MCPcopy Create free account
hub / github.com/InternRobotics/G2VLM / get_data_paths

Method get_data_paths

data/vlm_dataset.py:42–61  ·  view source on GitHub ↗
(
        self, 
        jsonl_path_list, 
        data_dir_list, 
        num_used_data, 
        shuffle_lines, 
        shuffle_seed,
    )

Source from the content-addressed store, hash-verified

40 self.set_epoch()
41
42 def get_data_paths(
43 self,
44 jsonl_path_list,
45 data_dir_list,
46 num_used_data,
47 shuffle_lines,
48 shuffle_seed,
49 ):
50 data_paths = []
51 for jsonl_path, image_dir, num_data_point in zip(
52 jsonl_path_list, data_dir_list, num_used_data
53 ):
54 with open(jsonl_path, 'r') as f:
55 raw_data = f.readlines()
56 if shuffle_lines:
57 self.rng.seed(shuffle_seed)
58 self.rng.shuffle(raw_data)
59 raw_data = raw_data[:num_data_point]
60 data_paths.extend([(json_data, image_dir) for json_data in raw_data])
61 return data_paths
62
63 def change_format(self, data, num_images):
64 elements = []

Callers 1

__init__Method · 0.95

Calls

no outgoing calls

Tested by

no test coverage detected