你好@nimashiri!

感谢你分享了关于gen_array_ops.quantize_and_dequantize_v4在2.11版本和Ubuntu上的使用方法。我在使用TF 2.11、Ubuntu 18和Cuda 11.2 + CudNN 8.1时遇到了以下错误：

Error:Value for attr 'T' of uint64 is not in the list of allowed values: bfloat16, half, float, double
	; NodeDef: {{node QuantizeAndDequantizeV4}}; Op<name=QuantizeAndDequantizeV4; signature=input:T, input_min:T, input_max:T -> output:T; attr=signed_input:bool,default=true; attr=num_bits:int,default=8; attr=range_given:bool,default=false; attr=T:type,allowed=[DT_BFLOAT16, DT_HALF, DT_FLOAT, DT_DOUBLE]; attr=round_mode:string,default="HALF_TO_EVEN",allowed=["HALF_TO_EVEN", "HALF_UP"]; attr=narrow_range:bool,default=false; attr=axis:int,default=-1> [Op:QuantizeAndDequantizeV4]

你能在Cuda 11.2、CudNN 8.1和TF 2.11的情况下检查一下吗？或者在Cuda 11.7、CudNN 8.2和TF-nightly的情况下也检查一下，然后告诉我们结果。如果可以的话，请也分享一下操作系统和CUDA的详细信息(Ubuntu Arm 64或X86_64,Cuda 11.2或11.7)。

非常感谢！

从Colab:

Error:Value for attr 'T' of uint64 is not in the list of allowed values: bfloat16, half, float, double
	; NodeDef: {{node QuantizeAndDequantizeV4}}; Op<name=QuantizeAndDequantizeV4; signature=input:T, input_min:T, input_max:T -> output:T; attr=signed_input:bool,default=true; attr=num_bits:int,default=8; attr=range_given:bool,default=false; attr=T:type,allowed=[DT_BFLOAT16, DT_HALF, DT_FLOAT, DT_DOUBLE]; attr=round_mode:string,default="HALF_TO_EVEN",allowed=["HALF_TO_EVEN", "HALF_UP"]; attr=narrow_range:bool,default=false; attr=axis:int,default=-1> [Op:QuantizeAndDequantizeV4]

Colab paid products - Cancel contracts here

从终端：

2023-01-09 10:11:50.975850: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2023-01-09 10:11:51.507938: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer.so.7'; dlerror: libnvinfer.so.7: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: :/home/nimashiri/anaconda3/envs/cuda11.2/lib/
2023-01-09 10:11:51.508097: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer_plugin.so.7'; dlerror: libnvinfer_plugin.so.7: cannot open shared object file: No such file or directory; LD_LIBRARY_PATH: :/home/nimashiri/anaconda3/envs/cuda11.2/lib/
2023-01-09 10:11:51.508104: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Cannot dlopen some TensorRT libraries. If you would like to use Nvidia GPU with TensorRT, please make sure the missing libraries mentioned above are installed properly.
2023-01-09 10:11:51.997861: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2023-01-09 10:11:52.023990: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2023-01-09 10:11:52.024113: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2023-01-09 10:11:52.024507: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2023-01-09 10:11:52.025015: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2023-01-09 10:11:52.025121: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2023-01-09 10:11:52.025214: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2023-01-09 10:11:52.639793: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2023-01-09 10:11:52.639941: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2023-01-09 10:11:52.640069: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2023-01-09 10:11:52.640201: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1613] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 4125 MB memory:  -> device: 0, name: NVIDIA GeForce GTX 1660 Ti, pci bus id: 0000:01:00.0, compute capability: 7.5
Error:can't convert negative int to unsigned

简单地通过VSCODE运行位于上述代码所在位置的脚本：

2023-01-09 10:13:02.292944: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2023-01-09 10:13:02.852646: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer.so.7'; dlerror: libnvinfer.so.7: cannot open shared object file: No such file or directory
2023-01-09 10:13:02.852689: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer_plugin.so.7'; dlerror: libnvinfer_plugin.so.7: cannot open shared object file: No such file or directory
2023-01-09 10:13:02.852696: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Cannot dlopen some TensorRT libraries. If you would like to use Nvidia GPU with TensorRT, please make sure the missing libraries mentioned above are installed properly.
2023-01-09 10:13:03.378474: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2023-01-09 10:13:03.415108: W tensorflow/compiler/xla/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libcudnn.so.8'; dlerror: libcudnn.so.8: cannot open shared object file: No such file or directory
2023-01-09 10:13:03.415132: W tensorflow/core/common_runtime/gpu/gpu_device.cc:1934] Cannot dlopen some GPU libraries. Please make sure the missing libraries mentioned above are installed properly if you would like to use GPU. Follow the guide at https://www.tensorflow.org/install/gpu for how to download and setup the required libraries for your platform.
Skipping registering GPU devices...
2023-01-09 10:13:03.415379: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
Error:can't convert negative int to unsigned

通过subprocess.run()运行：

2023-01-09 10:15:16.155685: F ./tensorflow/python/eager/pywrap_tensor_conversion.h:58] Check failed: !PyErr_Occurred() 
Aborted

@learning-to-play