RuntimeError: CUDA out of memory；ValueError: Input 0 is incompatible with layer sequential

RuntimeError: CUDA out of memory. Tried to allocate 22.00 MiB (GPU 0; 6.00 GiB total capacity; 260.44 MiB already allocated; 0 bytes free; 262.00 MiB reserved in total by PyTorch)

GPU 0 bytes free 看起来非常不正常

用的是yolo网络，经过修改以后程度应该是先调用torch 然后还用了tensorflow

在确保了torch中GPU可用，在运行的代码里增加一行

import os
os.environ["CUDA_VISIBLE_DEVICES"] = '1'

解决了问题

但又出了新的问题

expected shape=(None, 224, 224, 3), found shape=(32, 224, 3)

增加一行代码 reshape一下

img = np.reshape(img, [1, 224, 224, 3])

解决问题