Crate boostr 

boostr

ML framework built on numr — attention, quantization, model architectures.

boostr extends numr’s foundational numerical computing with ML-specific operations, quantized tensor support, and model building blocks. It uses numr’s runtime, tensors, and ops directly — no reimplementation, no wrappers.

Relationship to numr

┌─────────────────────────────────────────────────────────┐
│                    boostr ◄── YOU ARE HERE               │
│   (attention, RoPE, MoE, quantization, model loaders)   │
└──────────────────────────┬──────────────────────────────┘
│                      numr                                │
│     (tensors, ops, runtime, autograd, linalg, FFT)       │
└─────────────────────────────────────────────────────────┘

Design

• Extension traits: ML ops (AttentionOps, RoPEOps) implemented on numr’s clients
• QuantTensor: Separate type for block-quantized data (GGUF formats)
• impl_generic: Composite ops composed from numr primitives, same on all backends
• Custom kernels: Dequant, quantized matmul, fused attention (SIMD/PTX/WGSL)

Re-exports

pub use nn::Init;

pub use nn::VarBuilder;

pub use nn::VarMap;

pub use nn::Weight;

pub use nn::router_z_loss;

pub use ops::AttentionOps;

pub use ops::AttnOutLayout;

pub use ops::DeviceGrammarDfa;

pub use ops::FlashAttentionOps;

pub use ops::FusedFp8TrainingOps;

pub use ops::FusedOptimizerOps;

pub use ops::FusedQkvOps;

pub use ops::GatedDeltaNetOps;

pub use ops::GrammarDfaOps;

pub use ops::KvCacheOps;

pub use ops::KvSeqLen;

pub use ops::MlaOps;

pub use ops::PagedAttentionOps;

pub use ops::RoPEOps;

pub use ops::RoPEPackedOps;

pub use ops::SamplingOps;

pub use ops::SamplingScratch;

pub use ops::var_flash_attention;

pub use quant::DecomposedQuantLinear;

pub use quant::DecomposedQuantMethod;

pub use quant::DecomposedQuantTensor;

pub use quant::DequantOps;

pub use quant::FusedQuantOps;

pub use quant::QuantFormat;

pub use quant::QuantMatmulOps;

pub use quant::QuantTensor;

pub use quant::QuantizeOps;

pub use quant::ScheduleTuning;

pub use readback::scalar_f32;

pub use model::ExpertWeights;

pub use model::encoder::EmbeddingPipeline;

pub use model::encoder::Encoder;

pub use model::encoder::EncoderClient;

pub use model::encoder::EncoderConfig;

pub use model::encoder::Pooling;

Modules

autograd

Automatic differentiation (autograd)

data

distributed

error

boostr error types

format

inference

model

nn

ops

optimizer

quant

readback

runtime

Runtime backends for tensor computation

tcf

TCF: GGUF’s block-quantized payloads in a strict, self-describing container. See FORMAT.md beside this module for the binary layout.

tensor

Tensor types and operations

trainer

Macros

define_enum_u16

Generates a fallible u16-backed enum: exhaustive, a to_u16 encoder, and a TryFrom<u16> decoder that rejects every value not listed (E_UNKNOWN_ENUM_VALUE, naming $field — the wire field name, not the Rust type name). No catch-all variant: an unrecognized value is always a decode error, never a value the type can represent.

define_enum_u32

Generates a fallible u32-backed enum. See define_enum_u16!.

define_record

Generate a TCF v1 record from its spec offset table. See the module docs.

Structs

CpuClient

CPU client for operation dispatch

CpuDevice

CPU device (there’s only one: the host CPU)

CpuRuntime

CPU compute runtime

Tensor

N-dimensional array stored on a compute device

Enums

DType

Data types supported by numr tensors

NumrError

Errors that can occur in numr operations

PaddingMode

Padding mode for convolution operations.

Traits

ActivationOps

Activation operations

BinaryOps

Element-wise binary operations on tensors.

CompareOps

Comparison operations trait.

ConditionalOps

Conditional operations

ConvOps

Convolution operations.

HostCopyOps

Writing host data straight into a tensor the caller already owns.

IndexingOps

Indexing operations

LinalgOps

Linear algebra operations trait

MatmulOps

Matrix multiplication operations

NormalizationOps

Normalization operations

RandomOps

Random number generation operations

ReduceOps

Reduction operations

Runtime

Core trait for compute backends

RuntimeClient

Trait for runtime clients that handle operation dispatch

ScalarOps

Scalar operations trait for tensor-scalar operations

ShapeOps

Shape manipulation operations

SortingOps

Sorting and search operations trait

TensorOps

Core tensor operations trait

TypeConversionOps

Type conversion operations

UnaryOps

Unary operations

Type Aliases

NumrResult

Result type alias using numr’s Error