Fix package validation failure with llama.cpp

[shalinib-ibm]

Installing the .whl file and running llama binaries

1. pip show llama_cpp_python_package
2. pip install llama_cpp_python_package-1.0+ppc64le10145-py3-none-linux_ppc64le.whl
3. pip show llama_cpp_python_package
4. export LD_LIBRARY_PATH=/usr/local/lib/python3.9/site-packages/llama_cpp_python_package/lib:$LD_LIBRARY_PATH
5. /usr/local/lib/python3.9/site-packages/llama_cpp_python_package/bin/llama-bench -m /home/shalini/Models/granite-3.3-8b-instruct-f16.gguf -t 20

Checklist

• Have you checked and followed all the points mention in the CONTRIBUTING.MD
• Have you validated script on UBI 9 container
• Did you run the script(s) on fresh container with set -e option enabled and observe success ?
• Did you have Legal approvals for patch files ?

[cdeepali]

@shalinib-ibm the output wheel generated is llama_cpp_scripts-0.0.0-py3-none-any.whl. Can you please check.

Also, currently the script is using master branch. Can you please change it to build desired version?

[shalinib-ibm]

@shalinib-ibm the output wheel generated is llama_cpp_scripts-0.0.0-py3-none-any.whl. Can you please check.

Also, currently the script is using master branch. Can you please change it to build desired version?

Thank you @deepali . I have fixed the versioning number in wheel.

[cdeepali]

@shalinib-ibm the generated wheel file is still, pls check.

2026-07-06T10:17:54.9528111Z ============ Final wheel: llama_cpp_scripts-0.0.0+ppc64le1-py3-none-any.whl ===========.

Also there is an error while creating wheel:

2026-07-06T10:19:11.4959340Z ==================== Building wheel ====================
2026-07-06T10:19:11.4960572Z 
2026-07-06T10:19:11.4961181Z * Getting build dependencies for wheel...
2026-07-06T10:19:11.4963261Z ERROR Backend 'poetry.core.masonry.api' is not available.
2026-07-06T10:19:11.4964740Z 
2026-07-06T10:19:11.4965384Z Traceback (most recent call last):
2026-07-06T10:19:11.4968411Z   File "/home/tester/pyvenv_3.11/lib64/python3.11/site-packages/pyproject_hooks/_impl.py", line 402, in _call_hook
2026-07-06T10:19:11.4978681Z     raise BackendUnavailable(
2026-07-06T10:19:11.4981557Z pyproject_hooks._impl.BackendUnavailable: Cannot import 'poetry.core.masonry.api'
2026-07-06T10:19:11.4984530Z 
2026-07-06T10:19:11.4984631Z 
2026-07-06T10:19:11.4986825Z ===> Wheel Creation Failed for Python 3.11 (without isolation)
2026-07-06T10:19:11.4988907Z 
2026-07-06T10:19:11.4989827Z * Creating isolated environment: venv+pip...
2026-07-06T10:19:11.4991784Z * Installing packages in isolated environment:
2026-07-06T10:19:11.4993613Z   - poetry-core>=1.0.0
2026-07-06T10:19:11.4995038Z * Getting build dependencies for wheel...
2026-07-06T10:19:11.4996537Z * Building wheel...
2026-07-06T10:19:11.4998450Z Successfully built llama_cpp_scripts-0.0.0-py3-none-any.whl

[github-actions]

@shalinib-ibm

  **Reminder 1**

  This PR has had **no activity for 20 days**.
  Please respond to keep it active.

[shalinib-ibm]

@cdeepali I have fixed the package validation failure and the generataed wheel file also has build tag information as version number . Can you please take a look now ?

[cdeepali]

@shalinib-ibm The CI adds the ppc64le suffix to the wheel as part of postprocessing. The the wheel file being generated now is:

llama_cpp_python_package-1.0+ppc64le10158ppc64le1-py3-none-manylinux_2_34_ppc64le.whl

Can you skip added ppc64le in the version, because the CI will anyway add it, and then we will get wheel name like:
llama_cpp_python_package-1.0+10158ppc64le1-py3-none-manylinux_2_34_ppc64le.whl

[shalinib-ibm]

@shalinib-ibm The CI adds the ppc64le suffix to the wheel as part of postprocessing. The the wheel file being generated now is:

llama_cpp_python_package-1.0+ppc64le10158ppc64le1-py3-none-manylinux_2_34_ppc64le.whl

Can you skip added ppc64le in the version, because the CI will anyway add it, and then we will get wheel name like: llama_cpp_python_package-1.0+10158ppc64le1-py3-none-manylinux_2_34_ppc64le.whl

Thank you @cdeepali for pointing this out. Removed extra ppc64le from the generated .whl name. It now matches. what you have given above.
llama_cpp_python_package-1.0+10158ppc64le1-py3-none-manylinux_2_34_ppc64le.whl

[cdeepali]

@shalinib-ibm pls check minor comments wrt the echo statements.