[Bug]: [V1] Support Fallback For Unsupported Head Dim on FA

[robertgshaw2-redhat]

Your current environment

The output of `python collect_env.py`

Your output of `python collect_env.py` here

🐛 Describe the bug

Currently, if a model has an unsupported head_dim (e.g. microsoft/phi-2, which has head_dim=80), we raise a ValueError.

Options:

• Once we support Triton backend on V0, fall back
• Try to detect this at startup time and fall back to V0. The challenge with this is that the head_dim is not resolved until we get to the model architecture, which is too late

Before submitting a new issue...

• Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

[github-actions]

This issue has been automatically marked as stale because it has not had any activity within 90 days. It will be automatically closed if no further activity occurs within 30 days. Leave a comment if you feel this issue should remain open. Thank you!