Add accelerate API support for Word Language Model example

[framoncg]

Refactor Word Language Model example to utilize torch.accelerator API torch.accelerator API allows to abstract some of the accelerator specifics in the user scripts. By leveraging this API, the code becomes more adaptable to various hardware accelerators.

Updated word_language_model/main.py with accelerator flag Updated word_language_model/generate.py with accelerator flag Updated README to match word_language_model/main.py flags Updated run_python_examples.sh to add new accelerator flag

CC: @msaroufim, @malfet, @dvrogozh