeline.
-
8:09
03Data Members
Training data
Export and process.
-
13:27
04Tokens Members
Tokenization
Text encoding.
-
23:21
05Transformer Members
The transformer architecture
Build it.
-
32:25
06Pre-training Members
Pre-training
Learn language.
-
8:19
07Dataset Members
Fine-tuning dataset
Form the data.
-
33:12
08Instruct Members
Instruction fine-tuning
Compare approaches.
-
14:22
09LoRA Members
Fine-tuning with LoRA
Adapters.
-
49:01
10Scale Members
Scaling everything
Large dataset.
-
17:30
11Bonus Members
Bonus: conversational format
Try a new format.
-
2:01
12Wrap-up Members
Conclusion
Your own model.
Jobs this course suits
Our AI checked this course against 500 jobs; these get the most out of it. Each job links to its learning path.