peteromallet

/

Flux-Kontext-InScene

like 171

Thanks a lot for your research and dataset!

I'd like to know more details regarding your training if you are willing to share. From what I've gathered so far, I can see the following training details:

• trainer software: ostris/ai-toolkit
• steps: 3250
• epochs: 4
• dataset examples: 394 image pairs

I'm interested in figuring out the rest, like:

• LoRA/alpha rank
• Learning rate
• Batch size
• Learning rate scheduler
• Guidance scale
• The details of any failed runs you may have encountered and the things you've learned

Thank you again for your contribution, I hope to learn more from your experiences.