feat: add support for Intel XPU backend in device selection

[xiangyang-95]

This pull request enhances the device compatibility of the lerobot utility functions by adding support for Intel XPU. The changes include updates to device selection, validation, and data type handling to ensure smooth integration of the XPU backend.

Added support for Intel XPU:

• Device selection and logging:

  • Updated auto_select_torch_device to detect and select the XPU backend when available.
  • Enhanced get_safe_torch_device to handle "xpu" as a valid device option.

• Data type compatibility:

  • Modified get_safe_dtype to check XPU device capabilities for float64 support. If unsupported, it falls back to float32 with appropriate warnings.

• Device availability checks:

  • Extended is_torch_device_available to validate the availability of the XPU backend.

• AMP (Automatic Mixed Precision) support:

  • Updated is_amp_available to include XPU as a device supporting AMP.This pull request enhances the src/lerobot/utils/utils.py file by adding support for Intel XPU devices. The changes ensure that the utility functions can detect, validate, and handle the XPU backend alongside existing backends like CUDA, MPS, and CPU.

Added support for Intel XPU:

• auto_select_torch_device: Added logic to detect and select the XPU backend when available.
• get_safe_torch_device: Extended the function to handle "xpu" as a valid device option, ensuring proper assertions and device creation.
• get_safe_dtype: Introduced checks for XPU devices to handle cases where torch.float64 is unsupported, falling back to torch.float32 if necessary.
• is_torch_device_available: Updated to include detection of XPU availability and adjusted the error message to list XPU as a supported device.

How to checkout & try? (for the reviewer)

Provide a simple way for the reviewer to try out your changes.

Examples:

Install environment

python3 -m venv .venv
source .venv/bin/activate
git clone https://github.com/huggingface/lerobot.git
cd lerobot
pip install -e ".[smolvla]" --extra-index-url https://download.pytorch.org/whl/xpu

Run training

Tested on Intel B580 GPU

python3 src/lerobot/scripts/train.py \
  --policy.path=lerobot/smolvla_base \
  --policy.push_to_hub=false \
  --policy.device=xpu \
  --dataset.repo_id=lerobot/svla_so100_stacking \
  --batch_size=32 \
  --steps=20000 \
  --output_dir=outputs/train/my_smolvla \
  --job_name=my_smolvla_training \
  --wandb.enable=false

SECTION TO REMOVE BEFORE SUBMITTING YOUR PR

Note: Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR. Try to avoid tagging more than 3 people.

Note: Before submitting this PR, please read the contributor guideline.

[Copilot]

Pull Request Overview

This PR adds support for Intel XPU backend to the lerobot utility functions, expanding device compatibility beyond the existing CUDA, MPS, and CPU backends. The changes ensure that XPU devices can be automatically detected, validated, and properly configured with appropriate data type handling.

• Added XPU device detection and selection in auto device selection logic
• Extended device validation and creation functions to handle XPU devices
• Implemented XPU-specific data type compatibility checks with float64 fallback handling