Ritual ML Workflows
Supported Methods
Supported Models

🏫 Supported Models

Classical models

Ritual ML workflows support hooks for training and inference for most classical models that can be done in Python, such as most scikit-learn models. Some common choices below that can be found from the sklearn library.

Model-Type Task Sklearn library
Linear Regressionregressionsklearn.linear_model.LinearRegression
Logistic Regressionclassificationsklearn.linear_model.LogisticRegression
Decision Trees and Random Forests classification, regression sklearn.tree.DecisionTreeClassifier, sklearn.ensemble.RandomForestClassifier sklearn.ensemble.RandomForestRegressor
K-Meansclustering sklearn.cluster.KMeans

Large language models - Text

Ritual ML workflows support inference for open-source models that can be found on any model registry. For seamless user experience, Ritual currently offers optimized support for any large language model with HuggingFace model-ids. It lets users bring their own custom model or use fine-tuned model.

Model Namemodel-id exampleParameter Size
Mistral-7bmistralai/Mistral-7B-v0.17b
Llama, Llama2meta-llama/Llama-2-70b13b, 30b, 70b
Bloombigscience/bloom176b
Falcontiiuae/falcon-7b7b, 40b, 180b
Fine-tuned modelslink oss fine-tuned models
Your custom modelUpload your own

Multi-modal Support

Support for text to image and image to image models are coming soon!