How do I install the nightly build of PyTorch/XLA

1. Install PyTorch Nightly Build:
First, install the PyTorch nightly build using pip. You can specify the version and date if needed. For example, to install a recent nightly build, you can use:

bash
   pip3 install torch==2.7.0.dev20250124+cpu --index-url https://download.pytorch.org/whl/nightly/cpu
   

2. Install PyTorch/XLA Nightly Build:
Next, install the PyTorch/XLA nightly build. You can download the wheel directly from the Google Cloud Storage bucket. For example, for Python 3.10:

bash
   pip3 install https://storage.googleapis.com/pytorch-xla-releases/wheels/tpuvm/torch_xla-2.7.0.dev20250124+cxx11-cp310-cp310-linux_x86_64.whl
   

Make sure to replace the version and date with the one you need.

3. Optional Dependencies:
If you are using custom kernels, you might need to install additional dependencies like Pallas. You can do this by adding the Pallas option:

bash
   pip install 'torch_xla[pallas]' \
       -f https://storage.googleapis.com/jax-releases/jax_nightly_releases.html \
       -f https://storage.googleapis.com/jax-releases/jaxlib_nightly_releases.html
   

4. Additional Libraries:
Depending on your environment, you might need to install additional libraries such as `libopenblas-dev` or `libgl1`:

bash
   sudo apt-get update && sudo apt-get install libopenblas-dev -y
   sudo apt-get update && sudo apt-get install libgl1 -y
   

These steps should help you install the nightly build of PyTorch/XLA on your system[1][3].