Uses of Interface
io.github.kirstenali.deepj.training.Trainable

Packages that use Trainable

Package	Description
io.github.kirstenali.deepj.layers
io.github.kirstenali.deepj.layers.transformer
io.github.kirstenali.deepj.layers.transformer.attention
io.github.kirstenali.deepj.layers.transformer.blocks
io.github.kirstenali.deepj.layers.transformer.norm
io.github.kirstenali.deepj.models
io.github.kirstenali.deepj.models.deepseek
io.github.kirstenali.deepj.models.gpt
io.github.kirstenali.deepj.models.llama
io.github.kirstenali.deepj.transformer
io.github.kirstenali.deepj.transformer.embeddings

Uses of Trainable in io.github.kirstenali.deepj.layers

Subinterfaces of Trainable in io.github.kirstenali.deepj.layers

Modifier and Type	Interface	Description
interface	Layer	Differentiable module mapping Tensor -> Tensor.
interface	Projection	Marker interface for linear projection layers (fully-connected, no activation).

Classes in io.github.kirstenali.deepj.layers that implement Trainable

Modifier and Type	Class	Description
final class	FNN	Flexible fully-connected neural network (MLP) built from Linear projections.
final class	Linear	Fully-connected layer: y = xW + b x: [n x dIn], W: [dIn x dOut], b: [1 x dOut]

Uses of Trainable in io.github.kirstenali.deepj.layers.transformer

Classes in io.github.kirstenali.deepj.layers.transformer that implement Trainable

Modifier and Type	Class	Description
final class	SwiGLULayer	SwiGLU feed-forward layer — the FFN used in Llama, Mistral, Qwen, and DeepSeek.

Uses of Trainable in io.github.kirstenali.deepj.layers.transformer.attention

Classes in io.github.kirstenali.deepj.layers.transformer.attention that implement Trainable

Modifier and Type	Class	Description
final class	MultiHeadLatentAttention	Compact Multi-Head Latent Attention (MLA) inspired by DeepSeek-V2/V3.
class	MultiHeadSelfAttention	Multi-head causal self-attention for a single sequence (no batch dimension).
final class	RoPEMultiHeadSelfAttention	Multi-head self-attention with Rotary Positional Embedding (RoPE).

Uses of Trainable in io.github.kirstenali.deepj.layers.transformer.blocks

Classes in io.github.kirstenali.deepj.layers.transformer.blocks that implement Trainable

Modifier and Type	Class	Description
final class	DeepSeekTransformerBlock	Pre-LN transformer block with DeepSeek-style latent attention:
class	GPTTransformerBlock	Pre-LN GPT-style Transformer block:
final class	LlamaTransformerBlock	Pre-LN transformer block with Llama-style full-head attention:

Uses of Trainable in io.github.kirstenali.deepj.layers.transformer.norm

Subinterfaces of Trainable in io.github.kirstenali.deepj.layers.transformer.norm

Modifier and Type	Interface	Description
interface	NormLayer	Marker interface for row-wise normalisation layers (LayerNorm, RMSNorm, …).

Classes in io.github.kirstenali.deepj.layers.transformer.norm that implement Trainable

Modifier and Type	Class	Description
final class	LayerNorm1D	LayerNorm over feature dimension (cols) with trainable gamma/beta exposed as Parameters.
final class	RMSNorm1D	Root-Mean-Square Layer Normalisation — the normalisation used in Llama, Mistral, Qwen, and DeepSeek.

Uses of Trainable in io.github.kirstenali.deepj.models

Subinterfaces of Trainable in io.github.kirstenali.deepj.models

Modifier and Type	Interface	Description
interface	CausalLM	Common contract for decoder-only causal language models (GPT, Llama, DeepSeek, …).

Classes in io.github.kirstenali.deepj.models that implement Trainable

Modifier and Type	Class	Description
class	DecoderOnlyModel	Shared skeleton for decoder-only transformer models: token embedding → block stack → final norm → LM-head.

Uses of Trainable in io.github.kirstenali.deepj.models.deepseek

Classes in io.github.kirstenali.deepj.models.deepseek that implement Trainable

Modifier and Type	Class	Description
final class	DeepSeekModel	DeepSeek-style decoder-only transformer.

Uses of Trainable in io.github.kirstenali.deepj.models.gpt

Classes in io.github.kirstenali.deepj.models.gpt that implement Trainable

Modifier and Type	Class	Description
final class	GPTModel	Minimal GPT-style decoder-only transformer for educational/training use.

Uses of Trainable in io.github.kirstenali.deepj.models.llama

Classes in io.github.kirstenali.deepj.models.llama that implement Trainable

Modifier and Type	Class	Description
final class	LlamaModel	Llama-style decoder-only transformer.

Uses of Trainable in io.github.kirstenali.deepj.transformer

Classes in io.github.kirstenali.deepj.transformer that implement Trainable

Modifier and Type	Class	Description
final record	TransformerStack	A sequential stack of transformer blocks.

Uses of Trainable in io.github.kirstenali.deepj.transformer.embeddings

Classes in io.github.kirstenali.deepj.transformer.embeddings that implement Trainable

Modifier and Type	Class	Description
final class	Embedding	Token embedding: ids -> vectors.
final class	PositionalEmbedding	Learnable positional embeddings added to token embeddings.