Error When Using `STOIQONewRealityFLUXSD_F1DAlpha.safetensors`: Empty Sequence in `analyze_checkpoint_state`

Description:

Hello,

I'm encountering an issue when trying to train a model using the STOIQONewRealityFLUXSD_F1DAlpha.safetensors file from the TheImposterImposters/STOIQONewRealityFLUXSDXLLightning-F.1DAlpha repository.

Steps to Reproduce:

Configuring the model with the following:


newReality:
   repo: TheImposterImposters/STOIQONewRealityFLUXSDXLLightning-F.1DAlpha
   base: TheImposterImposters/STOIQONewRealityFLUXSDXLLightning-F.1DAlpha
   license: other
   file: STOIQONewrealityFLUXSD_F1DAlpha.safetensors

The model is successfully downloaded, but when running the training script, I encounter the following error:

ValueError: max() arg is an empty sequence

Specifically, this occurs in the flux_utils.analyze_checkpoint_state function at line 81.

[INFO] _, is_schnell, _, _ = flux_utils.analyze_checkpoint_state(args.pretrained_model_name_or_path)
[INFO] File "C:\Users\Gabriel\PhpstormProjects\fluxgym\sd-scripts\library\flux_utils.py", line 81, in analyze_checkpoint_state
[INFO] max_double_block_index = max(
[INFO] ValueError: max() arg is an empty sequence

Observations:

It seems that analyze_checkpoint_state is attempting to calculate the max() value from an empty sequence, potentially due to the model checkpoint file being improperly loaded or missing certain data.
When inspecting the state dict of the downloaded model, it seems to be missing certain keys, or the file format may not be compatible.

Questions:

What changes or configuration adjustments should I make to resolve this issue? Is this a known issue with the model checkpoint format, or is the model missing some necessary metadata?
Is the STOIQONewRealityFLUXSD_F1DAlpha.safetensors file fully compatible with flux_train_network.py, or does it require additional steps to load properly?
Could you provide guidance on how to debug this further, especially regarding the empty sequence in analyze_checkpoint_state?

I’ve also checked the paths and verified that all required files (clip_l.safetensors, t5xxl_fp16.safetensors, and ae.sft) are correctly set up.

Any help would be greatly appreciated. Thank you in advance!

cocktailpeanut / fluxgym