Implement initial CI configuration

[nathan-weinberg]

Supersedes #7

This PR adds the following CI jobs:

• Dependabot
• GitHub Action linting
• Markdown Linting (with fixes)
• Spellcheck (with fixes)
• Code Linting (with some fixes - see note below)

As well as a Makefile for local running of the checks.

Additional Notes

• I will do a followup for PyPI packaging but I want to work directly with the folks from Training Squad on that to keep them in the loop
• WRT the Python checks - @RobotSail @Maxusmusti @JamesKunstle I didn't go through all the failures (the fixes I did were auto-format fixes that have no functional impact) because I don't want to hamper y'alls velocity and break things currently working - so the checks remain but are non-blocking for now - I would encourage y'all to address these when it makes sense

[russellb]

know why the jobs aren't running on this PR? I'd prefer to see them run green before merging.

[bjhargrave]

know why the jobs aren't running on this PR?

I don't think github will run new actions in a PR since that is a security risk.

[cdoern]

I will do a followup for PyPI packaging but I want to work directly with the folks from Training Squad on that to keep them in the loop

any chance we can do this today @nathan-weinberg ? I'd really like to test this out

[nathan-weinberg]

know why the jobs aren't running on this PR? I'd prefer to see them run green before merging.

They won't run until they are merged - we saw this in instructlab/eval#1 - but these are a carbon-copy from the current eval config so I'm confident in them

[nathan-weinberg]

I will do a followup for PyPI packaging but I want to work directly with the folks from Training Squad on that to keep them in the loop

any chance we can do this today @nathan-weinberg ? I'd really like to test this out

Let's chat offline

[russellb]

for the Python linting issues you call out -- can you just update the linter config so that it passes? The typical approach I'd prefer is always have a config that runs green, but work over time to improve the config so more rules can be enforced.

[cdoern]

ERROR: Ignored the following versions that require a different python version: 0.52.0 Requires-Python >=3.6,<3.9; 0.52.0rc3 Requires-Python >=3.6,<3.9; 0.53.0 Requires-Python >=3.6,<3.10; 0.53.0rc1.post1 Requires-Python >=3.6,<3.10; 0.53.0rc2 Requires-Python >=3.6,<3.10; 0.53.0rc3 Requires-Python >=3.6,<3.10; 0.53.1 Requires-Python >=3.6,<3.10; 0.54.0 Requires-Python >=3.7,<3.10; 0.54.0rc2 Requires-Python >=3.7,<3.10; 0.54.0rc3 Requires-Python >=3.7,<3.10; 0.54.1 Requires-Python >=3.7,<3.10; 0.55.0 Requires-Python >=3.7,<3.11; 0.55.0rc1 Requires-Python >=3.7,<3.11; 0.55.1 Requires-Python >=3.7,<3.11; 0.55.2 Requires-Python >=3.7,<3.11
ERROR: Could not find a version that satisfies the requirement dolomite-engine (from instructlab-training) (from versions: none)
ERROR: No matching distribution found for dolomite-engine

Found this when pip installing, I think this will fail the linter as well

[nathan-weinberg]

ERROR: Ignored the following versions that require a different python version: 0.52.0 Requires-Python >=3.6,<3.9; 0.52.0rc3 Requires-Python >=3.6,<3.9; 0.53.0 Requires-Python >=3.6,<3.10; 0.53.0rc1.post1 Requires-Python >=3.6,<3.10; 0.53.0rc2 Requires-Python >=3.6,<3.10; 0.53.0rc3 Requires-Python >=3.6,<3.10; 0.53.1 Requires-Python >=3.6,<3.10; 0.54.0 Requires-Python >=3.7,<3.10; 0.54.0rc2 Requires-Python >=3.7,<3.10; 0.54.0rc3 Requires-Python >=3.7,<3.10; 0.54.1 Requires-Python >=3.7,<3.10; 0.55.0 Requires-Python >=3.7,<3.11; 0.55.0rc1 Requires-Python >=3.7,<3.11; 0.55.1 Requires-Python >=3.7,<3.11; 0.55.2 Requires-Python >=3.7,<3.11
ERROR: Could not find a version that satisfies the requirement dolomite-engine (from instructlab-training) (from versions: none)
ERROR: No matching distribution found for dolomite-engine

Found this when pip installing, I think this will fail the linter as well

@RobotSail @Maxusmusti @JamesKunstle any idea why this is happening? I'm hitting the same issue

[russellb]

ERROR: Ignored the following versions that require a different python version: 0.52.0 Requires-Python >=3.6,<3.9; 0.52.0rc3 Requires-Python >=3.6,<3.9; 0.53.0 Requires-Python >=3.6,<3.10; 0.53.0rc1.post1 Requires-Python >=3.6,<3.10; 0.53.0rc2 Requires-Python >=3.6,<3.10; 0.53.0rc3 Requires-Python >=3.6,<3.10; 0.53.1 Requires-Python >=3.6,<3.10; 0.54.0 Requires-Python >=3.7,<3.10; 0.54.0rc2 Requires-Python >=3.7,<3.10; 0.54.0rc3 Requires-Python >=3.7,<3.10; 0.54.1 Requires-Python >=3.7,<3.10; 0.55.0 Requires-Python >=3.7,<3.11; 0.55.0rc1 Requires-Python >=3.7,<3.11; 0.55.1 Requires-Python >=3.7,<3.11; 0.55.2 Requires-Python >=3.7,<3.11
ERROR: Could not find a version that satisfies the requirement dolomite-engine (from instructlab-training) (from versions: none)
ERROR: No matching distribution found for dolomite-engine

Found this when pip installing, I think this will fail the linter as well

@RobotSail @Maxusmusti @JamesKunstle any idea why this is happening? I'm hitting the same issue

ERROR: Ignored the following versions that require a different python version: 0.52.0 Requires-Python >=3.6,<3.9; 0.52.0rc3 Requires-Python >=3.6,<3.9; 0.53.0 Requires-Python >=3.6,<3.10; 0.53.0rc1.post1 Requires-Python >=3.6,<3.10; 0.53.0rc2 Requires-Python >=3.6,<3.10; 0.53.0rc3 Requires-Python >=3.6,<3.10; 0.53.1 Requires-Python >=3.6,<3.10; 0.54.0 Requires-Python >=3.7,<3.10; 0.54.0rc2 Requires-Python >=3.7,<3.10; 0.54.0rc3 Requires-Python >=3.7,<3.10; 0.54.1 Requires-Python >=3.7,<3.10; 0.55.0 Requires-Python >=3.7,<3.11; 0.55.0rc1 Requires-Python >=3.7,<3.11; 0.55.1 Requires-Python >=3.7,<3.11; 0.55.2 Requires-Python >=3.7,<3.11
ERROR: Could not find a version that satisfies the requirement dolomite-engine (from instructlab-training) (from versions: none)
ERROR: No matching distribution found for dolomite-engine

Found this when pip installing, I think this will fail the linter as well

@RobotSail @Maxusmusti @JamesKunstle any idea why this is happening? I'm hitting the same issue

dolomite_engine does not appear to be a valid package that exists on pypi

[cdoern]

also when bringing this into the cli:

ERROR: Cannot install instructlab, instructlab-training==0.1.dev22 and instructlab==0.16.1.dev140 because these package versions have conflicting dependencies.

The conflict is caused by:
    instructlab 0.16.1.dev140 depends on transformers<=4.38.2 and >=4.30.0
    peft 0.9.0 depends on transformers
    trl 0.7.11 depends on transformers>=4.31.0
    instructlab-training 0.1.dev22 depends on transformers==4.41.2

have these versions been vetted as ok to bump in the cli?

[nathan-weinberg]

also when bringing this into the cli:

ERROR: Cannot install instructlab, instructlab-training==0.1.dev22 and instructlab==0.16.1.dev140 because these package versions have conflicting dependencies.

The conflict is caused by:
    instructlab 0.16.1.dev140 depends on transformers<=4.38.2 and >=4.30.0
    peft 0.9.0 depends on transformers
    trl 0.7.11 depends on transformers>=4.31.0
    instructlab-training 0.1.dev22 depends on transformers==4.41.2

have these versions been vetted as ok to bump in the cli?

Can this be a separate issue/convo than the linting introduction?

[RobotSail]

Nice, I didn't know we could do this!

[RobotSail]

This looks good, LGTM

[russellb]

Thanks, @nathan-weinberg !

[russellb]

Well, I thought we were good. Most stuff works, but tox -e mypy is busted on the requirements.txt change made for dolomite_engine.

❯ tox -e mypy
.pkg: _optional_hooks> python /Users/rbryant/src/instructlab/venv/lib/python3.10/site-packages/pyproject_api/_backend.py True setuptools.build_meta
.pkg: get_requires_for_build_sdist> python /Users/rbryant/src/instructlab/venv/lib/python3.10/site-packages/pyproject_api/_backend.py True setuptools.build_meta
mypy: packaging backend failed (code=1), with InvalidRequirement: Expected end or semicolon (after name and no valid version specifier)
    git+https://github.com/ibm-granite/dolomite-engine.git

[russellb]

Well, I thought we were good. Most stuff works, but tox -e mypy is busted on the requirements.txt change made for dolomite_engine.

❯ tox -e mypy
.pkg: _optional_hooks> python /Users/rbryant/src/instructlab/venv/lib/python3.10/site-packages/pyproject_api/_backend.py True setuptools.build_meta
.pkg: get_requires_for_build_sdist> python /Users/rbryant/src/instructlab/venv/lib/python3.10/site-packages/pyproject_api/_backend.py True setuptools.build_meta
mypy: packaging backend failed (code=1), with InvalidRequirement: Expected end or semicolon (after name and no valid version specifier)
    git+https://github.com/ibm-granite/dolomite-engine.git

I got this working and pushed it to the branch

[russellb]

I'm going to merge and we'll keep working on this if anything doesn't pass right away