-
Notifications
You must be signed in to change notification settings - Fork 0
/
config.yaml
26 lines (24 loc) · 1.39 KB
/
config.yaml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
learning_rate: 0.001 # learning rate for model
dropout_rate: 0.1 # dropout rate for model
lambda_for_cl: 0.1 # control loss for encoder model(range: 0 ~ 1, allow float type)
substitution_rate: 0.3 # ratio for item substitution in batch
masking_rate: 0.3 # ratio for item masking in batch
cropping_rate: 0.3 # ratio for item cropping in batch
batch_size: 256
sequence_len: 1 # if run_mode == 'diversity' you should use 1, else using 30 will be good
embedding_dim: 128 # embedding size
early_stop_patience: 20 # control early stopping steps
warmup_steps: 10000 # control warmup steps for learning rate scheduling
epochs: 200
clip_norm: 5.0 # gradient clipping norm
log_interval: 10 # interval for logging for loguru and wandb
temperature: 0.1 # temperature for NT-Xent Loss
try_num: '9' # experiment number
run_mode: 'test' # running mode for main.py: ['train', 'test', 'diversity']
data_type: 'RentTheRunway' # 'RentTheRunway'
train_type: 'NCF' # Model which you want to train: ['CL', 'REC', 'ONLY_REC', 'NCF']
aug_mode: 'substitute' # Item augmentaion mode: ['substitute', 'crop', 'mask']
rec_num_dense_layer: 2
model_trainable: False # True, False
k: 20 # top-K item
sparsity: 0.5 # transaction count sparsity for checking diversity