mobileone-s1 prediction scores are low

[Phil-Lin]

分支

main 分支 (mmpretrain 版本)

描述该错误

hello，I pre-trained the model mobileone-s1 in Imagenet 1k，and get TOP ACC 75.2 in testSet of Imagenet 1k and loss during training is normal. I deploy the onnx model by mmdeploy for my pretrain checkpoint, and the prediction scores of model are low, prediction scores are low.

环境信息

{'sys.platform': 'linux', 'Python': '3.8.18 (default, Sep 11 2023, 13:40:15) [GCC 11.2.0]', 'CUDA available': True, 'numpy_random_seed': 2147483648, 'GPU 0,1,2,3,4,5,6,7': 'NVIDIA GeForce RTX 4090', 'CUDA_HOME': '/usr/local/cuda:', 'GCC': 'gcc (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0', 'PyTorch': '1.10.1+cu111', 'TorchVision': '0.11.2+cu111', 'OpenCV': '4.8.1', 'MMEngine': '0.9.0', 'MMCV': '2.0.1', 'MMPreTrain': '1.1.0+'}

其他信息

No response

[Phil-Lin]

There is also a problem that the training is completed without generating the best accuracy pth file（best_accuracy_top1.pth）