Examples
Example tutorials of how to use and troubleshoot with Arize.
LLM Examples
Setting up Tracing using Popular Integrations
LLM Models | Orchestration Frameworks | Other |
---|---|---|
Setting up Tracing using OTEL
LlamaIndex Tracing | |
Langchain Tracing | |
Logging data manually |
Evaluation
Running Evals with LlamaIndex | |
Running Evals with Langchain |
By Use Case
Prompt Template Iteration | |
Search and Retrieval with Llamaindex | |
Langchain Callback Integration | |
Text Summarization |
ML/CV Model Type Examples
Model Type | Pandas Batch | Python Single Record | CSV | Parquet |
---|---|---|---|---|
Binary Classification (Only Classification Metrics) | ||||
Binary Classification (Classification, AUC/Log Loss Metrics) | Download File *Open Parquet Reader Here | |||
Binary Classification (Classification, AUC/Log Loss, Regression) | Download File *Open Parquet Reader Here | |||
Multi-class Classification (Single-Label) | Download File *Open Parquet Reader Here | |||
Multi-class Classification (Multi-Label) | Download File *Open Parquet Reader Here | |||
Regression | ||||
Tabular Classification w/ Embeddings | ||||
Timeseries Forecasting | ||||
Ranking with Relevance Score | ||||
Ranking with Single Label | ||||
Ranking with Multiple Labels | ||||
NLP Classification | ||||
NLP Named Entity Recognition (NER) | ||||
CV Classification | ||||
Object Detection | ||||
Large Language Models (LLMs) - Text Summarization | ||||
Large Language Models (LLMs) - Search and Retrieval |
Explainability Tutorials
Examples for logging explainability metrics. Click here for more information on how to log feature importance and use explainability.
SHAP: Guide to Getting Started | |
SHAP: Neural Network on Tabular Data | |
Surrogate Model Explainability | |
One Hot Encoding Decomposition |
Cloud Storage Examples
Benchmark Test
Sending 10 Million Inferences to Arize in 90 Seconds |
Logging Predictions, Actuals, SHAP Values
Tutorials on how to log predictions, actuals, and feature importance.
Logging Predictions Only | |
Logging Predictions First, Then Logging Delayed Actuals | |
Logging Predictions First, Then Logging SHAPs After | |
Logging Predictions and Actuals Together | |
Logging Predictions and SHAP Together | |
Logging Predictions, Actuals, and SHAP Together | |
Logging PySpark DataFrames |
Example Integrations with Common ML/Data Platforms
Arize integrates with platforms across the MLOps toolchain. Don't see a platform you use? Reach out to add yours or ask our team to help!
ML Platform | Description | Example Integration | Blog |
---|---|---|---|
Algorithmia | MLOps platform with APIs to serve, host and manages models | ||
Anyscale | Integration tutorial for Anyscale's LLM Endpoints offering | ||
Azure ML & Databricks | Using Arize in an Azure ML Databricks workflow | ||
Bento ML | Use Bento’s ML service platform to turn ML models into production-worthy prediction services | ||
CML | Integrate Arize into the CI/CD workflow - Run checks on every new model version | ||
Deepnote | Deepnote is a Data Science Collaboration Platform | ||
Feast | Monitor & Troubleshoot any data inconsistency issue with feature stores Arize. | ||
Google Cloud ML (Vertex AI) | Integrate Arize with Vertex AI | Available on Request | |
Hugging Face | Use Arize to monitor embeddings generated from Hugging Face NLP or Transformer models | ||
Kafka | Use Arize Pandas SDK to consumes micro-batches of predictions | ||
Langchain | Effectively monitor the performance of your LLM agents | ||
MLFlow | Integrating Arize and MLflow to track the model across experimentation and deployment | ||
Neptune | Integrate Arize on models built using Neptune | ||
OpenAI | Build unstructured models with OpenAI | ||
Paperspace | Integrate Arize on models built using Paperspace | ||
PySpark | To log Spark DataFrames, which have | ||
Ray Serve (Anyscale) | Arize can be easily integrated with Ray Serve with at single entry point during | ||
Sagemaker | |||
Spell | Combine Spell model servers with Arize model observability | ||
UbiOps | Arize platform can easily integrate with UbiOps to enable model observability, explainability, and monitoring. | ||
Weights & Biases | Integrating Arize and W&B to track the model across experimentation and deployment |
Common Industry Use Cases
Common Industry Use CasesLast updated