Microsoft ML Wrappers
Published:
Responsible AI tools should be able to work with a broad spectrum of machine learning models and datasets. Much of this functionality is based on the ability to call predict or predict_proba on a model and get back the predicted values or probabilities in a specific format.
However, there are many different models outside of scikit-learn and even within scikit-learn which have unusual outputs or require the input in a specific format. Some, like pytorch, don’t even have the predict/predict_proba function specification.
These wrappers handle a variety of frameworks, including pytorch, tensorflow, keras wrappers on tensorflow, variations on scikit-learn models (such as the SVC classification model that doesn’t have a predict_proba function), lightgbm and xgboost, as well as certain strange pipelines we have encountered from customers and internal users in the past.
Contributions
- Extended LLM Support: Added the ability to specify the system prompt and chat history for LLMs.
- Asynchronous OpenAI Execution: Added support for asynchronous execution of OpenAI models, with a rate limiter to prevent going over the API limits.
- Compatibility Improvements: Expanded the compatibility of the wrapper to include all versions of the OpenAI API, from legacy to the latest.