Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
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.
BIT2024
1682f50c01
|
1 year ago | |
|---|---|---|
| Dataset | 1 year ago | |
| ckptFE | 1 year ago | |
| README1.md | 1 year ago | |
| dataset.py | 1 year ago | |
| models.py | 1 year ago | |
| path_config.py | 1 year ago | |
| predict.py | 1 year ago | |
| requirements.txt | 1 year ago | |
| run.py | 1 year ago | |
| template.py | 1 year ago | |
| test_clip.py | 1 year ago | |
| train_clip.py | 1 year ago | |
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。
Python Text Markdown