Fix offload tests for CogVideoX and CogView3

[a-r-r-o-w]

Addresses https://huggingface.slack.com/archives/C065E480NN9/p1736490500126669

[a-r-r-o-w]

@DN6 Seems like we're having infra issues? Tests are not queued and unable to SSH into GPU runners to check if my locally tested changes work on CI as well

[a-r-r-o-w]

Everything seems to be passing now on CI:

Logs

root@fbd443efc79b:/__w/diffusers/diffusers# pytest -s tests/models/transformers/test_models_transformer_cogview3plus.py
==================================================================================================================== test session starts ====================================================================================================================
platform linux -- Python 3.10.16, pytest-8.3.4, pluggy-1.5.0
rootdir: /__w/diffusers/diffusers
configfile: pyproject.toml
plugins: requests-mock-1.10.0, xdist-3.6.1, timeout-2.3.1
collected 29 items

.An error occurred while trying to fetch /tmp/tmpazip7w_o: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpazip7w_o.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpazip7w_o: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpazip7w_o.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
....An error occurred while trying to fetch /tmp/tmpptwbxmre: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpptwbxmre.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
.An error occurred while trying to fetch /tmp/tmpdiitv_e2: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpdiitv_e2.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpdiitv_e2: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpdiitv_e2.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpdmsqtcpe: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpdmsqtcpe.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpdmsqtcpe: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpdmsqtcpe.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpn2rp5bww: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpn2rp5bww.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpn2rp5bww: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpn2rp5bww.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
.sAn error occurred while trying to fetch /tmp/tmpg8efqwfw: Error no file named diffusion_pytorch_model.fp16.safetensors found in directory /tmp/tmpg8efqwfw.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpg8efqwfw: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpg8efqwfw.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
...An error occurred while trying to fetch /tmp/tmpk_pnkrad: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpk_pnkrad.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
.s.....ss......

===================================================================================================================== warnings summary ======================================================================================================================
tests/models/transformers/test_models_transformer_cogview3plus.py::CogView3PlusTransformerTests::test_disk_offload_without_safetensors
  /opt/venv/lib/python3.10/site-packages/accelerate/utils/modeling.py:1731: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
    return torch.load(checkpoint_file, map_location=torch.device("cpu"))

tests/models/transformers/test_models_transformer_cogview3plus.py::CogView3PlusTransformerTests::test_ema_training
  /__w/diffusers/diffusers/tests/models/test_modeling_common.py:799: UserWarning: Using a target size (torch.Size([2, 1, 4, 8, 8])) that is different to the input size (torch.Size([2, 4, 8, 8])). This will likely lead to incorrect results due to broadcasting. Please ensure they have the same size.
    loss = torch.nn.functional.mse_loss(output, noise)

tests/models/transformers/test_models_transformer_cogview3plus.py::CogView3PlusTransformerTests::test_training
  /__w/diffusers/diffusers/tests/models/test_modeling_common.py:780: UserWarning: Using a target size (torch.Size([2, 1, 4, 8, 8])) that is different to the input size (torch.Size([2, 4, 8, 8])). This will likely lead to incorrect results due to broadcasting. Please ensure they have the same size.
    loss = torch.nn.functional.mse_loss(output, noise)

-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
========================================================================================================= 25 passed, 4 skipped, 3 warnings in 5.27s =========================================================================================================
root@fbd443efc79b:/__w/diffusers/diffusers# pytest -s tests/models/transformers/test_models_transformer_cogvideox.py
==================================================================================================================== test session starts ====================================================================================================================
platform linux -- Python 3.10.16, pytest-8.3.4, pluggy-1.5.0
rootdir: /__w/diffusers/diffusers
configfile: pyproject.toml
plugins: requests-mock-1.10.0, xdist-3.6.1, timeout-2.3.1
collected 58 items

.An error occurred while trying to fetch /tmp/tmpkiuxwulf: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpkiuxwulf.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpkiuxwulf: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpkiuxwulf.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
....An error occurred while trying to fetch /tmp/tmpmw3n5hp2: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpmw3n5hp2.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
.An error occurred while trying to fetch /tmp/tmpw6z3ln60: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpw6z3ln60.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpw6z3ln60: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpw6z3ln60.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpmepparxv: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpmepparxv.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpmepparxv: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpmepparxv.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmph970w7u1: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmph970w7u1.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmph970w7u1: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmph970w7u1.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
.sAn error occurred while trying to fetch /tmp/tmpwrih7a3n: Error no file named diffusion_pytorch_model.fp16.safetensors found in directory /tmp/tmpwrih7a3n.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpwrih7a3n: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpwrih7a3n.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
...An error occurred while trying to fetch /tmp/tmppbwk5cjy: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmppbwk5cjy.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
.An error occurred while trying to fetch /tmp/tmpj4qm_r2o: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpj4qm_r2o.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpj4qm_r2o: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpj4qm_r2o.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
....An error occurred while trying to fetch /tmp/tmpr_yqyogw: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpr_yqyogw.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
.An error occurred while trying to fetch /tmp/tmp25833dfz: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmp25833dfz.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmp25833dfz: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmp25833dfz.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmprl9pn13v: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmprl9pn13v.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmprl9pn13v: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmprl9pn13v.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpma808kx_: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpma808kx_.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpma808kx_: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpma808kx_.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
.sAn error occurred while trying to fetch /tmp/tmpz03hdflr: Error no file named diffusion_pytorch_model.fp16.safetensors found in directory /tmp/tmpz03hdflr.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
An error occurred while trying to fetch /tmp/tmpz03hdflr: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpz03hdflr.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
...An error occurred while trying to fetch /tmp/tmpw40r1d4p: Error no file named diffusion_pytorch_model.safetensors found in directory /tmp/tmpw40r1d4p.
Defaulting to unsafe serialization. Pass `allow_pickle=False` to raise an error instead.
.s.....ss......

===================================================================================================================== warnings summary ======================================================================================================================
tests/models/transformers/test_models_transformer_cogvideox.py::CogVideoXTransformerTests::test_disk_offload_without_safetensors
tests/models/transformers/test_models_transformer_cogvideox.py::CogVideoX1_5TransformerTests::test_disk_offload_without_safetensors
  /opt/venv/lib/python3.10/site-packages/accelerate/utils/modeling.py:1731: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.
    return torch.load(checkpoint_file, map_location=torch.device("cpu"))

tests/models/transformers/test_models_transformer_cogvideox.py::CogVideoX1_5TransformerTests::test_ema_training
  /__w/diffusers/diffusers/tests/models/test_modeling_common.py:799: UserWarning: Using a target size (torch.Size([2, 1, 4, 8, 8])) that is different to the input size (torch.Size([2, 2, 4, 8, 8])). This will likely lead to incorrect results due to broadcasting. Please ensure they have the same size.
    loss = torch.nn.functional.mse_loss(output, noise)

tests/models/transformers/test_models_transformer_cogvideox.py::CogVideoX1_5TransformerTests::test_training
  /__w/diffusers/diffusers/tests/models/test_modeling_common.py:780: UserWarning: Using a target size (torch.Size([2, 1, 4, 8, 8])) that is different to the input size (torch.Size([2, 2, 4, 8, 8])). This will likely lead to incorrect results due to broadcasting. Please ensure they have the same size.
    loss = torch.nn.functional.mse_loss(output, noise)

-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
========================================================================================================= 50 passed, 8 skipped, 4 warnings in 8.49s =========================================================================================================