Train a product LoRA

Fine-tune a LoRA on a specific product (packaging, shape, logo placement) so it can be generated in new scenes and backgrounds without a full photoshoot.

Run this job Get help running this

Estimated $2–$8 per training run · 15–40 min

What inputs you need

15–30 product images from multiple angles
Product-focused captions
Training config

What Badgr returns

Trained product LoRA
Training loss curve
Sample generations in varied scenes

Recommended GPU routes

A100 40GB A6000 48GB A100 80GB

Estimated cost: $2–$8 per training run · Estimated runtime: 15–40 min

Example command

badgr run "python train_lora.py --config product_lora.yaml" \
  --gpu A100_40GB \
  --max-cost 8 \
  --max-runtime 50

Common setup failures Badgr avoids

Logo/label text distortion in generated scenes — dataset should include close-up label shots; training config flags datasets missing them.

Dataset upload times out on job start — pre-stage images to object storage and reference a URL instead of uploading inline.

Ready to run this?

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

Run this job Back to LoRA Training