Merge commit 'd803bfe2b1fe7f5e219e50ac20d6801a0a58ac75' as 'vendor/ruvector'

vendor/ruvector/crates/ruvllm/src/quantize/mod.rs

@@ -0,0 +1,68 @@
+//! Quantization Pipeline for RuvLTRA Models
+//!
+//! This module provides quantization capabilities for converting full-precision
+//! models to optimized quantized formats suitable for edge inference on Apple Silicon.
+//!
+//! ## Supported Quantization Formats
+//!
+//! | Format | Bits | Memory (0.5B) | Quality | Use Case |
+//! |--------|------|---------------|---------|----------|
+//! | Q4_K_M | 4.5  | ~300 MB       | Good    | Best quality/size tradeoff |
+//! | Q5_K_M | 5.5  | ~375 MB       | Better  | Higher quality, still compact |
+//! | Q8_0   | 8.5  | ~500 MB       | Best    | Near-lossless quantization |
+//!
+//! ## Apple Neural Engine (ANE) Optimization
+//!
+//! The quantization pipeline produces weights optimized for ANE inference:
+//! - 16-byte aligned weight layouts
+//! - Blocked quantization compatible with ANE tile operations
+//! - Optimized memory access patterns for M4 Pro's unified memory
+//!
+//! ## Example
+//!
+//! ```rust,ignore
+//! use ruvllm::quantize::{RuvltraQuantizer, QuantConfig, TargetFormat};
+//! use std::path::Path;
+//!
+//! // Create quantizer for Q4_K_M format
+//! let config = QuantConfig::default()
+//!     .with_format(TargetFormat::Q4_K_M)
+//!     .with_ane_optimization(true);
+//!
+//! let quantizer = RuvltraQuantizer::new(config)?;
+//!
+//! // Quantize a model
+//! quantizer.quantize_model(
+//!     Path::new("qwen-0.5b.safetensors"),
+//!     Path::new("ruvltra-small-q4.gguf"),
+//! )?;
+//! ```
+
+mod ruvltra_quant;
+
+pub use ruvltra_quant::{
+    dequantize_for_ane,
+
+    // Memory estimation
+    estimate_memory_q4,
+    estimate_memory_q5,
+    estimate_memory_q8,
+    // Quantization functions
+    quantize_ruvltra_q4,
+    quantize_ruvltra_q5,
+    quantize_ruvltra_q8,
+    MemoryEstimate,
+
+    // Block types
+    Q4KMBlock,
+    Q5KMBlock,
+    Q8Block,
+
+    QuantConfig,
+    // Progress tracking
+    QuantProgress,
+    QuantStats,
+    // Core quantizer
+    RuvltraQuantizer,
+    TargetFormat,
+};