dreamwell-intelligence 1.0.0

QuantumGPT (The Loom) — Quantum Information Pretrained Transformer. Density matrix attention with intrinsic thermodynamic loss, φ-scaled causal dephasing, and parameter shift gradient.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33

// dreamwell-intelligence — QIPT: Quantum Information Pretrained Transformer
//
// The Loom: density matrix attention with intrinsic thermodynamic loss,
// φ-scaled causal dephasing, Born rule measurement, and parameter shift gradient.
//
// Clean Compute: all hot-path allocations are pre-sized. Density matrices are
// contiguous Vec<Complex>. No autograd graph. No hidden amortization.
// Every operation is explicit, measurable, and benchmarkable.

pub mod adjoint;
pub mod attention;
pub mod checkpoint;
pub mod complex;
pub mod dataset;
pub mod density_matrix;
pub mod embed;
pub mod fock_forward;
pub mod generate;
pub mod golden_ratio_converter;
pub mod hamiltonian;
pub mod metrics;
pub mod spectral;
pub mod tokenizer;
pub mod train;
pub mod transformer;

pub use complex::Complex;
pub use dataset::TokenDataset;
pub use density_matrix::DensityMatrixN;
pub use golden_ratio_converter::GoldenRatioConverter;
pub use tokenizer::CharTokenizer;
pub use train::TrainConfig;
pub use transformer::{QCTConfig, QCT};