jsonl_path_list: list of jsonl file paths data_dir_list: list of image directories containing the images of each jsonl file num_used_data: list of number of sampled data points for each jsonl
(
self, dataset_name, dino_transform, tokenizer, frame_sampler,
jsonl_path_list, data_dir_list, num_used_data,
local_rank=0, world_size=1, num_workers=8, data_status=None,
shuffle_lines=False, shuffle_seed=0,
)
| 105 | |
| 106 | class SftJSONLIterableReconDataset(DistributedIterableDataset): |
| 107 | def __init__( |
| 108 | self, dataset_name, dino_transform, tokenizer, frame_sampler, |
| 109 | jsonl_path_list, data_dir_list, num_used_data, |
| 110 | local_rank=0, world_size=1, num_workers=8, data_status=None, |
| 111 | shuffle_lines=False, shuffle_seed=0, |
| 112 | ): |
| 113 | """ |
| 114 | jsonl_path_list: list of jsonl file paths |
| 115 | data_dir_list: list of image directories containing the images of each jsonl file |
| 116 | num_used_data: list of number of sampled data points for each jsonl |
| 117 | """ |
| 118 | super().__init__(dataset_name, local_rank, world_size, num_workers) |
| 119 | # self.transform = transform |
| 120 | # self.vit_transform = vit_transform |
| 121 | self.dino_transform = dino_transform |
| 122 | self.tokenizer = tokenizer |
| 123 | self.frame_sampler = frame_sampler |
| 124 | self.data_status = data_status |
| 125 | |
| 126 | self.img_size = 518 ### second stage |
| 127 | self.use_dinov3 = False |
| 128 | if self.use_dinov3: |
| 129 | self.patch_size = 16 |
| 130 | else: |
| 131 | self.patch_size = 14 |
| 132 | if self.use_dinov3: |
| 133 | self.img_size = 512 # |
| 134 | |
| 135 | self.aug_scale = [0.8, 1.2] |
| 136 | self.rescale = True |
| 137 | self.rescale_aug = True |
| 138 | self.landscape_check = False #True |
| 139 | self.training = True # hardcode |
| 140 | self.enable_random_image_num = True |
| 141 | self.ceph_read = True |
| 142 | |
| 143 | self._rng = np.random.default_rng(shuffle_seed) |
| 144 | |
| 145 | self.aug_crop = 16 ###aug_crop |
| 146 | self.aug_focal = 0.9 ####aug_focal |
| 147 | self.z_far = 0 ####z_far |
| 148 | self.random_sample_thres = 0.1 #random_sample_thre |
| 149 | |
| 150 | self.data_paths = self.get_data_paths( |
| 151 | jsonl_path_list, |
| 152 | data_dir_list, |
| 153 | num_used_data, |
| 154 | shuffle_lines, |
| 155 | shuffle_seed, |
| 156 | ) |
| 157 | self.base_seed = shuffle_seed |
| 158 | self.random_image_num = 0 |
| 159 | self.frame_num = 0 |
| 160 | self.random_aspect_ratio = 1.0 |
| 161 | self.resolution = [224, 224] |
| 162 | if self.use_dinov3: |
| 163 | self.resolution = [256, 256] |
| 164 |
nothing calls this directly
no test coverage detected