How to implement fine-tuning phase?

[LXYrsy]

Can you tell me how to implement this phase with your code? Looking forward to your reply!

[infinity7428]

I also uploaded a code for fine-tuning. Read the for fine-tune.

[LXYrsy]

sorry ,when i want to try fine-tuning ,it doesn't work,need some json,like seed1/3shot_novel.json,can you tell me how to get it?

[infinity7428]

JSON files are included in the support images. See Support images and JSON tab in README.

[LXYrsy]

when i want to try fine-tuning,it doesn't work, error is :ValueError: Sample larger than population or is negative
can you tell me how to solve it?

[LXYrsy]

error:
File "/home/lxy/hANMCL-main/lib/roi_data_layer/fs_loader.py", line 115, in getitem
support_dbs = random.sample(self.support_db[pos_cls_idx], k=self.support_shot)
File "/opt/conda/lib/python3.8/random.py", line 363, in sample
raise ValueError("Sample larger than population or is negative")

[infinity7428]

Did you specify the correct folder for the support images?
For example, --sup_dir your_target_support_dir

[Tara-Liu]

错误： 文件“/home/lxy/hANMCL-main/lib/roi_data_layer/fs_loader.py”，第 115 行，在getitem support_dbs = random.sample(self.support_db[pos_cls_idx], k=self.support_shot) 文件“/opt /conda/lib/python3.8/random.py"，第 363 行，样本中的 raise ValueError("Sample large than population or is negative")

我遇到了同样的问题， 指定了--sup_dir 也不起作用 请问你当时怎么解决的

[LXYrsy]

这个问题应该是用python train_ft.py，而不是python train.py

[Tara-Liu]

这个问题应该是用python train_ft.py，而不是python train.py

 好的 谢谢