diff --git a/ginka/generator/rnn.py b/ginka/generator/rnn.py
index c899f1e..7499545 100644
--- a/ginka/generator/rnn.py
+++ b/ginka/generator/rnn.py
@@ -35,13 +35,13 @@ class GinkaMapPatch(nn.Module):
             nn.BatchNorm2d(256),
             nn.ReLU(),
             
-            nn.Conv2d(256, 512, 3, padding=1),
+            nn.Conv2d(256, 512, 3),
             nn.BatchNorm2d(512),
             nn.ReLU(),
             
-            nn.AvgPool2d(kernel_size=(5, 5)),
             nn.Flatten()
         )
+        self.fc = nn.Linear(512 * 3 * 3, 256)
         
     def forward(self, map: torch.Tensor, x: int, y: int):
         """
@@ -66,7 +66,9 @@ class GinkaMapPatch(nn.Module):
         result[:, 4, 4] = 0
         result = F.one_hot(result, num_classes=32).permute(0, 3, 2, 1).float()
         
-        return self.patch_cnn(result)
+        feat = self.patch_cnn(result)
+        feat = self.fc(feat)
+        return feat
     
 class GinkaTileEmbedding(nn.Module):
     def __init__(self, tile_classes=32, embed_dim=256):
@@ -119,12 +121,11 @@ class GinkaInputFusion(nn.Module):
         cond_vec: [B, 256]
         row_embed: [B, 256]
         col_embed: [B, 256]
-        patch_vec: [B, 512]
+        patch_vec: [B, 256]
         """
-        vec = torch.cat([tile_embed, cond_vec, row_embed, col_embed, patch_vec], dim=1)
-        vec = torch.stack(torch.split(vec, 256, dim=1), dim=1)
+        vec = torch.stack([tile_embed, cond_vec, row_embed, col_embed, patch_vec], dim=1)
         feat = self.transformer(vec)
-        return torch.mean(feat, dim=1)
+        return feat[:, 0]
 
 class GinkaRNN(nn.Module):
     def __init__(self, tile_classes=32, input_dim=256, hidden_dim=2048):
@@ -141,7 +142,7 @@ class GinkaRNN(nn.Module):
         """
         hidden = self.gru(feat_fusion, hidden)
         logits = self.fc(hidden)
-        return F.sigmoid(logits), hidden
+        return logits, hidden
     
 class GinkaRNNModel(nn.Module):
     def __init__(self, device: torch.device, start_tile=31, width=13, height=13):
@@ -152,7 +153,7 @@ class GinkaRNNModel(nn.Module):
         self.height = height
         self.start_tile = start_tile
         
-        self.rnn_hidden = 2048
+        self.rnn_hidden = 512
         self.tile_classes = 32
         
         # 模型结构
@@ -196,8 +197,7 @@ class GinkaRNNModel(nn.Module):
                 # 处理输出
                 output_logits[:, y, x] = logits[:]
                 hidden = h
-                probs = F.softmax(logits, dim=1)
-                tile_id = torch.argmax(probs, dim=1).detach()
+                tile_id = torch.argmax(logits, dim=1).detach()
                 map[:, y, x] = tile_id[:]
                 now_tile = tile_id if use_self else target_map[:, y, x].detach()