[Bug] unpaired_image_dataset.py conduct random sampling in training + some urls are wrong in cyclegan's README.

[LeoXing1996]

Prerequisite

• I have searched Issues and Discussions but cannot get the expected help.
• I have read the FAQ documentation but cannot get the expected help.
• The bug has not been fixed in the latest version (master) or latest version (1.x).

Task

I'm using the official example scripts/configs for the officially supported tasks/models/datasets.

Branch

1.x branch https://github.com/open-mmlab/mmediting/tree/1.x

Environment

pytorch-1.11.0-pytorch
mmcv-2.0.0-rc1
mmengine-0.1.0

Reproduces the problem - code sample

    def get_data_info(self, idx) -> dict:
        img_a_path = self.data_infos_a[idx % self.len_a]['path']
        if self.test_mode:
            idx_b = np.random.randint(0, self.len_b)
            img_b_path = self.data_infos_b[idx_b]['path']
        else:
            img_b_path = self.data_infos_b[idx % self.len_b]['path']
        data_info = dict()
        data_info[f'img_{self.domain_a}_path'] = img_a_path
        data_info[f'img_{self.domain_b}_path'] = img_b_path
        return data_info

Reproduces the problem - command or script

bash tools/slurm_train.sh PARTITION JOB_NAME config/cyclegan/cyclegan_lsgan-id0-resnet-in_1xb1-250kiters_summer2winter.py work_dirs/train_benchmark

Reproduces the problem - error message

The performance is much worse than expected.

Additional information

No response

[zengyh1900]

Fixed in #1308.