tofix： 概率性有梯度

brep2sdf/networks/encoder.py

@@ -251,7 +251,6 @@ class BRepFeatureEmbedder(nn.Module):
         vertex_embed = vertex_embed.mean(dim=3)  # [B, F, E, embed_dim]
         
         # 2. 处理边特征
-        logger.info(f"edge_ncs shape: {edge_ncs.shape}")
         edge_embeds = self.edgez_embed(edge_ncs)  # [B, F, E, embed_dim]
         edge_p_embeds = self.edgep_embed(edge_pos)  # [B, F, E, embed_dim]
         
@@ -441,6 +440,13 @@ class BRepToSDF(nn.Module):
         """
         B, Q = query_points.shape[:2]  # B: batch_size, Q: num_queries
         
+        #logger.info(f"query_points requires_grad: {query_points.requires_grad}")
+        #logger.info(f"edge_ncs requires_grad: {edge_ncs.requires_grad}")
+        #logger.info(f"edge_pos requires_grad: {edge_pos.requires_grad}")
+        #logger.info(f"edge_mask requires_grad: {edge_mask.requires_grad}")
+        #logger.info(f"surf_ncs requires_grad: {surf_ncs.requires_grad}")
+        #logger.info(f"surf_pos requires_grad: {surf_pos.requires_grad}")
+
         try:
              # 确保query_points需要梯度
             if not query_points.requires_grad:
@@ -544,13 +550,13 @@ def main():
     
     # 生成测试数据
     test_data = {
-        'edge_ncs': torch.randn(batch_size, max_face, max_edge, num_edge_points, 3),
+        'edge_ncs': torch.randn(batch_size, max_face, max_edge, num_edge_points, 3, requires_grad=True),
-        'edge_pos': torch.randn(batch_size, max_face, max_edge, 6),
+        'edge_pos': torch.randn(batch_size, max_face, max_edge, 6, requires_grad=True),
         'edge_mask': torch.ones(batch_size, max_face, max_edge, dtype=torch.bool),
-        'surf_ncs': torch.randn(batch_size, max_face, num_surf_points, 3),
+        'surf_ncs': torch.randn(batch_size, max_face, num_surf_points, 3, requires_grad=True),
-        'surf_pos': torch.randn(batch_size, max_face, 6),
+        'surf_pos': torch.randn(batch_size, max_face, 6, requires_grad=True),
-        'vertex_pos': torch.randn(batch_size, max_face, max_edge, 2, 3),
+        'vertex_pos': torch.randn(batch_size, max_face, max_edge, 2, 3, requires_grad=True),
-        'query_points': torch.randn(batch_size, 1000, 3)  # 1000个查询点
+        'query_points': torch.randn(batch_size, 1000, 3, requires_grad=True)  # 1000个查询点
     }
     
     # 打印输入数据形状