MCPcopy Create free account
hub / github.com/OpenGVLab/EfficientQAT / TrainingArguments

Class TrainingArguments

deita_dataset/train.py:53–67  ·  view source on GitHub ↗

Source from the content-addressed store, hash-verified

51
52@dataclass
53class TrainingArguments(transformers.TrainingArguments):
54 cache_dir: Optional[str] = field(default=None)
55 optim: str = field(default="adamw_torch")
56 model_max_length: int = field(
57 default=512,
58 metadata={
59 "help": "Maximum sequence length. Sequences will be right padded (and possibly truncated)."
60 },
61 )
62 min_lr: float = field(
63 default = None
64 )
65 mask_user: bool = field(
66 default = True
67 )
68
69
70local_rank = None

Callers

nothing calls this directly

Calls

no outgoing calls

Tested by

no test coverage detected