A nice robot to help you forget that I messed up some things

Breaking changes

• The aspect bucket interval for SD3 is 64px, not 8px.
• The intermediary size for uncropped datasets was wrong for square images.
• --input_perturbation is gone, as it harmed models

Both of these issues will require removing:

• The vae cache contents (.pt files)
• The .json files from your data dir (aspect*.json)

You do not have to re-cache the text embeds.

Note: SD3 support is still experimental and it's recommended to avoid investing any large budgets into training runs until hyperparameter sweep can be conducted and ablation studies done.

What's Changed

• sdxl regression fixes by @bghira in https://github.com/bghira/SimpleTuner/pull/444
• sd3: clear T5 encoder out to save memory
• model card bugfixes by @bghira in https://github.com/bghira/SimpleTuner/pull/445
• sd3 timestep selection weighting changes by @bghira in https://github.com/bghira/SimpleTuner/pull/446
• remove --input_perturbation code
• fix tensor location error by @bghira in https://github.com/bghira/SimpleTuner/pull/448
• sd3: make sure we don't use the custom sdxl vae by default (whoops)
• nvidia: update dependencies by @bghira in https://github.com/bghira/SimpleTuner/pull/449
• sd3 vae default fix by @bghira in https://github.com/bghira/SimpleTuner/pull/450
• Unload LoRA weights before deleting pipeline object by @tolgacangoz in https://github.com/bghira/SimpleTuner/pull/451
• better error message for text embed failure
• fix for pipeline unload deleting pipeline before unloading the weights
• sd3: use 64px interval for bucketing instead of 8px by @bghira in https://github.com/bghira/SimpleTuner/pull/454

New Contributors

• @tolgacangoz made their first contribution in https://github.com/bghira/SimpleTuner/pull/451

Full Changelog: https://github.com/bghira/SimpleTuner/compare/v0.9.7...v0.9.7.1