SyncTree/parameters_sync.yaml at main · xlab-ub/SyncTree · GitHub

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
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
# General Params
wandb: 0                                         # use wandb track model training, 1:enable, 0:disable
exp: Sync_Delay_GCN_32                            # format: data_solver_model_nconvs
elmore_delay: 0                                  # show elmore delay results, 1:enable, 0:disable
solver: Delay                                    # Delay, Slew
model: GCN                                      # TREE, GCN, GAT, SAGE, DeepGCN, GraphTrans, NT, Transfer
batch_size: 32                                   # 32:v9
lr: 0.0002
seed: 1234
sh_gamma: 0.95                                   # lr_scheduler gamma
lamda: 0.001                                     # balance loss-l1 and loss_TC
epochs: 5
max_len: 52                                      # padding length 21:v10, 52:v9
leaf_num: 18                                     # max leafs 11:v10, 18:v9
save_pth: "trained_models/"
data_pth: "traindata/sync/dir/delay/graphs"        # dir: TREE+Transfer, undir: Other Models
jth_pth: "traindata/sync/jths/graphs"
csv_pth: "csvs/"
loss_type: 0                                     # 0:L1, 1:L1+TC, 2:L1+C
train_cut: 1                                     # measure data efficiency,[0.25, 0.75, 1]


# Pos Neg Samples Params
B: 64                                            # number of target leaf nodes samples in every batch
M: 1                                             # number of neg samples
N: 1                                             # number of pos samples
dist: 2

# TreeCIR Loss Params
m: 1
gamma: 1

# TREE & GAT & GCN & SAGE & DeepGCN Params
conv_in_dim: 1
conv_hi_dim: 32
conv_out_dim: 1
edge_dim: 1
n_convs: 32

# MLP Params
mlp_in_dim: 2
mlp_hi_dim: 32
mlp_out_dim: 1
n_mlps: 3

# Transfer Learning Params
transfer_gnn_model_pth: "trained_models/Sync_Delay_TREE_32"
transfer_n_mlps: 3
transfer_mlp_in_dim: 128
transfer_mlp_hi_dim: 64
transfer_mlp_out_dim: 1


# wandb setting
wandb_proj: "Sync-Delay"

# GraphTrans Params
dim_feedforward: 256
num_encoder_layers: 3
scheduler: cosine
pos_encoder: False
nhead: 4
transformer_activation: "relu"
transformer_norm_input: False
num_encoder_layers_masked: 0
transformer_prenorm: False
transformer_dropout: 0.1
gnn_dropout: 0.1
gnn_emb_dim: 128
gnn_num_layer: 16
gnn_type: gcn
d_model: 32
gnn_JK: "last"
gnn_residual: False
graph_pooling: 'mean'
num_features: 1
max_input_len: 60
max_seq_len: 60
trans_mlp_in_dim: 64
trans_mlp_hi_dim: 32
trans_mlp_out_dim: 1
trans_n_mlps: 3

# Neural Tree Params
radius: 50
min_diameter: 1
max_diameter: 1
input_dim: 1
output_dim: 1
conv_block: "GCN"
hidden_dim: 32
num_layers: 16
GAT_hidden_dims: [8, 8]
GAT_heads: [8, 1]
dropout: 0.1
GAT_concats: [True, False]