| from transformers import PretrainedConfig | |
| class PromptComplexityConfig(PretrainedConfig): | |
| model_type = "prompt-complexity" | |
| def __init__( | |
| self, | |
| base_model_name: str = "microsoft/deberta-v3-base", | |
| max_length: int = 512, | |
| dropout: float = 0.1, | |
| hidden: int | None = None, | |
| layernorm_after_pool: bool = True, | |
| use_projection: bool = False, | |
| proj_hidden_ratio: float = 1.0, | |
| output_sigmoid: bool = True, | |
| **kwargs, | |
| ): | |
| super().__init__(**kwargs) | |
| self.base_model_name = base_model_name | |
| self.max_length = max_length | |
| self.dropout = dropout | |
| self.hidden = hidden | |
| self.layernorm_after_pool = layernorm_after_pool | |
| self.use_projection = use_projection | |
| self.proj_hidden_ratio = proj_hidden_ratio | |
| self.output_sigmoid = output_sigmoid | |