| from transformers import PretrainedConfig | |
| from typing import List | |
| class InfMLLMChatConfig(PretrainedConfig): | |
| def __init__( | |
| self, | |
| image_size=448, | |
| vision_tower="eva_clip_g", | |
| mm_projector_type="pooler", | |
| pool_out_size="32+16+8", | |
| mm_hidden_size=1408, | |
| hidden_size=4096, | |
| **kwargs | |
| ): | |
| self.image_size = image_size | |
| self.vision_tower = vision_tower | |
| self.mm_projector_type = mm_projector_type | |
| self.pool_out_size = pool_out_size | |
| self.mm_hidden_size = mm_hidden_size | |
| self.hidden_size = hidden_size | |
| super().__init__(**kwargs) | |