Can not load fusionx_lora into original wan2.1-14b

hello, i am adding the fusionx_lora into original wan2.1-14b-i2v, my code is as follow:

> pipe = WanImageToVideoPipeline.from_pretrained(my_local_path + "Wan2.1-I2V-14B-480P-Diffusers", vae=vae, image_encoder=image_encoder, torch_dtype=torch.bfloat16)
> pipe.load_lora_weights(
> my_local_path + "Wan14BT2VFusioniX/FusionX_LoRa/Wan2.1_I2V_14B_FusionX_LoRA.safetensors"
> )

But i got some errors:

> File "/mmu_mllm_hdd_2/zuofei/infer_test/lora_infer_multi.py", line 60, in process_image
> pipe.load_lora_weights(
> File "/hetu_group/zuofei/env/wan_infer/lib/python3.12/site-packages/diffusers/loaders/lora_pipeline.py", line 4869, in load_lora_weights
> state_dict = self.lora_state_dict(pretrained_model_name_or_path_or_dict, **kwargs)
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> File "/hetu_group/zuofei/env/wan_infer/lib/python3.12/site-packages/huggingface_hub/utils/_validators.py", line 114, in _inner_fn
> return fn(*args, **kwargs)
> ^^^^^^^^^^^^^^^^^^^
> File "/hetu_group/zuofei/env/wan_infer/lib/python3.12/site-packages/diffusers/loaders/lora_pipeline.py", line 4796, in lora_state_dict
> state_dict = _convert_non_diffusers_wan_lora_to_diffusers(state_dict)
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> File "/hetu_group/zuofei/env/wan_infer/lib/python3.12/site-packages/diffusers/loaders/lora_conversion_utils.py", line 1564, in _convert_non_diffusers_wan_lora_to_diffusers
> num_blocks = len({k.split("blocks.")[1].split(".")[0] for k in original_state_dict})
> ~~~~~~~~~~~~~~~~~~^^^
> IndexError: list index out of range

Can you tell me how to fix it? Thank you so much!

Beginne in diffusers/loaders/lora_pipeline.py bei lora_state_dict und folge _convert_non_diffusers_wan_lora_to_diffusers in diffusers/loaders/lora_conversion_utils.py. Reproduziere den Fehler mit dem WanImageToVideoPipeline-Code und untersuche die State-Dict-Schlüssel von FusionX_LoRA; fertig ist es, wenn der Checkpoint ohne den IndexError geladen wird oder seine Inkompatibilität eindeutig dokumentiert ist.

Vom Indexierungsmodell aus dem Issue-Text verfasst.