log change

code/conversion/run.py

@@ -123,10 +123,16 @@ class ReconstructionRunner:
             self.network.train()  # 设置网络为训练模式
             self.adjust_learning_rate(epoch)  # 调整学习率
             nonmnfld_pnts = self.sampler.get_points(mnfld_pnts.unsqueeze(0), mnfld_sigma.unsqueeze(0)).squeeze()  # 生成非流形点
+            # nonmnfld_pnts: torch.Size([18432, 3])
+            #logger.info(f"mnfld_pnts: {mnfld_pnts.shape}") mnfld_pnts: torch.Size([16384, 3])
+            #logger.info(f"mnfld_sigma: {mnfld_sigma.shape}") mnfld_sigma: torch.Size([16384])
 
             # forward pass
             mnfld_pred_all = self.network(mnfld_pnts)  # 进行前向传播，计算流形点的预测值
             nonmnfld_pred_all = self.network(nonmnfld_pnts)  # 进行前向传播，计算非流形点的预测值
+            #logger.info(f"mnfld_pred_all: {mnfld_pred_all.shape}")
+            #logger.info(f"nonmnfld_pred_all: {nonmnfld_pred_all.shape}")
+
             mnfld_pred = mnfld_pred_all[:,0]  # 提取流形预测结果
             nonmnfld_pred = nonmnfld_pred_all[:,0]  # 提取非流形预测结果
             loss = 0.0  # 初始化损失为 0
@@ -166,6 +172,7 @@ class ReconstructionRunner:
             # last patch
             all_fi[(n_branch - 1) * n_patch_batch:, 0] = mnfld_pred_all[(n_branch - 1) * n_patch_batch:, n_branch]  # 填充最后一个分支的流形预测值
 
+            #logger.info(f"all_fi: {all_fi.shape}")
             # manifold loss for patches
             mnfld_loss_patch = torch.zeros(1).cuda()  # 初始化补丁流形损失
             if not args.ab == 'patch':  # 检查是否为补丁损失