forked from facebookresearch/jepa
-
Notifications
You must be signed in to change notification settings - Fork 0
/
vitl16.yaml
90 lines (90 loc) · 1.58 KB
/
vitl16.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
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
app: vjepa
nodes: 16
tasks_per_node: 8
data:
dataset_type: VideoDataset
datasets:
- /your_path_to_kinetics710_csv_file_index.csv
- /your_path_to_ssv2_csv_file_index.csv
- /your_path_to_howto100m_csv_file_index.csv
decode_one_clip: true
batch_size: 24
num_clips: 1
num_frames: 16
tubelet_size: 2
sampling_rate: 4
crop_size: 224
patch_size: 16
pin_mem: true
num_workers: 12
filter_short_videos: false
clip_duration: null
data_aug:
auto_augment: false
motion_shift: false
random_resize_aspect_ratio:
- 0.75
- 1.35
random_resize_scale:
- 0.3
- 1.0
reprob: 0.0
logging:
folder: /your_absolute_file_path_for_saving_logs_and_checkpoints/
write_tag: jepa
loss:
loss_exp: 1.0
reg_coeff: 0.0
mask:
- aspect_ratio:
- 0.75
- 1.5
num_blocks: 8
spatial_scale:
- 0.15
- 0.15
temporal_scale:
- 1.0
- 1.0
max_temporal_keep: 1.0
max_keep: null
- aspect_ratio:
- 0.75
- 1.5
num_blocks: 2
spatial_scale:
- 0.7
- 0.7
temporal_scale:
- 1.0
- 1.0
max_temporal_keep: 1.0
max_keep: null
meta:
load_checkpoint: false
read_checkpoint: null
seed: 234
eval_freq: 100
use_sdpa: true
dtype: bfloat16
model:
model_name: vit_large
pred_depth: 12
pred_embed_dim: 384
uniform_power: true
use_mask_tokens: true
zero_init_mask_tokens: true
optimization:
ipe: 300
ipe_scale: 1.25
clip_grad: 10.0
weight_decay: 0.04
final_weight_decay: 0.4
epochs: 300
warmup: 40
start_lr: 0.0002
lr: 0.000625
final_lr: 1.0e-06
ema:
- 0.998
- 1.0