AutoDeploy Transforms

This section documents the AutoDeploy transform interfaces and registered pipeline transforms. Use the stage pages to find where a transform runs in the optimization pipeline, what graph or runtime change it performs, and which configuration fields are available.

For an overview of how transforms fit into the AutoDeploy pipeline, see AutoDeploy (Beta). For information on configuring which transforms run and in what order, see Expert Configuration of LLM API.

• Core Transform APIs
• Factory Stage
• Export Stage
• Post-Export Stage
• Pattern Matching Stage
• Sharding Stage
• Weight Loading Stage
• Post-Load Fusion Stage
• Cache Initialization Stage
• Visualization Stage
• Compilation Stage
• Additional Registered Transforms