Upload modeling.py

modeling.py

@@ -0,0 +1,66 @@
+from transformers import CLIPVisionModel
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from dataclasses import dataclass
+
+@dataclass
+class VisualEncoderConfig:
+    n_embd: int = 2048
+    vision_tower_name: str = 'openai/clip-vit-large-patch14-336'
+    grid_size: int = -1 # -1: no grid pooling, 0: take cls token, 1: global avg pooling, 2, 3, 4, ...: grid pooling
+
+class VisualEncoder(nn.Module):
+    def __init__(self, args):
+        super().__init__()
+        self.args = args
+        self.vit = CLIPVisionModel.from_pretrained(args.vision_tower_name)
+        self.proj = nn.Linear(self.vit.config.hidden_size, args.n_embd, bias=False)
+
+    def encode_images(self, images):
+        B, N, C, H, W = images.shape
+        images = images.view(B*N, C, H, W)
+        image_features = self.vit(images).last_hidden_state
+        L, D = image_features.shape[1], image_features.shape[2]
+        # rerange [B*N, L, D] -> [B, N, L, D]
+        image_features = image_features.view(B, N, L, D)[:, 0, :, :]
+        image_features = self.grid_pooling(image_features)
+        return self.proj(image_features)
+    
+    def grid_pooling(self, image_features):
+        if self.args.grid_size == -1: # no grid pooling
+            return image_features
+        if self.args.grid_size == 0: # take cls token
+            return image_features[:, 0:1, :]
+        if self.args.grid_size == 1: # global avg pooling
+            return image_features.mean(dim=1, keepdim=True)
+        cls_features = image_features[:, 0:1, :]
+        image_features = image_features[:, 1:, :] #drop cls token
+        B, L, D = image_features.shape
+        H_or_W = int(L**0.5)
+        image_features = image_features.view(B, H_or_W, H_or_W, D)
+        grid_stride = H_or_W // self.args.grid_size
+        image_features = F.avg_pool2d(image_features.permute(0, 3, 1, 2), 
+                                      padding=0,
+                                      kernel_size=grid_stride, 
+                                      stride=grid_stride)
+        image_features = image_features.permute(0, 2, 3, 1).view(B, -1, D)
+        return torch.cat((cls_features, image_features), dim=1)
+
+
+class EmbeddingMixer(nn.Module):
+    def __init__(self, original_embedding, num_image_embeddings=4096):
+        super().__init__()
+        image_embedding = torch.zeros(num_image_embeddings, 
+                                      original_embedding.shape[1],
+                                      device=original_embedding.device,
+                                      dtype=original_embedding.dtype)
+        self.embedding = torch.cat((original_embedding, image_embedding), dim=0)
+        self.image_start_index = len(original_embedding)
+
+    def set_image_embeddings(self, image_embeddings):
+        end_index = self.image_start_index + image_embeddings.shape[0]
+        self.embedding[self.image_start_index:end_index] = image_embeddings
+
+    def get_input_embeddings(self):
+        return self.embedding