Hugging Face's logo

uav-vts
/
training_sem

Model card Files
xet
Community
1
training_sem
File size: 1,285 Bytes 
cb0ad2d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
 
import os
import torch
from libs.utils.vocab import Vocab

device = torch.device('cuda')

train_lrcs_path = [
    "/yrfs1/intern/pfhu6/TSR/Dataset/SciTSR/train/table.lrc"
]
train_data_dir = ''
train_max_pixel_nums = 400 * 400 * 5
train_bucket_seps = (50, 50, 50)
train_max_batch_size = 8
train_num_workers = 4

valid_lrc_path = '/yrfs1/intern/pfhu6/TSR/Dataset/SciTSR/test/table.lrc'
valid_data_dir = ''
valid_num_workers = 0
valid_batch_size = 1

vocab = Vocab()

# model params
# backbone
arch = "res34"
pretrained_backbone = True
backbone_out_channels = (64, 128, 256, 512)

# fpn
fpn_out_channels = 256

# pan
pan_num_levels = 4
pan_in_dim = 256
pan_out_dim = 256

# row segment predictor
rs_scale = 1

# col segment predictor
cs_scale = 1

# divide predictor
dp_head_nums = 8
dp_scale = 1

# cells extractor params
ce_scale = 1 / 8
ce_pool_size = (3, 3)
ce_dim = 512
ce_head_nums = 8
ce_heads = 1

# decoder
embed_dim = 512
feat_dim = 512
lm_state_dim = 512
proj_dim = 512
cover_kernel = 7
att_threshold = 0.5
spatial_att_weight_loss_wight = 1.0

# train params
base_lr = 0.0001
min_lr = 1e-6
weight_decay = 0

num_epochs = 20
sync_rate = 20

log_sep = 20

work_dir = './experiments/heads_1'

train_checkpoint = None

eval_checkpoint = os.path.join(work_dir, 'best_f1_model.pth')