MCPcopy Create free account
hub / github.com/MotrixLab/ViMoGen / TextDataset

Class TextDataset

models/transformer/wan/text_encoding_batch.py:8–34  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

6import json
7
8class TextDataset(torch.utils.data.Dataset):
9 def __init__(self, json_file, text_key):
10
11 self.data_list = json.load(open(json_file, 'r'))
12 self.data_list = [data for data in self.data_list if data[text_key] is not None]
13 self.text_key = text_key
14
15 def __getitem__(self, data_id):
16 data = self.data_list[data_id]
17 text = data[self.text_key]
18 if isinstance(text, list):
19 text = text[0]
20 if 'sample_id' in data:
21 sample_id = data['sample_id']
22 elif 'test_sample_id' in data:
23 sample_id = str(data['test_sample_id'])
24 elif 'global_id' in data:
25 sample_id = str(data['global_id'])
26 else:
27 raise ValueError("No sample_id or test_sample_id found in the data.")
28
29 data_dict = {"text": text, "sample_id": sample_id}
30
31 return data_dict
32
33 def __len__(self):
34 return len(self.data_list)
35
36if __name__ == "__main__":
37

Callers 1

Calls

no outgoing calls

Tested by

no test coverage detected