Error while importing unsloth in databricks

[BurakaKrishna]

Tried installing unsloth in databricks

%%capture
!pip install unsloth "xformers==0.0.28.post2"
# Also get the latest nightly Unsloth!
!pip uninstall unsloth -y && pip install --upgrade --no-cache-dir "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"

While importing from unsloth import FastLanguageModel

Got error

🦥 Unsloth: Will patch your computer to enable 2x faster free finetuning.
Unexpected internal error when monkey patching `PreTrainedModel.from_pretrained`: Failed to import transformers.modeling_utils because of the following error (look up to see its traceback):
operator torchvision::nms does not exist
Unexpected internal error when monkey patching `Trainer.train`: Failed to import transformers.trainer because of the following error (look up to see its traceback):
Failed to import transformers.integrations.integration_utils because of the following error (look up to see its traceback):
Failed to import transformers.modeling_utils because of the following error (look up to see its traceback):
partially initialized module 'torchvision' has no attribute 'extension' (most likely due to a circular import)
Unsloth: If you want to finetune Gemma 2, upgrade flash-attn to version 2.6.3 or higher!
Newer versions support faster and less memory usage kernels for Gemma 2's attention softcapping!
To update flash-attn, do the below:

pip install --no-deps --upgrade "flash-attn>=2.6.3"
AttributeError: partially initialized module 'torchvision' has no attribute 'extension' (most likely due to a circular import)
File <command-1593331561314759>, line 1
----> 1 from unsloth import FastLanguageModel
File /databricks/python/lib/python3.11/site-packages/torchvision/_meta_registrations.py:18, in register_meta.<locals>.wrapper(fn)
     17 def wrapper(fn):
---> 18     if torchvision.extension._has_ops():
     19         get_meta_lib().impl(getattr(getattr(torch.ops.torchvision, op_name), overload_name), fn)
     20     return fn

Anyone who has used unsloth in databricks have faced this issue ?

[BurakaKrishna]

Tried

pip install -U torch==2.3.1
pip install torchvision==0.17.2

Was able to import torchvision, but got a new issue

Unsloth: Will patch your computer to enable 2x faster free finetuning.
WARNING[XFORMERS]: xFormers can't load C++/CUDA extensions. xFormers was built for:
    PyTorch 2.5.0+cu121 with CUDA 1201 (you have 2.2.2+cu121)
    Python  3.11.10 (you have 3.11.0rc1)
  Please reinstall xformers (see https://github.com/facebookresearch/xformers#installing-xformers)
  Memory-efficient attention, SwiGLU, sparse and more won't be available.
  Set XFORMERS_MORE_DETAILS=1 for more details
Unsloth: Your Flash Attention 2 installation seems to be broken?
A possible explanation is you have a new CUDA version which isn't
yet compatible with FA2? Please file a ticket to Unsloth or FA2.
We shall now use Xformers instead, which does not have any performance hits!
We found this negligible impact by benchmarking on 1x A100.
ImportError: Unsloth: You have torch = 2.2.2+cu121 but xformers = 0.0.28.post2.
Please install xformers < 0.0.26 for torch = 2.2.2+cu121.

[danielhanchen]

Try running wget -qO- https://raw.githubusercontent.com/unslothai/unsloth/main/unsloth/_auto_install.py | python - then running that directly