Cannot install on Google Colab

[benjamin-marie]

Please check that this issue hasn't been reported before.

• I searched previous Bug Reports didn't find any similar reports.

Expected Behavior

axolotl should be installed.

Current behaviour

When I ran the installation command on Google Colab (L4 GPU):
!git clone https://github.com/axolotl-ai-cloud/axolotl.git
!cd axolotl && pip3 install packaging ninja && pip3 install -e '.[flash-attn,deepspeed]'

It yields:

ERROR: Cannot install None and axolotl because these package versions have conflicting dependencies.

The conflict is caused by:
axolotl 0.4.1 depends on torch==2.4.1+cu121
accelerate 0.34.2 depends on torch>=1.10.0
bitsandbytes 0.44.0 depends on torch
liger-kernel 0.3.0 depends on torch>=2.1.2
optimum 1.16.2 depends on torch>=1.11
peft 0.13.0 depends on torch>=1.13.0
trl 0.9.6 depends on torch>=1.4.0
xformers 0.0.27 depends on torch==2.3.1

To fix this you could try to:

1. loosen the range of package versions you've specified
2. remove package versions to allow pip to attempt to solve the dependency conflict

ERROR: ResolutionImpossible: for help visit https://pip.pypa.io/en/latest/topics/dependency-resolution/#dealing-with-dependency-conflicts

Steps to reproduce

Go on Google Colab.
Select a GPU instance, e.g. L4
And run the installation commands
!git clone https://github.com/axolotl-ai-cloud/axolotl.git
!cd axolotl && pip3 install packaging ninja && pip3 install -e '.[flash-attn,deepspeed]'

Config yaml

No response

Possible solution

Update the requirements.

Which Operating Systems are you using?

• Linux
• macOS
• Windows

Python Version

3.10

axolotl branch-commit

main

Acknowledgements

• My issue title is concise, descriptive, and in title casing.
• I have searched the existing issues to make sure this bug has not been reported yet.
• I am using the latest version of axolotl.
• I have provided enough information for the maintainers to reproduce and diagnose the issue.

[nicokim]

Same thing happened to me on runpod.
I updated the version of the transformers and xformers in the requirements.txt

xformers==0.0.28
transformers==4.45.1

and it worked somehow haha! Can you try it?

[benjamin-marie]

Yes, it works.
The requirements must be updated.
I expect new installations of Axolotl to be impossible until it's fixed.

[Hasan-Demez]

i also ran across the same issue in vscode and the same solution worked for me as well, thank you.

[NanoCode012]

Hey, sorry for this. We're trying to make things more clear on this. Colab's version of torch may be incompatible with the specific version required by xformers. Previously, you should install torch==2.3.1 and now 2.4.1 for xformers to be happy. Currently, Colab uses 2.4.1, so this issue should not exist.

I plan to remove dependency on xformers out as it's creating this locking issue.

[Amit30swgoh]

Python 3.10.12

pip install torch==2.4.1+cu124 torchvision==0.19.1+cu124 torchaudio==2.4.1+cu124 torchtext==0.18.0 torchdata==0.8.0 --extra-index-url https://download.pytorch.org/whl/cu124
pip install xformers==0.0.28.post1
pip install torch==2.4.0+cu124 torchvision==0.19.0+cu124 torchaudio==2.4.0+cu124 torchtext==0.18.0 torchdata==0.8.0 --extra-index-url https://download.pytorch.org/whl/cu124
pip install xformers==0.0.27.post2
pip install torch==2.3.1+cu121 torchvision==0.18.1+cu121 torchaudio==2.3.1+cu121 torchtext==0.18.0 torchdata==0.8.0 --extra-index-url https://download.pytorch.org/whl/cu121
pip install xformers==0.0.27

Enjoy !

[NanoCode012]

I updated the colab example in the linked PR. It uses the instructions on the readme which installs fine. I would recommend anyone having this issue to check it out. Sorry for any confusion with this.