- summary
- Ollama Model Tuner: Locally fine-tune prompts, LoRAs, and models with Ollama for custom tasks.
- description
- Optimize Ollama models/prompts using local datasets, eval metrics, and iterative tuning. No cloud needed.
- triggers
- read_when
Ollama Model Tuner v1.0.0
🎯 Purpose
- Prompt engineering & A/B testing
- Modelfile customization
- LoRA fine-tuning with local data
- Performance benchmarking
🚀 Quick Start
!ollama-model-tuner --model llama3 --dataset ./data.json --task classificationFiles
scripts/tune.py: Python tuner with eval loopprompts/system.md: Base system prompts
Supported
Ollama 0.3+, Python 3.10+, datasets in JSONL/CSV.