OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Evaluation & Observability

What If Tool

What If Tool is an interactive interface for investigating trained classification and regression models. It runs inference on example sets, visualizes outputs, and lets users edit examples and rerun them to explore changes in predictions, model performance, and fairness across dataset subsets.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

What If Tool — OpenViST project profile in AI Evaluation & Observability.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when you have a trained model and example data and want to explore predictions, compare the effects of edited inputs, or examine performance and fairness across subsets.

Who it is for

  • Machine-learning practitioners examining trained classification or regression models.
  • Notebook and TensorBoard users who want to inspect predictions and model behavior.

Documented uses

  • Run inference over a set of examples and visualize the results.
  • Edit examples manually or programmatically and compare rerun predictions.
  • Investigate model performance and fairness across dataset subsets.
  • Inspect attribution values when using supported AI Platform models or custom prediction functions.
Before you choose

Requirements & limits

The project repository identifies the software as open source under the Apache-2.0 license.

  • TensorBoard integration uses a model served through TensorFlow Serving; users provide its host and port.
  • Notebook support includes TensorFlow Estimators accepting TensorFlow Example or SequenceExample inputs, hosted prediction models with supported input formats, and custom prediction functions with correct input and output
  • For TensorBoard Predict API models, classification outputs must include class probabilities and regression outputs must provide one score per example.
  • The project includes example workflows spanning binary and multiclass classification and regression.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Build and open the Census salary demo

Build and open the What-If Tool binary-classification demo for UCI Census salary prediction.

Before you start

  • The What-If Tool repository and its demo source code.
  • Bazel, to run the documented demo target.

Steps

  1. Run bazel run wit_dashboard/demo:demoserver.

  2. Navigate to http://localhost:6006/wit-dashboard/demo.html.

Expected outcome

The Census salary-prediction demo is available at the documented local address.

Guide limitations

  • The demo predicts whether a person earns more or less than $50,000 using Census information.
  • The source does not specify Bazel installation or other setup steps.
Documentation 1 ↗
COMPARE OPTIONS

Compare options for Model inference and serving

These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.

ProjectDocumented purposePlatformAccess and requirements
What If ToolRun inference over a set of examples and visualize the results.Cloud, WebThe project repository identifies the software as open source under the Apache-2.0 license.
FMBenchCompare inference latency and transactions per minute across model-serving configurations.CLI, CloudFMBench is distributed as a Python package and used from the command line. The project is published under the MIT-0 license.
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
13 October 2026
Original website title
PAIR-code/what-if-tool · Official homepage title · retained in its source language
How we review sourcesSuggest a correction