|
9 | 9 |
|
10 | 10 | logger = logging.getLogger(__file__) |
11 | 11 |
|
12 | | -list_coyo = [ |
13 | | - "/vlm/data/coyo400m_part1/coyo700m_00", |
14 | | - "/vlm/data/coyo400m_part1/coyo700m_01", |
15 | | - "/vlm/data/coyo400m_part1/coyo700m_02", |
16 | | - "/vlm/data/coyo400m_part1/coyo700m_03", |
17 | | - "/vlm/data/coyo400m_part1/coyo700m_04", |
18 | | - "/vlm/data/coyo400m_part1/coyo700m_05", |
19 | | - "/vlm/data/coyo400m_part1/coyo700m_06", |
20 | | - "/vlm/data/coyo400m_part1/coyo700m_07", |
21 | | - "/vlm/data/coyo400m_part1/coyo700m_08", |
22 | | - "/vlm/data/coyo400m_part1/coyo700m_09", |
23 | | - "/vlm/data/coyo400m_part2/coyo700m_10", |
24 | | - "/vlm/data/coyo400m_part2/coyo700m_11", |
25 | | - "/vlm/data/coyo400m_part2/coyo700m_12", |
26 | | - "/vlm/data/coyo400m_part2/coyo700m_13", |
27 | | - "/vlm/data/coyo400m_part2/coyo700m_14", |
28 | | - "/vlm/data/coyo400m_part2/coyo700m_15", |
29 | | - "/vlm/data/coyo400m_part2/coyo700m_16", |
30 | | - "/vlm/data/coyo400m_part2/coyo700m_17", |
31 | | - "/vlm/data/coyo400m_part2/coyo700m_18", |
32 | | - "/vlm/data/coyo400m_part2/coyo700m_19", |
33 | | - "/vlm/data/coyo400m_part3/coyo700m_20", |
34 | | - "/vlm/data/coyo400m_part3/coyo700m_21", |
35 | | - "/vlm/data/coyo400m_part3/coyo700m_22", |
36 | | - "/vlm/data/coyo400m_part3/coyo700m_24", |
37 | | - "/vlm/data/coyo400m_part3/coyo700m_25", |
38 | | - "/vlm/data/coyo400m_part3/coyo700m_26", |
39 | | - "/vlm/data/coyo400m_part3/coyo700m_27", |
40 | | - "/vlm/data/coyo400m_part3/coyo700m_28", |
41 | | - "/vlm/data/coyo400m_part3/coyo700m_29", |
42 | | - "/vlm/data/coyo400m_part4/coyo700m_30", |
43 | | - "/vlm/data/coyo400m_part4/coyo700m_31", |
44 | | - "/vlm/data/coyo400m_part4/coyo700m_31", |
45 | | -] |
46 | | -list_laion = [ |
47 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-20-tencentos", |
48 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-21-tencentos", |
49 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-23-tencentos", |
50 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-28-tencentos", |
51 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-34-tencentos", |
52 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-58-tencentos", |
53 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-62-tencentos", |
54 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-85-tencentos", |
55 | | -] |
| 12 | +# data paths - configure via environment or replace with your paths |
| 13 | +DATA_ROOT = os.getenv("DATA_ROOT", "/path/to/data") |
| 14 | +list_coyo = [f"{DATA_ROOT}/coyo400m/coyo700m_{i:02d}" for i in range(32)] |
| 15 | +list_laion = [f"{DATA_ROOT}/laion/laion_part_{i:02d}" for i in range(8)] |
56 | 16 |
|
57 | 17 |
|
58 | 18 | @DATASET_REGISTRY.register() |
59 | 19 | def llava_vit_si_2025_12_12(): |
60 | | - list_coyo = [ |
61 | | - "/vlm/data/coyo400m_part1/coyo700m_00", |
62 | | - "/vlm/data/coyo400m_part1/coyo700m_01", |
63 | | - "/vlm/data/coyo400m_part1/coyo700m_02", |
64 | | - "/vlm/data/coyo400m_part1/coyo700m_03", |
65 | | - "/vlm/data/coyo400m_part1/coyo700m_04", |
66 | | - "/vlm/data/coyo400m_part1/coyo700m_05", |
67 | | - "/vlm/data/coyo400m_part1/coyo700m_06", |
68 | | - "/vlm/data/coyo400m_part1/coyo700m_07", |
69 | | - "/vlm/data/coyo400m_part1/coyo700m_08", |
70 | | - "/vlm/data/coyo400m_part1/coyo700m_09", |
71 | | - "/vlm/data/coyo400m_part2/coyo700m_10", |
72 | | - "/vlm/data/coyo400m_part2/coyo700m_11", |
73 | | - "/vlm/data/coyo400m_part2/coyo700m_12", |
74 | | - "/vlm/data/coyo400m_part2/coyo700m_13", |
75 | | - "/vlm/data/coyo400m_part2/coyo700m_14", |
76 | | - "/vlm/data/coyo400m_part2/coyo700m_15", |
77 | | - "/vlm/data/coyo400m_part2/coyo700m_16", |
78 | | - "/vlm/data/coyo400m_part2/coyo700m_17", |
79 | | - "/vlm/data/coyo400m_part2/coyo700m_18", |
80 | | - "/vlm/data/coyo400m_part2/coyo700m_19", |
81 | | - "/vlm/data/coyo400m_part3/coyo700m_20", |
82 | | - "/vlm/data/coyo400m_part3/coyo700m_21", |
83 | | - "/vlm/data/coyo400m_part3/coyo700m_22", |
84 | | - "/vlm/data/coyo400m_part3/coyo700m_24", |
85 | | - "/vlm/data/coyo400m_part3/coyo700m_25", |
86 | | - "/vlm/data/coyo400m_part3/coyo700m_26", |
87 | | - "/vlm/data/coyo400m_part3/coyo700m_27", |
88 | | - "/vlm/data/coyo400m_part3/coyo700m_28", |
89 | | - "/vlm/data/coyo400m_part3/coyo700m_29", |
90 | | - "/vlm/data/coyo400m_part4/coyo700m_30", |
91 | | - "/vlm/data/coyo400m_part4/coyo700m_31", |
92 | | - "/vlm/data/coyo400m_part4/coyo700m_31", |
93 | | - ] |
94 | | - list_laion = [ |
95 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-20-tencentos", |
96 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-21-tencentos", |
97 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-23-tencentos", |
98 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-28-tencentos", |
99 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-34-tencentos", |
100 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-58-tencentos", |
101 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-62-tencentos", |
102 | | - "/vlm/data/LAION224M_HOI31M_IN13M_labeled_2024_03_05/LAION224M_HOI31M_IN13M_labeled_2024_03_05_VM-2-85-tencentos", |
103 | | - ] |
| 20 | + # use module-level list_coyo and list_laion |
104 | 21 | rank = int(os.getenv("RANK", "0")) |
105 | 22 | world_size = int(os.getenv("WORLD_SIZE", "1")) # Expected to be 128 |
106 | 23 |
|
|
0 commit comments