Add default training configuration
Browse files- configs/default.yaml +56 -0
configs/default.yaml
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# NeuroName Default Training Configuration
|
| 2 |
+
# ==========================================
|
| 3 |
+
|
| 4 |
+
# Model Architecture
|
| 5 |
+
d_model: 256
|
| 6 |
+
d_latent: 128
|
| 7 |
+
d_semantic: 128
|
| 8 |
+
d_control: 64
|
| 9 |
+
nhead: 4
|
| 10 |
+
enc_layers: 4
|
| 11 |
+
dec_layers: 6
|
| 12 |
+
max_len: 32
|
| 13 |
+
memory_slots: 8
|
| 14 |
+
char_vocab_size: 44
|
| 15 |
+
semantic_vocab_size: 10000
|
| 16 |
+
num_styles: 8
|
| 17 |
+
num_languages: 8
|
| 18 |
+
dropout: 0.1
|
| 19 |
+
|
| 20 |
+
# Training
|
| 21 |
+
learning_rate: 0.0003
|
| 22 |
+
weight_decay: 0.01
|
| 23 |
+
batch_size: 64
|
| 24 |
+
num_epochs: 100
|
| 25 |
+
warmup_steps: 1000
|
| 26 |
+
max_grad_norm: 1.0
|
| 27 |
+
|
| 28 |
+
# Loss Weights
|
| 29 |
+
kl_weight_start: 0.0
|
| 30 |
+
kl_weight_end: 0.1
|
| 31 |
+
kl_anneal_steps: 5000
|
| 32 |
+
kl_free_bits: 0.5
|
| 33 |
+
style_loss_weight: 0.3
|
| 34 |
+
lang_loss_weight: 0.3
|
| 35 |
+
phon_loss_weight: 0.2
|
| 36 |
+
|
| 37 |
+
# Phonotactic Discriminator
|
| 38 |
+
phon_lr: 0.001
|
| 39 |
+
phon_train_every: 5
|
| 40 |
+
|
| 41 |
+
# Generation
|
| 42 |
+
temperature: 0.8
|
| 43 |
+
top_k: 15
|
| 44 |
+
top_p: 0.92
|
| 45 |
+
repetition_penalty: 1.2
|
| 46 |
+
|
| 47 |
+
# Latent Space Control
|
| 48 |
+
guidance_method: "langevin"
|
| 49 |
+
guidance_steps: 50
|
| 50 |
+
guidance_step_size: 0.01
|
| 51 |
+
guidance_noise: 0.005
|
| 52 |
+
|
| 53 |
+
# Data
|
| 54 |
+
train_data_path: "data/names_train.txt"
|
| 55 |
+
val_data_path: "data/names_val.txt"
|
| 56 |
+
semantic_vocab_path: "data/semantic_vocab.json"
|