fast.h

Go to the documentation of this file.

// Copyright © 2023-2024 Apple Inc.
 
#pragma once
 
#include <optional>
 
#include "mlx/utils.h"
 
namespace mlx::core::fast {
 
array rms_norm(
    const array& x,
    const array& weight,
    float eps,
    StreamOrDevice s = {});
 
array layer_norm(
    const array& x,
    const std::optional<array>& weight,
    const std::optional<array>& bias,
    float eps,
    StreamOrDevice s = {});
 
array rope(
    const array& x,
    int dims,
    bool traditional,
    float base,
    float scale,
    int offset,
    StreamOrDevice s = {});
 
array scaled_dot_product_attention(
    const array& queries,
    const array& keys,
    const array& values,
    const float scale,
    const std::optional<array>& mask = std::nullopt,
    StreamOrDevice s = {});
 
std::tuple<array, array, array> affine_quantize(
    const array& w,
    int group_size = 64,
    int bits = 4,
    StreamOrDevice s = {});
 
array affine_quantize(
    const array& w,
    const array& scales,
    const array& biases,
    int group_size = 64,
    int bits = 4,
    StreamOrDevice s = {});
 
array affine_dequantize(
    const array& w,
    const array& scales,
    const array& biases,
    int group_size = 64,
    int bits = 4,
    StreamOrDevice s = {});
 
} // namespace mlx::core::fast