How to Fine-Tune or Create my Own Model from Existing or Scratch ?

[codezeros]

Hi i would also like to fine-tune and train my own LLMs is there any guidelines for that with either from any existing model like llama or any other model or from scratch and also for stable diffusion models with fully base model or as loras on locally using GPU? with own images and datas in .safetensors or onnx or .pickle tensors like file formats