◉ Checking local model...
Inference Engine: Ollama
Add Model · Ollama
Downloads from the internet into Ollama's local model store. Check available disk space and model licensing before installing. Model files may be several GB.
Model installation is optional.
Advanced: Train a model (optional)
For advanced users only. This requires a separate local training setup, a compatible model, and a JSONL dataset. It is not needed to use the agents.
Compatible model directory
Place each compatible local model folder here.
Loading local path…
Training dataset directory
Place each validated .jsonl training dataset here.
Loading local path…
Checking optional training environment.
To install the optional training environment, open your extracted Lab folder and run setup-training-windows.bat once. When it finishes, close and restart the Lab.
Which technology should I use?
Ollama: local inference. PyTorch: numerical training. Transformers: compatible local pretrained models. PEFT / LoRA: efficient adapters. TRL: supervised fine-tuning. CUDA: supported NVIDIA GPU acceleration. Training changes model parameters; agent exercises and evaluation do not.