Hello, please ask me about running through your code With my own dataset, the dataset format is the same, that is, you need to change the num_classes in the config.py to 3 num_labels to 12, but why does it fail? It belongs to multi-label sentiment analysis

[taozixiangpichipeach]

Hello, please ask me about running through your code With my own dataset, the dataset format is the same, that is, you need to change the num_classes in the config.py to 3 num_labels to 12, but why does it fail? It belongs to multi-label sentiment analysis The following are the error messages

[2023-07-24 22:30:44] Epoch: [0][420/846] Batch Time 0.062 (0.076) Loss 0.6371 (0.6336) Accuracy 83.333 (77.189) C:\actions-runner_work\pytorch\pytorch\builder\windows\pytorch\aten\src\ATen\native\cuda\Loss.cu:242: block: [0,0,0], thread: [0,0,0] Assertion t >= 0 && t < n_classes failed. Traceback (most recent call last): File "C:\Users\Admin\Desktop\tlt\Sentiment-Analysis-master\train.py", line 194, in main() File "C:\Users\Admin\Desktop\tlt\Sentiment-Analysis-master\train.py", line 167, in main train(epoch, train_data, encoder, optimizer) File "C:\Users\Admin\Desktop\tlt\Sentiment-Analysis-master\train.py", line 55, in train acc += accuracy(outputs[:, :, idx], target_variable[:, idx]) / len(label_names) File "C:\Users\Admin\Desktop\tlt\Sentiment-Analysis-master\utils.py", line 105, in accuracy return correct_total.item() * (100.0 / batch_size) RuntimeError: CUDA error: device-side assert triggered [W CUDAGuardImpl.h:46] Warning: CUDA warning: driver shutting down (function uncheckedGetDevice) [W CUDAGuardImpl.h:62] Warning: CUDA warning: driver shutting down (function uncheckedSetDevice)