AI Training

Your local model lab — fine-tune LLMs, diffusion, voices and video LoRAs on your own GPU, all offline.

• Fine-tune LLMs with QLoRA/DoRA and six alignment methods (SFT, DPO, GRPO, ORPO, SimPO, KTO) — or chain them SFT→DPO→merge→eval in one pipeline
• Train LoRAs for Stable Diffusion / SDXL / FLUX images, XTTS voices, VLM captioners, and video/temporal motion — one workbench, every modality
• Browse models across HuggingFace, CivitAI and Ollama, watch live loss curves and GPU load, then run Optuna sweeps and LLM-as-judge scoring
• Merge adapters and export to GGUF or model-dock — everything runs on your own machine; your data never leaves your disk