fix: enabling block-by-block evaluation for granite-3.x-models

[bayo-ibm]

Description of the change

The block-by-block model evaluation does not work with granite-3.x models. This is because the eval_llm_1GPU does not take into consideration the peculiarities associated with granite-3.x-models.

The quant/ptq.py was updated to reflect the current naming conventions for the various blocks in granite-3.x models.
The eval_llm_1GPU function was updated to correctly calculate the logits for granite-3. x models.

Related issues or PRs

#155

How to verify the PR

Was the PR tested

The fix was validated by performing FP8 DQ-SQ using the granite-3.0-8B-Instruct/granite-3.3-8B-Instruct models. The evaluation of the quantized and the unquantized models using eval_llm_1GPU achieved the same results as using the "evaluator.evaluate" method

• I have added >=1 unit test(s) for every new method I have added (if that coverage is difficult, please briefly explain the reason)
• I have ensured all unit tests pass

Checklist for passing CI/CD:

• All commits are signed showing "Signed-off-by: Name <[email protected]>" with git commit -signoff or equivalent
• PR title and commit messages adhere to Conventional Commits
• Contribution is formatted with tox -e fix
• Contribution passes linting with tox -e lint
• Contribution passes spellcheck with tox -e spellcheck
• Contribution passes all unit tests with tox -e unit

Note: CI/CD performs unit tests on multiple versions of Python from a fresh install. There may be differences with your local environment and the test environment.

[andrea-fasoli]

I worked with Bayo to troubleshoot this Granite evaluation issue. The solution looks functional.

[chichun-charlie-liu]

pylint issue regarding triton kernel will be fixed by #166 .