| from transformers import PretrainedConfig | |
| class InternImageConfig(PretrainedConfig): | |
| model_type = "intern_image" | |
| def __init__( | |
| self, | |
| core_op='DCNv3_pytorch', | |
| channels=64, | |
| depths=(4, 4, 18, 4), | |
| groups=(4, 8, 16, 32), | |
| num_classes=1000, | |
| mlp_ratio=4., | |
| drop_rate=0., | |
| drop_path_rate=0.1, | |
| drop_path_type='linear', | |
| act_layer='GELU', | |
| norm_layer='LN', | |
| layer_scale=None, | |
| offset_scale=1.0, | |
| post_norm=False, | |
| cls_scale=1.5, | |
| with_cp=False, | |
| **kwargs, | |
| ): | |
| self.core_op = core_op | |
| self.channels = channels | |
| self.depths = depths | |
| self.groups = groups | |
| self.num_classes = num_classes | |
| self.mlp_ratio = mlp_ratio | |
| self.drop_rate = drop_rate | |
| self.drop_path_rate = drop_path_rate | |
| self.drop_path_type = drop_path_type | |
| self.act_layer = act_layer | |
| self.norm_layer = norm_layer | |
| self.layer_scale = layer_scale | |
| self.offset_scale = offset_scale | |
| self.post_norm = post_norm | |
| self.cls_scale = cls_scale | |
| self.with_cp = with_cp | |
| super().__init__(**kwargs) | |