zhengye

I meet the same problem on my dataset，i tried deceasing the lr，but it does not works.

Do you have any methods to solve it?I have meet the same question.

下载后的文件是分卷压缩的形式，请按照分卷压缩文件的解压方式解压，之前有很多人都下载解压过没有问题

这个比赛有点久远了，我记忆中是比赛的时候，主办方合并了一些相似的类目，所以直接上类别减少了，并不需要我们处理，给定的标签文件几个类就是几类，具体的数据预处理在readme都写了。

In training, the semantic_score is calucated in [here](https://github.com/zhengye1995/Zero-shot-Instance-Segmentation/blob/45ee140205ce1f4acaaa6e72ddb526c962068378/mmdet/models/bbox_heads/convfc_bbox_semantic_head.py#L213)： `semantic_score = torch.mm(semantic_score, self.vec)`. Therefore the test code you mentioned above is in line with training.

In training, the classification loss function is ```F.Cross_entropy(input, target)```, which contains the softmax. Therefore, the semantic_score is also processed with softmax.

I got your problem, in inference： for seen classes, using: ``` seen_scores = torch.mm(scores, self.vec.t()) seen_scores = torch.mm(seen_scores, self.vec) ``` or directly the scores only have the numerical difference, the...

For unseen scores, this line ```unseen_scores = torch.mm(scores, self.vec.t())``` projects the scores back to the semantic space: e.g., scores (100, s) mm vec.t (s, 300) -> (100,300) then this line...

I did not try this before, you can have a try and good luck.

generated during training process automaticly.