MCPcopy Create free account
hub / github.com/zai-org/CodeGeeX / PadDataset

Class PadDataset

codegeex/mindspore/src/sat_dataset.py:56–69  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

54
55
56class PadDataset(Dataset):
57 def __init__(self, dataset, seq_len, eod_id):
58 self.dataset = dataset
59 self.seq_len = seq_len + 1
60 self.eod_id = eod_id
61
62 def __len__(self):
63 return len(self.dataset)
64
65 def __getitem__(self, idx):
66 item = self.dataset[idx][0]
67 return (item[:self.seq_len],) if self.seq_len <= len(item) else (
68 np.concatenate((item, np.ones(self.seq_len - len(item)) * self.eod_id), axis=0),)
69 # return (np.pad(item, (0, 1), constant_values=self.eod_id),)
70
71
72class BinaryDataset(Dataset):

Callers 2

get_code_data_trainFunction · 0.85
get_code_data_evalFunction · 0.85

Calls

no outgoing calls

Tested by

no test coverage detected