oa::ByteEmbedding

BYTE EMBEDDING - The 256-entry lookup table Maps each byte value (0-255) to a d_model dimensional vector. input: [batch, seq_len] of UInt8 output: [batch, seq_len, d_model] of Float32 This replaces the massive 30k-100k token embedding tables in GPT/LLaMA. Ours is always 256 x d_model. Tiny. Fast. Universal.

Inheritance

public Module

Public Methods

oa::I32 oa::ByteEmbedding::dModel()
Matrix oa::ByteEmbedding::forward(const Matrix & inByteIds)

Constructor & Destructor Documentation

oa::ByteEmbedding::ByteEmbedding( oa::I32 inDModel )
No public source comment is attached to this declaration.

Parameters

inDModel
oa::I32

Public Method Documentation

oa::I32 oa::ByteEmbedding::dModel()
No public source comment is attached to this declaration.

Returns

oa::I32

The declared return value.

Matrix oa::ByteEmbedding::forward( const Matrix & inByteIds )
[batch, seq] -> [batch, seq, d_model]

Parameters

inByteIds
const Matrix &

Returns

Matrix

The declared return value.