Skip requirement inference for PEFT model

[B-Step62]

🛠 DevTools 🛠

Install mlflow from this PR

pip install git+https://github.com/mlflow/mlflow.git@refs/pull/11184/merge

Checkout with GitHub CLI

gh pr checkout 11184

Related Issues/PRs

#xxx

What changes are proposed in this pull request?

Pip requirement inference calls mlflow.pyfunc.load_model() for the model to be logged, which loads a full copy of the model weight onto the memory in a subprocess. This is not preferable for PEFT use case as it is designed for fine-tuning large model under limited GPU memory. At worst case, it crashes the training process due to OOM. Also it is common technique to use lower precision for the base model during fine-tuning (QLoRA), but the requirement inference always use the default precision, leading to much larger GPU memory than required for fine-tuning itself.

Ideally, we can implement more light-weight inference logic, for example, patch model loading method, but it needs to be designed more carefully not to cause regression. The default requirements should work for most of the PEFT use cases, because there is very little space for users to write custom inference logic over PEFT's abstraction.

How is this PR tested?

• [x] Existing unit/integration tests
• [x] New unit/integration tests
• [x] Manual tests

Does this PR require documentation update?

• [x] No. You can skip the rest of this section.
• [ ] Yes. I've updated:
  • [ ] Examples
  • [ ] API references
  • [ ] Instructions

This behavior will be added to documentation in follow-up.

Release Notes

Is this a user-facing change?

• [x] No. You can skip the rest of this section.
• [ ] Yes. Give a description of this change to be included in the release notes for MLflow users.

What component(s), interfaces, languages, and integrations does this PR affect?

Components

• [ ] area/artifacts: Artifact stores and artifact logging
• [ ] area/build: Build and test infrastructure for MLflow
• [ ] area/deployments: MLflow Deployments client APIs, server, and third-party Deployments integrations
• [ ] area/docs: MLflow documentation pages
• [ ] area/examples: Example code
• [ ] area/model-registry: Model Registry service, APIs, and the fluent client calls for Model Registry
• [x] area/models: MLmodel format, model serialization/deserialization, flavors
• [ ] area/recipes: Recipes, Recipe APIs, Recipe configs, Recipe Templates
• [ ] area/projects: MLproject format, project running backends
• [ ] area/scoring: MLflow Model server, model deployment tools, Spark UDFs
• [ ] area/server-infra: MLflow Tracking server backend
• [ ] area/tracking: Tracking Service, tracking client APIs, autologging

Interface

• [ ] area/uiux: Front-end, user experience, plotting, JavaScript, JavaScript dev server
• [ ] area/docker: Docker use across MLflow's components, such as MLflow Projects and MLflow Models
• [ ] area/sqlalchemy: Use of SQLAlchemy in the Tracking Service or Model Registry
• [ ] area/windows: Windows support

Language

• [ ] language/r: R APIs and clients
• [ ] language/java: Java APIs and clients
• [ ] language/new: Proposals for new client languages

Integrations

• [ ] integrations/azure: Azure and Azure ML integrations
• [ ] integrations/sagemaker: SageMaker integrations
• [ ] integrations/databricks: Databricks integrations

How should the PR be classified in the release notes? Choose one:

• [x] rn/none - No description will be included. The PR will be mentioned only by the PR number in the "Small Bugfixes and Documentation Updates" section
• [ ] rn/breaking-change - The PR will be mentioned in the "Breaking Changes" section
• [ ] rn/feature - A new user-facing feature worth mentioning in the release notes
• [ ] rn/bug-fix - A user-facing bug fix worth mentioning in the release notes
• [ ] rn/documentation - A user-facing documentation change worth mentioning in the release notes

[github-actions[bot]]

Documentation preview for 80b4819d36c1e2c6fbc4878c330ad9dbd0d63f7d will be available when this CircleCI job completes successfully.

• Top page
• Changed pages

More info

• Ignore this comment if this PR does not change the documentation.
• It takes a few minutes for the preview to be available.
• The preview is updated when a new commit is pushed to this PR.
• This comment was created by https://github.com/mlflow/mlflow/actions/runs/7961685521.