Interactive Tool
What Model Size Do I Need?
Answer 4 questions. Get a specific recommendation for model family, quantization, and expected performance on your target hardware.
1. What's your task?
What will the model do?
2. Where will it run?
Target deployment hardware.
3. How fast does it need to be?
Acceptable inference latency.
4. How much training data do you have?
Number of labeled examples or instruction pairs.
Answer all 4 questions above to see your recommendation.
Fine-Tune the Right Model, First Try
Ertas handles model selection, training configuration, and GGUF export. Upload your data, pick a model, and ship.