fix colab notebooks

[tburleyinfo]

Summary

Type of contribution

• New worker
• New analyzer
• Bug fix
• Other (describe below)

Hit cuda13 compatibility and dependency errors in colab environment. Resolved by uninstalling cuda13 and related dependencies and using cuda128 and related dependencies. Requires the install cell to be ran twice to have the new packages in the runtime environment. Run the install cell once, then restart the runtime (do not disconnect and delete) and then run all cells.

Files modified

notebooks/*colab.ipynb

The following files are core infrastructure. Please discuss with maintainers before modifying:
hook_llm.py, _hook_plugin.py, registry.py, hook_client.py, run_utils.py, shm_utils.py, workers/_common.py

• I have NOT modified core files, OR I have discussed the change with maintainers
• If I added a new worker/analyzer, I registered it in __init__.py

Plugin architecture checklist

• New workers/analyzers are registered via PluginRegistry in __init__.py
• New workers are implemented as worker_extension_cls (for in-band artifact retrieval and concurrent request support)
• Examples or notebooks are included for new features

Testing

Ran each notebook to ensure each would complete.

Related issue

Contribution acknowledgement

If this contribution is included in a future version of the vLLM-Hook technical report, would you like to be credited as a co-author?

• Yes, please include me as a contributor
• No, thanks

If yes, please provide:

• Name: Timothy Burley
• Affiliation: Independent
• One-sentence description of your contribution: Temporary fix for cuda13 compatibility and dependency errors in colab environment.