Module binary 

Binary: 1-bit quantization for maximum compression

This module implements binary (1-bit) quantization where each value is represented by a single bit based on its sign relative to the mean. This provides 32x compression and is particularly effective for similarity search tasks.

Functions

dequantize_binary

Dequantizes a binary quantized vector back to f32

quantize_binary

Quantizes an embedding vector to binary (1-bit per value)