LoRA Training

Train a person LoRA

Fine-tune a LoRA on a specific person's likeness from reference photos, for consistent portraits and scenes — subject to Badgr's usage policy for likeness training.

Run this jobGet help running this

Estimated $2–$9 per training run · 15–45 min

What inputs you need

  • 15–30 reference photos of the person (varied lighting/angles)
  • Captions
  • Training config

What Badgr returns

  • Trained person LoRA
  • Training loss curve
  • Sample portrait generations

Recommended GPU routes

A100 40GBA6000 48GBA100 80GB

Estimated cost: $2–$9 per training run · Estimated runtime: 15–45 min

Example command

badgr run "python train_lora.py --config person_lora.yaml" \
  --gpu A100_40GB \
  --max-cost 9 \
  --max-runtime 60

Common setup failures Badgr avoids

Likeness drift across lighting conditions — dataset should span varied lighting; training config warns if the input set is too uniform.

Training job exceeds max-cost before convergence — lower batch size or rank, or split into checkpointed resumable runs.

Ready to run this?

Launch from the dashboard, CLI, or Compute API. Max-cost protection included.

Run this jobBack to LoRA Training