BIT2024
/
clip
Not watched
1
0
Fork 0
Code
Releases 0 Wiki evaluate Activity
Issues 0 Pull Requests 0 Datasets Model Cloudbrain HPC
You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
45 Commits
1 Branch
3.0 MB
0 Download
Tree: 7cc4e5c1bf
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '7cc4e5c1bf'
${ noResults }
clip/logs/train.log

train.log 12 kB
Raw Normal View History

ADD file via upload
1 year ago
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
​
 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152 
  1. 2024-07-30 09:42:54 train_clip.py [line:144] INFO: Model will be saved at ckptFE/07-30/version_1

  2. 2024-07-30 09:42:54 train_clip.py [line:73] INFO: model parameters: 

  3. 

  4. 2024-07-30 09:42:54 train_clip.py [line:74] INFO: +-----------------------+-------------------+

  5. | Parameter             | Value             |

  6. +=======================+===================+

  7. | seed                  | 123               |

  8. +-----------------------+-------------------+

  9. | optimizer             | Adan              |

  10. +-----------------------+-------------------+

  11. | lr                    | 3e-06             |

  12. +-----------------------+-------------------+

  13. | betas                 | [0.9, 0.98, 0.98] |

  14. +-----------------------+-------------------+

  15. | eps                   | 1e-08             |

  16. +-----------------------+-------------------+

  17. | weight_decay          | 0.2               |

  18. +-----------------------+-------------------+

  19. | batch_size            | 256               |

  20. +-----------------------+-------------------+

  21. | num_workers           | 8                 |

  22. +-----------------------+-------------------+

  23. | epochs                | 100               |

  24. +-----------------------+-------------------+

  25. | early_stop            | False             |

  26. +-----------------------+-------------------+

  27. | patience              | 10                |

  28. +-----------------------+-------------------+

  29. | delta                 | 0.0001            |

  30. +-----------------------+-------------------+

  31. | caption_version       | 2                 |

  32. +-----------------------+-------------------+

  33. | data_augment          | 0                 |

  34. +-----------------------+-------------------+

  35. | model_save_path       | ckptFE            |

  36. +-----------------------+-------------------+

  37. | log_save_path         | logs              |

  38. +-----------------------+-------------------+

  39. | compute_acc_frequency | 0                 |

  40. +-----------------------+-------------------+

  41. | use_scheduler         | False             |

  42. +-----------------------+-------------------+

  43. | save_frequency        | 5                 |

  44. +-----------------------+-------------------+

  45. | freeze_version        | 2                 |

  46. +-----------------------+-------------------+

  47. 2024-07-30 09:42:54 train_clip.py [line:76] INFO: 

  48. 

  49. Start training...

  50. 2024-07-30 09:43:15 train_clip.py [line:116] INFO: Epoch: 1, Loss: 6.7326 Min Loss: 6.7326

  51. 2024-07-30 09:43:26 train_clip.py [line:116] INFO: Epoch: 2, Loss: 4.8350 Min Loss: 4.8350

  52. 2024-07-30 09:43:36 train_clip.py [line:116] INFO: Epoch: 3, Loss: 4.2236 Min Loss: 4.2236

  53. 2024-07-30 09:43:47 train_clip.py [line:116] INFO: Epoch: 4, Loss: 3.6406 Min Loss: 3.6406

  54. 2024-07-30 09:44:00 train_clip.py [line:116] INFO: Epoch: 5, Loss: 3.1923 Min Loss: 3.1923

  55. 2024-07-30 09:44:10 train_clip.py [line:116] INFO: Epoch: 6, Loss: 2.7900 Min Loss: 2.7900

  56. 2024-07-30 09:44:21 train_clip.py [line:116] INFO: Epoch: 7, Loss: 2.7613 Min Loss: 2.7613

  57. 2024-07-30 09:44:31 train_clip.py [line:116] INFO: Epoch: 8, Loss: 2.4837 Min Loss: 2.4837

  58. 2024-07-30 09:44:42 train_clip.py [line:116] INFO: Epoch: 9, Loss: 2.3107 Min Loss: 2.3107

  59. 2024-07-30 09:44:52 train_clip.py [line:116] INFO: Epoch: 10, Loss: 2.5804 Min Loss: 2.3107

  60. 2024-07-30 09:45:03 train_clip.py [line:116] INFO: Epoch: 11, Loss: 2.3105 Min Loss: 2.3105

  61. 2024-07-30 09:45:13 train_clip.py [line:116] INFO: Epoch: 12, Loss: 2.2390 Min Loss: 2.2390

  62. 2024-07-30 09:45:24 train_clip.py [line:116] INFO: Epoch: 13, Loss: 2.1971 Min Loss: 2.1971

  63. 2024-07-30 09:45:34 train_clip.py [line:116] INFO: Epoch: 14, Loss: 2.1353 Min Loss: 2.1353

  64. 2024-07-30 09:45:44 train_clip.py [line:116] INFO: Epoch: 15, Loss: 2.1982 Min Loss: 2.1353

  65. 2024-07-30 09:45:55 train_clip.py [line:116] INFO: Epoch: 16, Loss: 2.1203 Min Loss: 2.1203

  66. 2024-07-30 09:46:06 train_clip.py [line:116] INFO: Epoch: 17, Loss: 2.0903 Min Loss: 2.0903

  67. 2024-07-30 09:46:12 train_clip.py [line:116] INFO: Epoch: 18, Loss: 2.2247 Min Loss: 2.0903

  68. 2024-07-30 09:46:23 train_clip.py [line:116] INFO: Epoch: 19, Loss: 1.9555 Min Loss: 1.9555

  69. 2024-07-30 09:46:34 train_clip.py [line:116] INFO: Epoch: 20, Loss: 2.1647 Min Loss: 1.9555

  70. 2024-07-30 09:46:40 train_clip.py [line:116] INFO: Epoch: 21, Loss: 2.1485 Min Loss: 1.9555

  71. 2024-07-30 09:46:48 train_clip.py [line:116] INFO: Epoch: 22, Loss: 1.9868 Min Loss: 1.9555

  72. 2024-07-30 09:46:55 train_clip.py [line:116] INFO: Epoch: 23, Loss: 2.0948 Min Loss: 1.9555

  73. 2024-07-30 09:47:02 train_clip.py [line:116] INFO: Epoch: 24, Loss: 2.2866 Min Loss: 1.9555

  74. 2024-07-30 09:47:14 train_clip.py [line:116] INFO: Epoch: 25, Loss: 2.0188 Min Loss: 1.9555

  75. 2024-07-30 09:47:20 train_clip.py [line:116] INFO: Epoch: 26, Loss: 1.9677 Min Loss: 1.9555

  76. 2024-07-30 09:47:27 train_clip.py [line:116] INFO: Epoch: 27, Loss: 2.1724 Min Loss: 1.9555

  77. 2024-07-30 09:47:35 train_clip.py [line:116] INFO: Epoch: 28, Loss: 2.1688 Min Loss: 1.9555

  78. 2024-07-30 09:47:42 train_clip.py [line:116] INFO: Epoch: 29, Loss: 2.1263 Min Loss: 1.9555

  79. 2024-07-30 09:47:53 train_clip.py [line:116] INFO: Epoch: 30, Loss: 1.9607 Min Loss: 1.9555

  80. 2024-07-30 09:48:00 train_clip.py [line:116] INFO: Epoch: 31, Loss: 2.1069 Min Loss: 1.9555

  81. 2024-07-30 09:48:07 train_clip.py [line:116] INFO: Epoch: 32, Loss: 2.1050 Min Loss: 1.9555

  82. 2024-07-30 09:48:14 train_clip.py [line:116] INFO: Epoch: 33, Loss: 1.9982 Min Loss: 1.9555

  83. 2024-07-30 09:48:22 train_clip.py [line:116] INFO: Epoch: 34, Loss: 2.0697 Min Loss: 1.9555

  84. 2024-07-30 09:48:33 train_clip.py [line:116] INFO: Epoch: 35, Loss: 2.0341 Min Loss: 1.9555

  85. 2024-07-30 09:48:39 train_clip.py [line:116] INFO: Epoch: 36, Loss: 2.1605 Min Loss: 1.9555

  86. 2024-07-30 09:48:47 train_clip.py [line:116] INFO: Epoch: 37, Loss: 1.9961 Min Loss: 1.9555

  87. 2024-07-30 09:48:54 train_clip.py [line:116] INFO: Epoch: 38, Loss: 2.1738 Min Loss: 1.9555

  88. 2024-07-30 09:49:01 train_clip.py [line:116] INFO: Epoch: 39, Loss: 2.1491 Min Loss: 1.9555

  89. 2024-07-30 09:49:12 train_clip.py [line:116] INFO: Epoch: 40, Loss: 1.9688 Min Loss: 1.9555

  90. 2024-07-30 09:49:19 train_clip.py [line:116] INFO: Epoch: 41, Loss: 2.0219 Min Loss: 1.9555

  91. 2024-07-30 09:49:26 train_clip.py [line:116] INFO: Epoch: 42, Loss: 1.9816 Min Loss: 1.9555

  92. 2024-07-30 09:49:34 train_clip.py [line:116] INFO: Epoch: 43, Loss: 2.0341 Min Loss: 1.9555

  93. 2024-07-30 09:49:41 train_clip.py [line:116] INFO: Epoch: 44, Loss: 2.1563 Min Loss: 1.9555

  94. 2024-07-30 09:49:52 train_clip.py [line:116] INFO: Epoch: 45, Loss: 1.9851 Min Loss: 1.9555

  95. 2024-07-30 09:49:59 train_clip.py [line:116] INFO: Epoch: 46, Loss: 2.0096 Min Loss: 1.9555

  96. 2024-07-30 09:50:10 train_clip.py [line:116] INFO: Epoch: 47, Loss: 1.9379 Min Loss: 1.9379

  97. 2024-07-30 09:50:17 train_clip.py [line:116] INFO: Epoch: 48, Loss: 2.0642 Min Loss: 1.9379

  98. 2024-07-30 09:50:24 train_clip.py [line:116] INFO: Epoch: 49, Loss: 2.0583 Min Loss: 1.9379

  99. 2024-07-30 09:50:35 train_clip.py [line:116] INFO: Epoch: 50, Loss: 1.9452 Min Loss: 1.9379

  100. 2024-07-30 09:50:42 train_clip.py [line:116] INFO: Epoch: 51, Loss: 2.0361 Min Loss: 1.9379

  101. 2024-07-30 09:50:49 train_clip.py [line:116] INFO: Epoch: 52, Loss: 1.9722 Min Loss: 1.9379

  102. 2024-07-30 09:50:56 train_clip.py [line:116] INFO: Epoch: 53, Loss: 1.9925 Min Loss: 1.9379

  103. 2024-07-30 09:51:04 train_clip.py [line:116] INFO: Epoch: 54, Loss: 2.1328 Min Loss: 1.9379

  104. 2024-07-30 09:51:15 train_clip.py [line:116] INFO: Epoch: 55, Loss: 2.0497 Min Loss: 1.9379

  105. 2024-07-30 09:51:21 train_clip.py [line:116] INFO: Epoch: 56, Loss: 2.0061 Min Loss: 1.9379

  106. 2024-07-30 09:51:29 train_clip.py [line:116] INFO: Epoch: 57, Loss: 2.2227 Min Loss: 1.9379

  107. 2024-07-30 09:51:36 train_clip.py [line:116] INFO: Epoch: 58, Loss: 1.9891 Min Loss: 1.9379

  108. 2024-07-30 09:51:43 train_clip.py [line:116] INFO: Epoch: 59, Loss: 1.9929 Min Loss: 1.9379

  109. 2024-07-30 09:51:54 train_clip.py [line:116] INFO: Epoch: 60, Loss: 2.0990 Min Loss: 1.9379

  110. 2024-07-30 09:52:01 train_clip.py [line:116] INFO: Epoch: 61, Loss: 2.0332 Min Loss: 1.9379

  111. 2024-07-30 09:52:12 train_clip.py [line:116] INFO: Epoch: 62, Loss: 1.9359 Min Loss: 1.9359

  112. 2024-07-30 09:52:19 train_clip.py [line:116] INFO: Epoch: 63, Loss: 2.0810 Min Loss: 1.9359

  113. 2024-07-30 09:52:26 train_clip.py [line:116] INFO: Epoch: 64, Loss: 2.1124 Min Loss: 1.9359

  114. 2024-07-30 09:52:37 train_clip.py [line:116] INFO: Epoch: 65, Loss: 2.0724 Min Loss: 1.9359

  115. 2024-07-30 09:52:44 train_clip.py [line:116] INFO: Epoch: 66, Loss: 2.0832 Min Loss: 1.9359

  116. 2024-07-30 09:52:52 train_clip.py [line:116] INFO: Epoch: 67, Loss: 2.0640 Min Loss: 1.9359

  117. 2024-07-30 09:53:03 train_clip.py [line:116] INFO: Epoch: 68, Loss: 1.9184 Min Loss: 1.9184

  118. 2024-07-30 09:53:10 train_clip.py [line:116] INFO: Epoch: 69, Loss: 1.9776 Min Loss: 1.9184

  119. 2024-07-30 09:53:22 train_clip.py [line:116] INFO: Epoch: 70, Loss: 1.9696 Min Loss: 1.9184

  120. 2024-07-30 09:53:28 train_clip.py [line:116] INFO: Epoch: 71, Loss: 2.1936 Min Loss: 1.9184

  121. 2024-07-30 09:53:35 train_clip.py [line:116] INFO: Epoch: 72, Loss: 2.1248 Min Loss: 1.9184

  122. 2024-07-30 09:53:43 train_clip.py [line:116] INFO: Epoch: 73, Loss: 2.1000 Min Loss: 1.9184

  123. 2024-07-30 09:53:50 train_clip.py [line:116] INFO: Epoch: 74, Loss: 1.9791 Min Loss: 1.9184

  124. 2024-07-30 09:54:02 train_clip.py [line:116] INFO: Epoch: 75, Loss: 1.9313 Min Loss: 1.9184

  125. 2024-07-30 09:54:08 train_clip.py [line:116] INFO: Epoch: 76, Loss: 2.1005 Min Loss: 1.9184

  126. 2024-07-30 09:54:15 train_clip.py [line:116] INFO: Epoch: 77, Loss: 2.1343 Min Loss: 1.9184

  127. 2024-07-30 09:54:23 train_clip.py [line:116] INFO: Epoch: 78, Loss: 2.0381 Min Loss: 1.9184

  128. 2024-07-30 09:54:30 train_clip.py [line:116] INFO: Epoch: 79, Loss: 2.0349 Min Loss: 1.9184

  129. 2024-07-30 09:54:41 train_clip.py [line:116] INFO: Epoch: 80, Loss: 2.1082 Min Loss: 1.9184

  130. 2024-07-30 09:54:48 train_clip.py [line:116] INFO: Epoch: 81, Loss: 1.9580 Min Loss: 1.9184

  131. 2024-07-30 09:54:55 train_clip.py [line:116] INFO: Epoch: 82, Loss: 2.0330 Min Loss: 1.9184

  132. 2024-07-30 09:55:02 train_clip.py [line:116] INFO: Epoch: 83, Loss: 2.0308 Min Loss: 1.9184

  133. 2024-07-30 09:55:09 train_clip.py [line:116] INFO: Epoch: 84, Loss: 1.9555 Min Loss: 1.9184

  134. 2024-07-30 09:55:21 train_clip.py [line:116] INFO: Epoch: 85, Loss: 2.0020 Min Loss: 1.9184

  135. 2024-07-30 09:55:28 train_clip.py [line:116] INFO: Epoch: 86, Loss: 2.0153 Min Loss: 1.9184

  136. 2024-07-30 09:55:39 train_clip.py [line:116] INFO: Epoch: 87, Loss: 1.9062 Min Loss: 1.9062

  137. 2024-07-30 09:55:46 train_clip.py [line:116] INFO: Epoch: 88, Loss: 2.1427 Min Loss: 1.9062

  138. 2024-07-30 09:55:53 train_clip.py [line:116] INFO: Epoch: 89, Loss: 2.1196 Min Loss: 1.9062

  139. 2024-07-30 09:56:08 train_clip.py [line:116] INFO: Epoch: 90, Loss: 1.8230 Min Loss: 1.8230

  140. 2024-07-30 09:56:14 train_clip.py [line:116] INFO: Epoch: 91, Loss: 2.1242 Min Loss: 1.8230

  141. 2024-07-30 09:56:22 train_clip.py [line:116] INFO: Epoch: 92, Loss: 2.0993 Min Loss: 1.8230

  142. 2024-07-30 09:56:29 train_clip.py [line:116] INFO: Epoch: 93, Loss: 1.9123 Min Loss: 1.8230

  143. 2024-07-30 09:56:36 train_clip.py [line:116] INFO: Epoch: 94, Loss: 2.0217 Min Loss: 1.8230

  144. 2024-07-30 09:56:48 train_clip.py [line:116] INFO: Epoch: 95, Loss: 1.9521 Min Loss: 1.8230

  145. 2024-07-30 09:56:54 train_clip.py [line:116] INFO: Epoch: 96, Loss: 2.0764 Min Loss: 1.8230

  146. 2024-07-30 09:57:02 train_clip.py [line:116] INFO: Epoch: 97, Loss: 1.9347 Min Loss: 1.8230

  147. 2024-07-30 09:57:09 train_clip.py [line:116] INFO: Epoch: 98, Loss: 1.9534 Min Loss: 1.8230

  148. 2024-07-30 09:57:16 train_clip.py [line:116] INFO: Epoch: 99, Loss: 2.0558 Min Loss: 1.8230

  149. 2024-07-30 09:57:28 train_clip.py [line:116] INFO: Epoch: 100, Loss: 1.9141 Min Loss: 1.8230

  150. 2024-07-30 09:57:28 train_clip.py [line:125] INFO: 

  151. 

  152. Finish training...

冻结ViT-B/32版本的CLIP模型中的全部图像层，用Adan优化器训练模型，训练100个epoch，每隔5个epoch对模型进行保存；完成CLIP模型训练后，运行test_clip.py用测试集中的数据和自定义的提示词对保存的模型进行测试，选取测试精度最好的模型和对应的提示词，运行predict.py文件，选择“min_loss.pth”模型，提交官方系统测试，top1的精度是0.6788。

Contributors (1)
All