PyTorch回归训练报错：矩阵形状不匹配且输入维度异常

我用PyTorch对5个形状均为10×3361的数据集训练回归模型，运行时返回错误：

RuntimeError: mat1 and mat2 shapes cannot be multiplied (10x[some multiple of 1499, varying from dataset to dataset] and 3361x64)

我知道这是神经网络第一层维度不匹配导致的，但疑惑为什么数据集维度会变成1499的倍数且随数据集变化，明明所有数据集形状应该一致。

以下是相关代码：

神经网络模型定义 #

# define neural network model for regression
class Regression(nn.Module): 

  def __init__(self):
    super().__init__()
    self.layers = nn.Sequential(
      nn.Linear(3361, 64), 
      nn.ReLU(),
      nn.Linear(64, 32),
      nn.ReLU(),
      nn.Linear(32, 1)
    )

  def forward(self, x):
    '''
      Forward pass
    '''
    return self.layers(x)

训练循环 #

def training_loop(n_epochs, train_loader):
# training loop

    for epoch in range(n_epochs):

        # Set current loss value
        current_loss = 0.0
        # Iterate over the DataLoader for training data
        for i, data in enumerate(train_loader, 0):
          # Get and prepare inputs
          inputs, targets = data
          inputs, targets = inputs.float(), targets.float()
          targets = targets.reshape((targets.shape[0],1))
            # Zero the gradients
          optimizer.zero_grad()
          # Perform forward pass
          outputs = model(inputs)
          
          # Compute loss
          loss = loss_function(outputs, targets)
      
          # Perform backward pass
          loss.backward()
      
          # Perform optimization
          optimizer.step()
    return model

调用上下文 #

exo_file="exo_data_rp2500_set.csv"
exo_data=pd.read_csv(exo_file)
    
input_size = exo_data.shape[1]
batch_size = 10
model=Regression()
n_epochs=100
# loss function and optimizer
loss_function =nn.MSELoss()
# mean square error
optimizer = torch.optim.Adam(model.parameters(), lr=1e-4)

...
metrics=[#list of the nine categories of features I'm investigating]

for metric in metrics:
#split and format data
    exo_target=exo_metrics[metric]
    X_train, X_test, y_train, y_test = train_test_split(exo_spectra,exo_target, test_size=0.2, random_state=23) 
    X_train_tensor = torch.from_numpy(X_train)
    X_train_tensor = torch.tensor(X_train_tensor,dtype=torch.float32)
    y_train_tensor = torch.from_numpy(y_train.values)
    y_train_tensor = torch.tensor(y_train_tensor,dtype=torch.float32)
    X_test_tensor = torch.from_numpy(X_test)
    X_test_tensor = torch.tensor(X_test_tensor,dtype=torch.float32)
    y_test_tensor= torch.from_numpy(y_test.values)
    y_test_tensor = torch.tensor(y_test_tensor,dtype=torch.float32)
    # load data
    train_dataset = TensorDataset(X_train_tensor, y_train_tensor)
    train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True) 
    # train data and assess accuracy
    model=training_loop(n_epochs,train_loader)
    y_pred = model(X_test_tensor)
    model_score=r2_score(y_test_tensor.detach().numpy(),y_pred.detach().numpy())
    scores.append(model_score)

1. 模型输入维度硬编码，未匹配实际数据特征数 #

你的模型第一层nn.Linear(3361, 64)直接写死了输入特征数为3361，但实际训练时输入数据的特征数是1499的倍数，和模型期望的3361不匹配。你声称所有数据集都是10×3361，但实际可能存在以下问题：

• 读取CSV时误读了额外列（比如索引列、冗余特征列），导致特征数变成1499的倍数；
• 在代码省略的...部分处理exo_spectra时，错误修改了特征维度（比如合并特征、转置矩阵、误删列），使得特征数偏离3361。

2. 模型未在每次训练任务前重新初始化 #

你在metrics循环外只初始化了一次模型，后续训练不同metric时直接复用旧模型。如果某次循环中exo_spectra的特征数意外变化，就会触发维度不匹配的错误。即使所有数据集形状一致，这种复用模型的方式也会埋下隐患。

3. 全局变量滥用导致的逻辑混乱 #

训练循环training_loop直接使用了全局的model、optimizer和loss_function，而非通过参数传入。这种写法会导致循环训练不同任务时，模型和优化器的状态混乱，一旦某次输入维度异常，就会报错。

1. 动态设置模型输入维度
   修改模型定义，用参数接收输入特征数，避免硬编码：

   class Regression(nn.Module): 
     def __init__(self, input_size):
       super().__init__()
       self.layers = nn.Sequential(
         nn.Linear(input_size, 64), 
         nn.ReLU(),
         nn.Linear(64, 32),
         nn.ReLU(),
         nn.Linear(32, 1)
       )
     def forward(self, x):
       return self.layers(x)
   

2. 每次训练前重新初始化模型和优化器
   在metrics循环内，根据当前数据的实际特征数初始化模型，并重新创建优化器：

   for metric in metrics:
       # ... 数据处理代码 ...
       # 获取当前训练数据的实际特征数
       current_input_size = X_train.shape[1]
       # 重新初始化模型
       model = Regression(current_input_size)
       # 重新初始化优化器
       optimizer = torch.optim.Adam(model.parameters(), lr=1e-4)
       # 训练模型（需修改training_loop接收参数）
       model = training_loop(n_epochs, train_loader, model, optimizer, loss_function)
       # ... 评估代码 ...
   

3. 修改训练循环，避免使用全局变量
   更新training_loop函数，让它接收模型、优化器和损失函数作为参数：

   def training_loop(n_epochs, train_loader, model, optimizer, loss_function):
       for epoch in range(n_epochs):
           current_loss = 0.0
           for i, data in enumerate(train_loader, 0):
               inputs, targets = data
               inputs, targets = inputs.float(), targets.float()
               targets = targets.reshape((targets.shape[0],1))
               optimizer.zero_grad()
               outputs = model(inputs)
               loss = loss_function(outputs, targets)
               loss.backward()
               optimizer.step()
       return model
   

4. 检查数据形状
   在数据处理部分添加打印，确认每次循环中exo_spectra、X_train的形状是否符合预期：

   print(f"exo_spectra形状: {exo_spectra.shape}")
   print(f"X_train形状: {X_train.shape}")
   

内容的提问来源于stack exchange，提问作者Tessa