Managed Evaluations
Run models against environments on RLMesh-hosted GPUs and compare the results.
The managed platform runs model–environment evaluations on an RLMesh-hosted cluster. Use the dashboard to choose from the catalog, launch an evaluation, and inspect its results. If you build with the open-source framework locally, upload your model or environment to make it available in the dashboard.
The catalog uses EnvFactory and Model subclasses, the recommended starting
point for your own integrations. Keep environment construction in make() and
model setup and inference in load() and predict(). The same classes support
local evaluation and become container entrypoints through rlmesh.serve; add
environment tags and a model spec to describe how they connect.
New to the platform? Start with Your First Evaluation.