This project (fine-tune.app) provides a full workflow for end-to-end fine-tuning of LLMs using user-uploaded data. The web UI and backend allow users to upload TXT, PDF, and DOCX files. The system automatically converts these files into high-quality JSONL training datasets, fine-tunes the Gemma 270M model, converts the result to GGUF, and deploys it to Ollama. The output model can then run fully locally, even on smaller devices and without an internet connection. This is particularly useful in domain-specific environments, privacy-sensitive settings, and cases where external APIs cannot be used.
Team
Placement
Finalist
