what is s2 in vision load tower and how it effects training

[Jayantverma2]

import os from .clip_encoder import CLIPVisionTower, CLIPVisionTowerS2

def build_vision_tower(vision_tower_cfg, kwargs): vision_tower = getattr(vision_tower_cfg, 'mm_vision_tower', getattr(vision_tower_cfg, 'vision_tower', None)) is_absolute_path_exists = os.path.exists(vision_tower) use_s2 = getattr(vision_tower_cfg, 's2', False) if is_absolute_path_exists or vision_tower.startswith("openai") or vision_tower.startswith("laion") or "ShareGPT4V" in vision_tower: if use_s2: return CLIPVisionTowerS2(vision_tower, args=vision_tower_cfg, kwargs) else: return CLIPVisionTower(vision_tower, args=vision_tower_cfg, **kwargs)

raise ValueError(f'Unknown vision tower: {vision_tower}')

[mmaaz60]

Hi @Jayantverma2,

Thank you for your interest in our work. In order to use S2, you have to use the lates LLaVA code base. Its implemented here.