01Unread
Fine-Tuning Without Touching the Weights
What a LoRA adapter actually is, why it is so much smaller than the model it changes, and how to work out its size before you train.
Start here →Standalone essays for a senior developer moving into ML and AI engineering
The Program Is Now a Model
1 standalone essay. Each explains one mechanism with a worked example you can recompute, and closes with an exercise you can do without a GPU, a paid API, or the source.
1 essays published
The shelf
Reading order
Each essay is standalone; read in any order.
Part I
The program is now a set of weights
No essays published yet.
Part II
Data and training
Part III
Evaluation and monitoring
No essays published yet.
Part IV
Serving and inference
No essays published yet.
Part V
LLM systems: retrieval, fine-tuning, agents
No essays published yet.
Part VI
The job
No essays published yet.