Pytorch/Diffuser/fastai compatibility issues

Part 2 2022/23
1

!pip install torch==1.13.1 torchvision==0.14.1 torchaudio==0.13.1 torchtext==0.14.1 fastai==2.7.11

!pip install tokenizers

!pip install torchdata==0.5.1

from diffusers import UNet2DModel

ERROR: pip’s dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. gcsfs 2024.10.0 requires fsspec==2024.10.0, but you have fsspec 2024.9.0 which is incompatible.

from diffusers import UNet2DModel
RuntimeError: Failed to import diffusers.models.unets.unet_2d because of the following error (look up to see its traceback):
module ‘torch’ has no attribute ‘float8_e4m3fn’

The AI tool explains the error as follows “The error message AttributeError: module ‘torch’ has no attribute ‘float8_e4m3fn’ indicates that the version of PyTorch you are using does not support the float8_e4m3fn data type. This data type was introduced in a recent version of PyTorch, and your installed version might be older.”
So it means that the installs in the beginning have become outdated. Can you please tell me the updated versions to install for the code segment given below.

!pip install torch==1.13.1 torchvision==0.14.1 torchaudio==0.13.1 torchtext==0.14.1 fastai==2.7.11

!pip install tokenizers

!pip install torchdata==0.5.1

2 
!pip install torch==2.0.0 torchvision==0.16.0 torchaudio==2.1.0 torchtext==0.15.1 fastai==2.7.11

!pip install tokenizers

!pip install torchdata==0.6.0

These versions should be compatible with each other and resolve the dependency conflicts you were encountering. Let me know if you need any further assistance!

Best regards,
Roxanne
Home Depot Survey

3

Hi. Unfortunately the solution didn’t work. I am still getting dependency issues viz !pip install torch==2.0.0 torchvision==0.16.0 torchaudio==2.1.0 torchtext==0.15.1 fastai==2.7.11

!pip install tokenizers

!pip install torchdata==0.6.0ERROR: Cannot install torch==2.0.0 and torchvision==0.16.0 because these package versions have conflicting dependencies.The conflict is caused by:
The user requested torch==2.0.0
torchvision 0.16.0 depends on torch==2.1.0ERROR: pip’s dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
gcsfs 2024.10.0 requires fsspec==2024.10.0, but you have fsspec 2024.9.0 which is incompatible.