Spaces:
Running
on
Zero
Running
on
Zero
File size: 1,805 Bytes
bc752b1 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 |
import torch
import torch.nn as nn
from .eva_clip_processors import EvaClipImageTrainProcessor
from .eva_vit import Eva2LargePlusEncoder
class EvaClipVisionTower(nn.Module):
def __init__(self, vision_tower, args, delay_load=False):
super().__init__()
self.is_loaded = False
self.vision_tower_path = vision_tower
self.config = VisionTowerConfig()
if not delay_load:
self.load_model()
else:
self.cfg_only = self.config
def load_model(self):
self.image_processor = EvaClipImageTrainProcessor(self.config.image_size)
self.vision_tower = Eva2LargePlusEncoder(self.vision_tower_path)
self.vision_tower.requires_grad_(False)
self.is_loaded = True
@torch.no_grad()
def forward(self, images):
if type(images) is list:
image_features = []
for image in images:
image_feature = self.vision_tower(
image.to(device=self.device, dtype=self.dtype).unsqueeze(0)
).to(image.dtype)
image_features.append(image_feature)
else:
image_features = self.vision_tower(images.to(device=self.device, dtype=self.dtype)).to(
images.dtype
)
return image_features
@property
def dtype(self):
return self.vision_tower.dtype
@property
def device(self):
return self.vision_tower.device
@property
def hidden_size(self):
return self.config.hidden_size
@property
def num_patches(self):
return (self.config.image_size // self.config.patch_size) ** 2
class VisionTowerConfig:
def __init__(self):
self.image_size = 336
self.patch_size = 14
self.hidden_size = 1024
|