我正在使用TensorFlow 2.0.0及其Keras版本。
我已经成功地训练了我的大型模型，批处理大小为16，但是当我尝试predict时，我从终端运行此错误：
2020-01-27 16:54:01.298606: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cudart64_100.dll
Using TensorFlow backend.
2020-01-27 16:54:05.417321: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library nvcuda.dll
2020-01-27 16:54:05.438414: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties:
name: GeForce RTX 2080 Ti major: 7 minor: 5 memoryClockRate(GHz): 1.665
pciBusID: 0000:01:00.0
2020-01-27 16:54:05.443695: I tensorflow/stream_executor/platform/default/dlopen_checker_stub.cc:25] GPU libraries are statically linked, skip dlopen check.
2020-01-27 16:54:05.448672: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1746] Adding visible gpu devices: 0
2020-01-27 16:54:05.451627: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2
2020-01-27 16:54:05.459043: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties:
name: GeForce RTX 2080 Ti major: 7 minor: 5 memoryClockRate(GHz): 1.665
pciBusID: 0000:01:00.0
2020-01-27 16:54:05.464931: I tensorflow/stream_executor/platform/default/dlopen_checker_stub.cc:25] GPU libraries are statically linked, skip dlopen check.
2020-01-27 16:54:05.468931: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1746] Adding visible gpu devices: 0
2020-01-27 16:54:05.953379: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1159] Device interconnect StreamExecutor with strength 1 edge matrix:
2020-01-27 16:54:05.957463: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1165] 0
2020-01-27 16:54:05.961678: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1178] 0: N
2020-01-27 16:54:05.965985: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1304] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 8685 MB memory) -> physical GPU (device: 0, name: GeForce RTX 2080 Ti, pci bus id: 0000:01:00.0, compute capability: 7.5)
2020-01-27 17:03:36.790437: W tensorflow/core/grappler/optimizers/implementation_selector.cc:310] Skipping optimization due to error while loading function libraries: Invalid argument: Functions '__inference_cudnn_lstm_with_fallback_10134_specialized_for_model_sequential_bidirectional_backward_lstm_StatefulPartitionedCall_at___inference_distributed_function_15069' and '__inference_cudnn_lstm_with_fallback_10134' both implement 'lstm_157a770f-be75-4a59-a000-e5bd1949e961' but their signatures do not match.
2020-01-27 17:03:37.145999: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cublas64_100.dll
2020-01-27 17:03:37.404156: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library cudnn64_7.dll
2020-01-27 17:03:39.152137: E tensorflow/stream_executor/dnn.cc:588] CUDNN_STATUS_EXECUTION_FAILED
in tensorflow/stream_executor/cuda/cuda_dnn.cc(1796): 'cudnnRNNForwardTraining( cudnn.handle(), rnn_desc.handle(), model_dims.max_seq_length, input_desc.handles(), input_data.opaque(), input_h_desc.handle(), input_h_data.opaque(), input_c_desc.handle(), input_c_data.opaque(), rnn_desc.params_handle(), params.opaque(), output_desc.handles(), output_data->opaque(), output_h_desc.handle(), output_h_data->opaque(), output_c_desc.handle(), output_c_data->opaque(), workspace.opaque(), workspace.size(), reserve_space.opaque(), reserve_space.size())'
2020-01-27 17:03:39.152182: F tensorflow/stream_executor/cuda/cuda_dnn.cc:189] Check failed: status == CUDNN_STATUS_SUCCESS (7 vs. 0)Failed to set cuDNN stream.
在jupyter笔记本上运行时，内核崩溃而没有任何错误消息。
对于predict，我还使用与训练中相同的批处理大小（我也尝试了较小的批处理大小，并使用较小的数据集进行预测）。
有什么想法吗？