# TrainConfig
base_init:
- normal
- 0.01
batch_size: 64
clip: 1.0
dim_inner: 1000
dim_model: 500
data_dir: data/jsb_chorales_numpy
dropatt: 0.1
dropout: 0.1
embed_init:
- normal
- 0.01
eval_batch_size: 2
eval_interval: 500
eval_mem_len: 512
eval_tgt_len: 128
log_interval: 100
lr: 0.0005
lr_min: 2.5e-05
max_step: 100000
mem_len: 512
num_heads: 10
num_layers: 4
optim: adam
plot_interval: 500
plot_losses_while_training: false
save_all_test_losses: true
save_path: checkpoints/
scheduler: inv_sqrt
seed: 101
tgt_len: 128
tie_embedding: true
warmup_step: 4000
weight_decay: 0.0