Skip to main content

Mojo trait

ScoreModTrait

The ScoreMod trait desctribes score_mod for mha kernel like alibi bias.

Implemented traits

AnyType, Copyable, DevicePassable, UnknownDestructibility

Aliases

__copyinit__is_trivial

alias __copyinit__is_trivial

A flag (often compiler generated) to indicate whether the implementation of __copyinit__ is trivial.

The implementation of __copyinit__ is considered to be trivial if:

  • The struct has a compiler-generated trivial __copyinit__ and all its fields have a trivial __copyinit__ method.

In practice, it means the value can be copied by copying the bits from one location to another without side effects.

__del__is_trivial

alias __del__is_trivial

A flag (often compiler generated) to indicate whether the implementation of __del__ is trivial.

The implementation of __del__ is considered to be trivial if:

  • The struct has a compiler-generated trivial destructor and all its fields have a trivial __del__ method.

In practice, it means that the __del__ can be considered as no-op.

device_type

alias device_type

Indicate the type being used on accelerator devices.

name_str

alias name_str

Required methods

__copyinit__

__copyinit__(out self: _Self, existing: _Self, /)

Create a new instance of the value by copying an existing one.

Args:

  • existing (_Self): The value to copy.

Returns:

_Self

score_mod

score_mod[dtype: DType, width: Int, //, *, element_type: DType = DType.int32](self: _Self, coord: IndexList[4, element_type=element_type], score_vec: SIMD[dtype, width], max_prompt_len: Int = 0) -> SIMD[dtype, width]

Return score vector at given coordinates given a score_mod.

Arguments: coord is (seq_id, head, q_idx, k_idx) score_vec is at coord of the score matrix

Score_mod calculates a tensor given the functor and adds to score_vec.

Returns:

SIMD

get_type_name

static get_type_name() -> String

Gets the name of the host type (the one implementing this trait). For example, Int would return "Int", DeviceBuffer[DType.float32] would return "DeviceBuffer[DType.float32]". This is used for error messages when passing types to the device. TODO: This method will be retired soon when better kernel call error messages arrive.

Returns:

String: The host type's name.

get_device_type_name

static get_device_type_name() -> String

Gets device_type's name. For example, because DeviceBuffer's device_type is UnsafePointer, DeviceBuffer[DType.float32]'s get_device_type_name() should return something like "UnsafePointer[Scalar[DType.float32]]". This is used for error messages when passing types to the device. TODO: This method will be retired soon when better kernel call error messages arrive.

Returns:

String: The device type's name.

Provided methods

copy

copy(self: _Self) -> _Self

Explicitly construct a copy of self.

Returns:

_Self: A copy of this value.

Was this page helpful?