Index

B

backend (transformer_implementation.TransformerConfig.TransformerConfig attribute) batch_size (transformer_implementation.TransformerConfig.TransformerConfig attribute) beta1 (transformer_implementation.TransformerConfig.TransformerConfig attribute)

beta2 (transformer_implementation.TransformerConfig.TransformerConfig attribute) bias (transformer_implementation.TransformerConfig.TransformerConfig attribute) block_size (transformer_implementation.TransformerConfig.TransformerConfig attribute) BOS_IDX (transformer_implementation.TransformerConfig.TransformerConfig attribute)

C

compile (transformer_implementation.TransformerConfig.TransformerConfig attribute)

configure_optimizers() (transformer_implementation.Transformer.Transformer method)

D

DataLoaderFactory (class in transformer_implementation.DataLoaderFactory) ddp (transformer_implementation.TransformerConfig.TransformerConfig attribute) Decoder (class in transformer_implementation.Decoder) DecoderBlock (class in transformer_implementation.blocks.DecoderBlock)

device (transformer_implementation.TransformerConfig.TransformerConfig attribute) device_type (transformer_implementation.TransformerConfig.TransformerConfig attribute) dropout (transformer_implementation.TransformerConfig.TransformerConfig attribute) dtype (transformer_implementation.TransformerConfig.TransformerConfig attribute)

E

Encoder (class in transformer_implementation.Encoder) EncoderBlock (class in transformer_implementation.blocks.EncoderBlock) EOS_IDX (transformer_implementation.TransformerConfig.TransformerConfig attribute)

eps (transformer_implementation.TransformerConfig.TransformerConfig attribute) estimate_loss() (in module utils.estimate_loss) eval_iters (transformer_implementation.TransformerConfig.TransformerConfig attribute)

F

FeedForward (class in transformer_implementation.blocks.layers.FeedForward) forward() (transformer_implementation.blocks.DecoderBlock.DecoderBlock method) (transformer_implementation.blocks.EncoderBlock.EncoderBlock method) (transformer_implementation.blocks.layers.FeedForward.FeedForward method) (transformer_implementation.blocks.layers.LayerNorm.LayerNorm method) (transformer_implementation.blocks.layers.MultiHeadAttention.MultiHeadAttention method) (transformer_implementation.Decoder.Decoder method) (transformer_implementation.Encoder.Encoder method) (transformer_implementation.Transformer.Transformer method)

G

generate() (transformer_implementation.Transformer.Transformer method) generate_padding_mask() (transformer_implementation.Tokenizer.Tokenizer method) get_batch() (transformer_implementation.DataLoaderFactory.DataLoaderFactory method)

get_linear_schedule_with_warmup() (in module utils.training_loop) get_num_params() (transformer_implementation.Decoder.Decoder method) (transformer_implementation.Encoder.Encoder method) grad_accumulation_steps (transformer_implementation.TransformerConfig.TransformerConfig attribute)

L

LayerNorm (class in transformer_implementation.blocks.layers.LayerNorm)

learning_rate (transformer_implementation.TransformerConfig.TransformerConfig attribute) load_model() (transformer_implementation.Transformer.Transformer method)

M

max_epochs (transformer_implementation.TransformerConfig.TransformerConfig attribute) max_iters (transformer_implementation.TransformerConfig.TransformerConfig attribute) module transformer_implementation transformer_implementation.blocks transformer_implementation.blocks.DecoderBlock transformer_implementation.blocks.EncoderBlock transformer_implementation.blocks.layers transformer_implementation.blocks.layers.FeedForward transformer_implementation.blocks.layers.LayerNorm transformer_implementation.blocks.layers.MultiHeadAttention transformer_implementation.DataLoaderFactory transformer_implementation.Decoder transformer_implementation.Encoder transformer_implementation.Tokenizer transformer_implementation.Transformer transformer_implementation.TransformerConfig utils utils.estimate_loss utils.plot_losses utils.training_loop

MultiHeadAttention (class in transformer_implementation.blocks.layers.MultiHeadAttention)

N

n_embd (transformer_implementation.TransformerConfig.TransformerConfig attribute)

n_head (transformer_implementation.TransformerConfig.TransformerConfig attribute) n_layer (transformer_implementation.TransformerConfig.TransformerConfig attribute)

P

PAD_IDX (transformer_implementation.TransformerConfig.TransformerConfig attribute)

plot_losses() (in module utils.plot_losses)

S

save_model() (transformer_implementation.Transformer.Transformer method) scaled_dot_product_attention() (transformer_implementation.blocks.layers.MultiHeadAttention.MultiHeadAttention method)

sequence_cleaner() (transformer_implementation.Tokenizer.Tokenizer method) sequence_padding() (transformer_implementation.Tokenizer.Tokenizer method)

T

tokenize() (transformer_implementation.Tokenizer.Tokenizer method) tokenize_from_str() (transformer_implementation.Tokenizer.Tokenizer method) Tokenizer (class in transformer_implementation.Tokenizer) train_data_size (transformer_implementation.TransformerConfig.TransformerConfig attribute) training_loop() (in module utils.training_loop) Transformer (class in transformer_implementation.Transformer) transformer_implementation module transformer_implementation.blocks module transformer_implementation.blocks.DecoderBlock module transformer_implementation.blocks.EncoderBlock module transformer_implementation.blocks.layers module transformer_implementation.blocks.layers.FeedForward module

transformer_implementation.blocks.layers.LayerNorm module transformer_implementation.blocks.layers.MultiHeadAttention module transformer_implementation.DataLoaderFactory module transformer_implementation.Decoder module transformer_implementation.Encoder module transformer_implementation.Tokenizer module transformer_implementation.Transformer module transformer_implementation.TransformerConfig module TransformerConfig (class in transformer_implementation.TransformerConfig) TranslationDataset (class in transformer_implementation.DataLoaderFactory)

U

utils module utils.estimate_loss module

utils.plot_losses module utils.training_loop module

V

vocab_size (transformer_implementation.TransformerConfig.TransformerConfig attribute)

vocab_size() (transformer_implementation.Tokenizer.Tokenizer method)

W

weight_decay (transformer_implementation.TransformerConfig.TransformerConfig attribute)