MCPcopy Create free account
hub / github.com/THUDM/GLM / __init__

Method __init__

data_utils/corpora.py:67–75  ·  view source on GitHub ↗
(self, prompt_loader, text_loader, tokenizer=None, to_tokenize=False, **kwargs)

Source from the content-addressed store, hash-verified

65
66class PromptDataset(data.Dataset):
67 def __init__(self, prompt_loader, text_loader, tokenizer=None, to_tokenize=False, **kwargs):
68 self.prompts = prompt_loader
69 self.texts = text_loader
70 self.tokenizer = tokenizer
71 self.to_tokenize = to_tokenize
72 if isinstance(self.prompts, LazyLoader) and isinstance(self.texts, LazyLoader):
73 self.prompt_lens = self.prompts.lens
74 self.text_lens = self.texts.lens
75 self.is_lazy = True
76
77 def get_text_len(self, idx):
78 return self.prompt_lens[idx] + self.text_lens[idx]

Callers

nothing calls this directly

Calls

no outgoing calls

Tested by

no test coverage detected