Sign In

Dataster Documentation

Dataster helps you build Generative AI applications with better accuracy and lower latency.
Back to documentation

Model End-Of-Life Planning

Strategic AI Asset Management is a key part of LLMOps and GenAIOps. A crucial aspect of this is planning for the lifecycle of models, especially as cloud providers retire specific models. When this happens, your API calls may start returning errors, potentially disrupting your operations. Dataster allows you to visualize upcoming retirement or end-of-life dates for your models across various cloud providers. It also helps you identify suitable replacement candidates, ensuring you have ample time for proper planning, benchmarking, and making informed decisions. With this feature, you can seamlessly transition to new models without unexpected surprises, maintaining the smooth operation of your production environment.


The models that Dataster provides access to out-of-the-box are shared. For private access, users can bring their own models into Dataster as assets, ensuring that only they can access those models. The catalog displays Bring Your Own (BYO) LLMs differently from the Dataster-provided LLMs. Instead of showing the Dataster badge, it gives the user the option to edit or delete the model from the catalog.


Dataster provided vs BYO LLM

Model End-Of-Life Planning allows you to visualize the upcoming retirement or end-of-life dates for your models across several cloud providers for both the Dataster-provided models and the BYO models.

Prerequisites

  1. A Dataster account.
  2. Optionally, one or more BYO Models.

Step 1: Navigate to the Lifecycle Page

  1. Navigate to the Lifecycle page by clicking "Lifecycle" in the left navigation pane.

Step 2: Strategize

  1. Note the upcoming retirement or end-of-life dates for your models.
  2. Hover over the models that concern you and identify potential candidates.
  3. Run performance tests on the candidates to ensure they meet your requirements.


Model End-Of-Life Planning

Conclusion

By effectively planning for the end-of-life of your models, you can ensure a smooth transition and maintain the reliability of your operations. Dataster's Model End-Of-Life Planning feature provides you with the tools to visualize upcoming retirement dates, identify suitable replacement candidates, and perform necessary performance tests. This proactive approach helps you avoid unexpected disruptions and keeps your production environment running seamlessly. Whether using Dataster-provided models or your own, you can confidently manage your AI assets and stay ahead of any potential challenges.


If you encounter any issues or need further assistance, please contact our support team at support@dataster.com.