Overview

Analyse runs on prompts and workflows

After configuring your prompts/workflows, datasets, and evaluators in Maxim, you can easily test your prompts or workflow on the test dataset and the selected evaluators. This allows you to evaluate your application seamlessly in just a few clicks.

Understand trends

Looking at historical data of runs can help you gain the following kind of insights:

1. Comparison across configurations: Compare all your historical test runs across different prompt and workflow configurations. This allows you to track progress and trends over time as your datasets and prompts/workflows evolve.
2. Insight into queries: Gain valuable insights into which queries are performing well and which are not. This helps in pinpointing areas that need enhancement and identifying the right fallback mechanisms.
3. Performance trends: Observe how the performance of your workflows changes over time with changes in prompts, or with respect to a new model or against your evolving datasets. This helps in iterating and improving the efficiency and accuracy of your AI applications.

Customise your view

The reports allow you to easily filer and search, show/hide columns, re-order columns, pin columns and even add evaluators post run in order to view the right information for you to take informed decisions. Learn more about report customisability here.

Share reports

Share the test run reports easily with anyone via a simple share link. This feature facilitates collaboration and communication within your team or with external stakeholders. This report is visible to anyone who you share with and they don’t need to be on the platform to view the report.

Organise all reports

By default the list of all runs is searchable and can be filtered for certain criteria. This can help you reach the relevant report from the past quickly. To further organise your reports, you can now add tags on them Eg. Finalised tag on the experiments which made their way into production.