Hugging Face's logo

younggi
/
videomae-base-finetuned-ucf101-subset

Video Classification
Transformers
PyTorch
TensorBoard
videomae
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
younggi commited on
Commit
b5d59c4
·
1 Parent(s): ad73c25

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +10 -56
README.md CHANGED
@@ -16,8 +16,8 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  This model is a fine-tuned version of [MCG-NJU/videomae-base](https://huggingface.co/MCG-NJU/videomae-base) on an unknown dataset.
18
  It achieves the following results on the evaluation set:
19
- - Loss: 0.2709
20
- - Accuracy: 0.9540
21
 
22
  ## Model description
23
 
@@ -43,67 +43,21 @@ The following hyperparameters were used during training:
43
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
44
  - lr_scheduler_type: linear
45
  - lr_scheduler_warmup_ratio: 0.1
46
- - training_steps: 3750
47
 
48
  ### Training results
49
 
50
  | Training Loss | Epoch | Step | Validation Loss | Accuracy |
51
  |:-------------:|:-----:|:----:|:---------------:|:--------:|
52
- | 2.2549 | 0.02 | 75 | 2.2707 | 0.0270 |
53
- | 1.7546 | 1.02 | 150 | 1.8893 | 0.3514 |
54
- | 0.8462 | 2.02 | 225 | 0.8723 | 0.6216 |
55
- | 0.551 | 3.02 | 300 | 0.4068 | 0.8108 |
56
- | 0.6627 | 4.02 | 375 | 0.6916 | 0.7297 |
57
- | 0.4383 | 5.02 | 450 | 0.5512 | 0.7568 |
58
- | 0.3398 | 6.02 | 525 | 0.4060 | 0.8378 |
59
- | 0.0769 | 7.02 | 600 | 0.8299 | 0.8108 |
60
- | 0.0077 | 8.02 | 675 | 0.0570 | 0.9730 |
61
- | 0.0055 | 9.02 | 750 | 0.0168 | 1.0 |
62
- | 0.002 | 10.02 | 825 | 0.0497 | 0.9730 |
63
- | 0.1242 | 11.02 | 900 | 0.2132 | 0.9459 |
64
- | 0.0022 | 12.02 | 975 | 0.0026 | 1.0 |
65
- | 0.0074 | 13.02 | 1050 | 0.0577 | 0.9730 |
66
- | 0.0038 | 14.02 | 1125 | 0.0137 | 1.0 |
67
- | 0.0011 | 15.02 | 1200 | 0.0022 | 1.0 |
68
- | 0.001 | 16.02 | 1275 | 0.0025 | 1.0 |
69
- | 0.0009 | 17.02 | 1350 | 0.0059 | 1.0 |
70
- | 0.0024 | 18.02 | 1425 | 0.1411 | 0.9730 |
71
- | 0.1709 | 19.02 | 1500 | 0.0041 | 1.0 |
72
- | 0.0008 | 20.02 | 1575 | 0.0489 | 0.9730 |
73
- | 0.0007 | 21.02 | 1650 | 0.0116 | 1.0 |
74
- | 0.0008 | 22.02 | 1725 | 0.0741 | 0.9730 |
75
- | 0.0008 | 23.02 | 1800 | 0.1699 | 0.9730 |
76
- | 0.0007 | 24.02 | 1875 | 0.1828 | 0.9730 |
77
- | 0.0006 | 25.02 | 1950 | 0.1652 | 0.9730 |
78
- | 0.0006 | 26.02 | 2025 | 0.1608 | 0.9730 |
79
- | 0.0005 | 27.02 | 2100 | 0.1595 | 0.9730 |
80
- | 0.0005 | 28.02 | 2175 | 0.1445 | 0.9730 |
81
- | 0.0006 | 29.02 | 2250 | 0.1488 | 0.9730 |
82
- | 0.0005 | 30.02 | 2325 | 0.1202 | 0.9730 |
83
- | 0.0005 | 31.02 | 2400 | 0.1238 | 0.9730 |
84
- | 0.0004 | 32.02 | 2475 | 0.1225 | 0.9730 |
85
- | 0.0005 | 33.02 | 2550 | 0.2320 | 0.9459 |
86
- | 0.0004 | 34.02 | 2625 | 0.0791 | 0.9730 |
87
- | 0.0005 | 35.02 | 2700 | 0.1285 | 0.9730 |
88
- | 0.0004 | 36.02 | 2775 | 0.1719 | 0.9730 |
89
- | 0.0007 | 37.02 | 2850 | 0.1799 | 0.9730 |
90
- | 0.0004 | 38.02 | 2925 | 0.1936 | 0.9730 |
91
- | 0.0004 | 39.02 | 3000 | 0.1844 | 0.9730 |
92
- | 0.0004 | 40.02 | 3075 | 0.1790 | 0.9730 |
93
- | 0.0004 | 41.02 | 3150 | 0.1747 | 0.9730 |
94
- | 0.0004 | 42.02 | 3225 | 0.1359 | 0.9730 |
95
- | 0.0004 | 43.02 | 3300 | 0.1283 | 0.9730 |
96
- | 0.0004 | 44.02 | 3375 | 0.1209 | 0.9730 |
97
- | 0.0004 | 45.02 | 3450 | 0.0876 | 0.9730 |
98
- | 0.0004 | 46.02 | 3525 | 0.0933 | 0.9730 |
99
- | 0.0004 | 47.02 | 3600 | 0.0976 | 0.9730 |
100
- | 0.0004 | 48.02 | 3675 | 0.1011 | 0.9730 |
101
- | 0.0004 | 49.02 | 3750 | 0.1050 | 0.9730 |
102
 
103
 
104
  ### Framework versions
105
 
106
- - Transformers 4.24.0
107
- - Pytorch 1.8.0+cu111
108
- - Datasets 2.7.1
109
  - Tokenizers 0.13.2
 
16
 
17
  This model is a fine-tuned version of [MCG-NJU/videomae-base](https://huggingface.co/MCG-NJU/videomae-base) on an unknown dataset.
18
  It achieves the following results on the evaluation set:
19
+ - Loss: 1.4743
20
+ - Accuracy: 0.5419
21
 
22
  ## Model description
23
 
 
43
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
44
  - lr_scheduler_type: linear
45
  - lr_scheduler_warmup_ratio: 0.1
46
+ - training_steps: 300
47
 
48
  ### Training results
49
 
50
  | Training Loss | Epoch | Step | Validation Loss | Accuracy |
51
  |:-------------:|:-----:|:----:|:---------------:|:--------:|
52
+ | 2.3247 | 0.25 | 75 | 2.2102 | 0.2143 |
53
+ | 2.2528 | 1.25 | 150 | 2.0479 | 0.1571 |
54
+ | 1.8267 | 2.25 | 225 | 1.7605 | 0.3286 |
55
+ | 1.5876 | 3.25 | 300 | 1.6022 | 0.3714 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
56
 
57
 
58
  ### Framework versions
59
 
60
+ - Transformers 4.26.1
61
+ - Pytorch 1.8.1+cu111
62
+ - Datasets 2.9.0
63
  - Tokenizers 0.13.2