ferrum_engine/pipeline/mod.rs
1//! Inference Pipeline Implementations
2//!
3//! This module provides optimized pipeline implementations for different
4//! inference patterns:
5//!
6//! - Chunked Prefill: Split long prompts into chunks for better memory efficiency
7//! - Prefill-Decode Separation: Dedicated handling for each phase
8//! - Batch Pipeline: Efficient processing of multiple requests
9
10pub mod chunked_prefill;
11pub mod executor;
12
13pub use chunked_prefill::{ChunkedPrefillConfig, ChunkedPrefillExecutor};
14pub use executor::{ExecutionPhase, PipelineConfig, PipelineExecutor};