| language: | |
| - en | |
| tags: | |
| - sdxl | |
| - lora | |
| - kaggle | |
| - workflow | |
| library_name: diffusers | |
| pipeline_tag: text-to-image | |
| license: mit | |
| # mihai-lora-workflow (workflow-only) | |
| This repository contains the reproducible training workflow used to run a personal SDXL LoRA pipeline on free Kaggle GPU, including checkpoint continuation, checkpoint evaluation, and LinkedIn-style gallery generation. | |
| No personal training images, captions, generated portraits, or model checkpoints are included. | |
| ## Included | |
| - Kaggle training script (`07_kaggle/train_flux_lora.py`) | |
| - Config/run automation scripts (`03_configs/*`) | |
| - Evaluation script templates (`08_kaggle_eval/*`) | |
| - Runbook and process notes | |
| ## Excluded | |
| - Raw/curated personal photos | |
| - Captions tied to personal data | |
| - Checkpoints and model weights | |
| - Generated output galleries | |
| - API tokens and credentials | |
| ## Privacy note | |
| If you publish similar work, keep biometric data and personal LoRA weights private unless you explicitly want public distribution. | |