Module pipeline 

Inference Pipeline Implementations

This module provides optimized pipeline implementations for different inference patterns:

• Chunked Prefill: Split long prompts into chunks for better memory efficiency
• Prefill-Decode Separation: Dedicated handling for each phase
• Batch Pipeline: Efficient processing of multiple requests

Re-exports

pub use chunked_prefill::ChunkedPrefillConfig;

pub use chunked_prefill::ChunkedPrefillExecutor;

pub use executor::ExecutionPhase;

pub use executor::PipelineConfig;

pub use executor::PipelineExecutor;

Modules

chunked_prefill

Chunked Prefill Implementation

executor

Pipeline Executor