timestep issue in svd

[bilalmashooq]

(venv) PS C:\Users\Students\PycharmProjects\pythonProject> python examples/svd/image_to_video.py --mode=1 --SVD.config=examples/svd/configs/svd.yaml --SVD.checkpoint=checkpoints/svd-d19a808f.ckpt --SVD.num_frames=5 --SVD.fps=15 --image=download.jpeg --SVD.decode_chunk_size=1 flash attention is available. Image size is not divisible by 64 (300x168). Resizing it to (256x128)! Initialized embedder #0: FrozenOpenCLIPImagePredictionEmbedder with 683800065 params. Trainable: False Initialized embedder #1: ConcatTimestepEmbedderND with 0 params. Trainable: False Initialized embedder #2: ConcatTimestepEmbedderND with 0 params. Trainable: False Initialized embedder #3: VideoPredictionEmbedderWithEncoder with 83653863 params. Trainable: False Initialized embedder #4: ConcatTimestepEmbedderND with 0 params. Trainable: False Loading model from C:\Users\Students\PycharmProjects\pythonProject\checkpoints/svd-d19a808f.ckpt INFO:main:Starting video generation, this may take a while... Traceback (most recent call last): File "examples/svd/image_to_video.py", line 202, in main(cfg) File "examples/svd/image_to_video.py", line 181, in main vid = pipeline(Tensor(image)) File "C:\Users\Students\PycharmProjects\pythonProject\venv\lib\site-packages\mindspore\nn\cell.py", line 705, in call raise err File "C:\Users\Students\PycharmProjects\pythonProject\venv\lib\site-packages\mindspore\nn\cell.py", line 701, in call output = self._run_construct(args, kwargs) File "C:\Users\Students\PycharmProjects\pythonProject\venv\lib\site-packages\mindspore\nn\cell.py", line 482, in _run_construct output = self.construct(cast_inputs, kwargs) File "examples/svd/image_to_video.py", line 135, in construct samples_x = self.model.decode_first_stage(samples_z) File "C:\Users\Students\PycharmProjects\pythonProject\examples/stable_diffusion_xl\gm\models\diffusion.py", line 139, in decode_first_stage out = self.first_stage_model.decode(z) File "C:\Users\Students\PycharmProjects\pythonProject\examples/stable_diffusion_xl\gm\models\autoencoder.py", line 100, in decode x = self.decoder(z, kwargs) File "C:\Users\Students\PycharmProjects\pythonProject\venv\lib\site-packages\mindspore\nn\cell.py", line 705, in call raise err File "C:\Users\Students\PycharmProjects\pythonProject\venv\lib\site-packages\mindspore\nn\cell.py", line 701, in call output = self._run_construct(args, kwargs) File "C:\Users\Students\PycharmProjects\pythonProject\venv\lib\site-packages\mindspore\nn\cell.py", line 482, in _run_construct output = self.construct(cast_inputs, kwargs) File "C:\Users\Students\PycharmProjects\pythonProject\examples/stable_diffusion_xl\gm\modules\diffusionmodules\model.py", line 509, in construct h = self.mid.block_1(h, temb, kwargs) File "C:\Users\Students\PycharmProjects\pythonProject\venv\lib\site-packages\mindspore\nn\cell.py", line 705, in call raise err File "C:\Users\Students\PycharmProjects\pythonProject\venv\lib\site-packages\mindspore\nn\cell.py", line 701, in call output = self._run_construct(args, kwargs) File "C:\Users\Students\PycharmProjects\pythonProject\venv\lib\site-packages\mindspore\nn\cell.py", line 482, in _run_construct output = self.construct(*cast_inputs, **kwargs) File "C:\Users\Students\PycharmProjects\pythonProject\examples\svd\modules\temporal_ae.py", line 69, in construct x_spat = x.reshape(-1, timesteps, c, h, w).swapaxes(1, 2) # (b t) c h w -> b c t h w File "C:\Users\Students\PycharmProjects\pythonProject\venv\lib\site-packages\mindspore\common\tensor.py", line 2043, in swapaxes return tensor_operator_registry.get('swapaxes')(self, axis0, axis1) File "C:\Users\Students\PycharmProjects\pythonProject\venv\lib\site-packages\mindspore\ops\function\array_func.py", line 7334, in swapaxes axis0, axis1 = _check_swapaxes_axis((axis0, axis1), input.ndim) return len(self.shape) File "C:\Users\Students\PycharmProjects\pythonProject\venv\lib\site-packages\mindspore\common_stub_tensor.py", line 85, in shape self.stub_shape = self.stub.get_shape() TypeError: For primitive[Reshape], the input[shape] must be one of ['int', 'tuple', 'list'] with all Int elements, but got (-1, None, 512, 16, 32)

---

• C++ Call Stack: (For framework developers)

  mindspore\core\utils\check_convert_utils.cc:1204 mindspore::CheckAndConvertUtils::CheckIntOrTupleInt

I am unable to resolve this issue, can you tell me why this is happening, using mindspore with cpu

[feesavage]

do u solve it ? i hve the same problem

[HaFred]

Did you check your input img situation? Make sure it's a valid img and the tensor read in correctly as described in the readme.