Using --enable_t23d with the gradio app breaks

[maunvz]

Using one of the example gradio commands from the readme:

python3 gradio_app.py --model_path tencent/Hunyuan3D-2mini --subfolder hunyuan3d-dit-v2-mini --texgen_model_path tencent/Hunyuan3D-2 --low_vram_mode

The web page shows the following warning:

Warning: Text to 3D is disable. To activate it, please run python gradio_app.py --enable_t23d.

Adding that flag to the command as follows:

python3 gradio_app.py --model_path tencent/Hunyuan3D-2mini --subfolder hunyuan3d-dit-v2-mini --texgen_model_path tencent/Hunyuan3D-2 --low_vram_mode --enable_t23d

Results in a new round of downloads, and a failure to load the models with the following error:

~/git/Hunyuan3D-2 (main ✔) python3 gradio_app.py --model_path tencent/Hunyuan3D-2mini --subfolder hunyuan3d-dit-v2-mini --texgen_model_path tencent/Hunyuan3D-2 --low_vram_mode --enable_t23d
Loading example img list ...
Loading example txt list ...
Loading example mv list ...
Fetching 13 files: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████| 13/13 [00:00<00:00, 49255.60it/s]
Fetching 20 files: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████| 20/20 [00:00<00:00, 61908.55it/s]
Loading pipeline components...: 100%|██████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:00<00:00, 10.16it/s]
Loading pipeline components...:  33%|██████████████████████████████                                                            | 2/6 [00:00<00:00,  4.52it/s]
An error occurred while trying to fetch /home/nvz/.cache/huggingface/hub/models--tencent--Hunyuan3D-2/snapshots/34e28261f71c32975727be8db0eace439a280f82/hunyuan3d-paint-v2-0-turbo/vae: Error no file named diffusion_pytorch_model.safetensors found in directory /home/nvz/.cache/huggingface/hub/models--tencent--Hunyuan3D-2/snapshots/34e28261f71c32975727be8db0eace439a280f82/hunyuan3d-paint-v2-0-turbo/vae.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
Loading pipeline components...: 100%|██████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:09<00:00,  1.50s/it]
Expected types for unet: (<class 'diffusers_modules.local.unet.modules.UNet2p5DConditionModel'>,), got <class 'diffusers_modules.local.modules.UNet2p5DConditionModel'>.
Loading checkpoint shards: 100%|███████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00,  3.54it/s]
Loading pipeline components...:  43%|██████████████████████████████████████▌                                                   | 3/7 [00:00<00:01,  3.53it/s]
Traceback (most recent call last):
  File "/home/nvz/git/Hunyuan3D-2/gradio_app.py", line 717, in <module>
    t2i_worker = HunyuanDiTPipeline('Tencent-Hunyuan/HunyuanDiT-v1.1-Diffusers-Distilled', device=args.device)
  File "/home/nvz/git/Hunyuan3D-2/hy3dgen/text2image.py", line 37, in __init__
    self.pipe = AutoPipelineForText2Image.from_pretrained(
                ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
        model_path,
        ^^^^^^^^^^^
    ...<2 lines>...
        pag_applied_layers=["blocks.(16|17|18|19)"]
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    ).to(device)
    ^
  File "/home/nvz/.local/lib/python3.13/site-packages/huggingface_hub/utils/_validators.py", line 114, in _inner_fn
    return fn(*args, **kwargs)
  File "/home/nvz/.local/lib/python3.13/site-packages/diffusers/pipelines/auto_pipeline.py", line 443, in from_pretrained
    return text_2_image_cls.from_pretrained(pretrained_model_or_path, **kwargs)
           ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/nvz/.local/lib/python3.13/site-packages/huggingface_hub/utils/_validators.py", line 114, in _inner_fn
    return fn(*args, **kwargs)
  File "/home/nvz/.local/lib/python3.13/site-packages/diffusers/pipelines/pipeline_utils.py", line 961, in from_pretrained
    loaded_sub_model = load_sub_model(
        library_name=library_name,
    ...<20 lines>...
        provider_options=provider_options,
    )
  File "/home/nvz/.local/lib/python3.13/site-packages/diffusers/pipelines/pipeline_loading_utils.py", line 709, in load_sub_model
    raise ValueError(
    ...<2 lines>...
    )
ValueError: The component <class 'transformers.models.t5.tokenization_t5._LazyModule.__getattr__.<locals>.Placeholder'> of <class 'diffusers.pipelines.pag.pipeline_pag_hunyuandit.HunyuanDiTPAGPipeline'> cannot be loaded as it does not seem to have any of the loading methods defined in {'ModelMixin': ['save_pretrained', 'from_pretrained'], 'SchedulerMixin': ['save_pretrained', 'from_pretrained'], 'DiffusionPipeline': ['save_pretrained', 'from_pretrained'], 'OnnxRuntimeModel': ['save_pretrained', 'from_pretrained'], 'PreTrainedTokenizer': ['save_pretrained', 'from_pretrained'], 'PreTrainedTokenizerFast': ['save_pretrained', 'from_pretrained'], 'PreTrainedModel': ['save_pretrained', 'from_pretrained'], 'FeatureExtractionMixin': ['save_pretrained', 'from_pretrained'], 'ProcessorMixin': ['save_pretrained', 'from_pretrained'], 'ImageProcessingMixin': ['save_pretrained', 'from_pretrained'], 'ORTModule': ['save_pretrained', 'from_pretrained']}.

This occurs on the latest main branch, commit 2ee4770

[uanu2002]

Hi @maunvz , I ran into the same issue. pip install sentencepiece works for me.