From 3a4fa0f1cf3c381a3d81f39e6e9439ec192b2509 Mon Sep 17 00:00:00 2001
From: Gustaf Rydholm <gustaf.rydholm@gmail.com>
Date: Mon, 11 Sep 2023 22:14:43 +0200
Subject: Refactor vit to encoder only

---
 text_recognizer/network/vit.py | 49 ++++++------------------------------------
 1 file changed, 6 insertions(+), 43 deletions(-)

diff --git a/text_recognizer/network/vit.py b/text_recognizer/network/vit.py
index 1fbf3fc..a596792 100644
--- a/text_recognizer/network/vit.py
+++ b/text_recognizer/network/vit.py
@@ -1,16 +1,11 @@
-"""Transformer module."""
-from typing import Type
-
 from einops.layers.torch import Rearrange
 from torch import Tensor, nn
 
-from .transformer.embedding.token import TokenEmbedding
 from .transformer.embedding.sincos import sincos_2d
-from .transformer.decoder import Decoder
 from .transformer.encoder import Encoder
 
 
-class VisionTransformer(nn.Module):
+class Vit(nn.Module):
     def __init__(
         self,
         image_height: int,
@@ -18,16 +13,11 @@ class VisionTransformer(nn.Module):
         patch_height: int,
         patch_width: int,
         dim: int,
-        num_classes: int,
         encoder: Encoder,
-        decoder: Decoder,
-        token_embedding: TokenEmbedding,
-        pos_embedding: Type[nn.Module],
-        tie_embeddings: bool,
-        pad_index: int,
+        channels: int = 1,
     ) -> None:
         super().__init__()
-        patch_dim = patch_height * patch_width
+        patch_dim = patch_height * patch_width * channels
         self.to_patch_embedding = nn.Sequential(
             Rearrange(
                 "b c (h ph) (w pw) -> b (h w) (ph pw c)",
@@ -41,36 +31,9 @@ class VisionTransformer(nn.Module):
         self.patch_embedding = sincos_2d(
             h=image_height // patch_height, w=image_width // patch_width, dim=dim
         )
-        self.pos_embedding = pos_embedding
-        self.token_embedding = token_embedding
-        self.to_logits = (
-            nn.Linear(dim, num_classes)
-            if not tie_embeddings
-            else lambda t: t @ self.token_embedding.to_embedding.weight.t()
-        )
         self.encoder = encoder
-        self.decoder = decoder
-        self.pad_index = pad_index
 
-    def encode(self, img: Tensor) -> Tensor:
-        x = self.to_patch_embedding(img)
-        x += self.patch_embedding.to(img.device, dtype=img.dtype)
+    def forward(self, images: Tensor) -> Tensor:
+        x = self.to_patch_embedding(images)
+        x = x + self.patch_embedding.to(images.device, dtype=images.dtype)
         return self.encoder(x)
-
-    def decode(self, text: Tensor, img_features: Tensor) -> Tensor:
-        text = text.long()
-        mask = text != self.pad_index
-        tokens = self.token_embedding(text)
-        tokens = tokens + self.pos_embedding(tokens)
-        output = self.decoder(tokens, context=img_features, mask=mask)
-        return self.to_logits(output)
-
-    def forward(
-        self,
-        img: Tensor,
-        text: Tensor,
-    ) -> Tensor:
-        """Applies decoder block on input signals."""
-        img_features = self.encode(img)
-        logits = self.decode(text, img_features)
-        return logits  # [B, N, C]
-- 
cgit v1.2.3-70-g09d2