 在 TensorFlow.org 上查看
|
 在 Google Colab 中运行 |
 在 GitHub 上查看源代码
|
 下载笔记本 |
TensorFlow Lite Authoring API 提供了一种方式来维护与 TensorFlow Lite 兼容的 tf.function 模型。
安装
import tensorflow as tf
2023-11-07 21:48:11.853226: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:9261] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered 2023-11-07 21:48:11.853276: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:607] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered 2023-11-07 21:48:11.855004: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1515] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
TensorFlow 到 TensorFlow Lite 的兼容性问题
如果您想在设备端使用您的 TF 模型，您需要将其转换为 TFLite 模型，以便从 TFLite 解释器使用它。在转换过程中，您可能会遇到兼容性错误，因为 TFLite 内置运算集不支持 TensorFlow 运算。
这是一个令人讨厌的问题。如何能够更早地检测到该问题呢，比如在模型创作时间？
请注意，以下代码将在调用 converter.convert() 时失败。
@tf.function(input_signature=[
tf.TensorSpec(shape=[None], dtype=tf.float32)
])
def f(x):
return tf.cosh(x)
# Evaluate the tf.function
result = f(tf.constant([0.0]))
print (f"result = {result}")
result = [1.]
# Convert the tf.function
converter = tf.lite.TFLiteConverter.from_concrete_functions(
[f.get_concrete_function()], f)
try:
fb_model = converter.convert()
except Exception as e:
print(f"Got an exception: {e}")
Got an exception: Could not translate MLIR to FlatBuffer. UNKNOWN: /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:695:1: error: 'tf.Cosh' op is neither a custom op nor a flex op
self._concrete_variable_creation_fn = tracing_compilation.trace_function(
^
/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:178:1: note: called from
concrete_function = _maybe_define_function(
^
/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:283:1: note: called from
concrete_function = _create_concrete_function(
^
/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:310:1: note: called from
traced_func_graph = func_graph_module.func_graph_from_py_func(
^
/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/framework/func_graph.py:1059:1: note: called from
func_outputs = python_func(*func_args, **func_kwargs)
^
/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:598:1: note: called from
out = weak_wrapped_fn().__wrapped__(*args, **kwds)
^
/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/autograph_util.py:41:1: note: called from
return api.converted_call(
^
/tmpfs/tmp/ipykernel_34429/885400331.py:5:1: note: called from
/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/weak_tensor_ops.py:88:1: note: called from
return op(*args, **kwargs)
^
/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/gen_math_ops.py:2612:1: note: called from
_, _, _op, _outputs = _op_def_library._apply_op_helper(
^
/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:695:1: note: Error code: ERROR_NEEDS_FLEX_OPS
self._concrete_variable_creation_fn = tracing_compilation.trace_function(
^
<unknown>:0: error: failed while converting: 'main':
Some ops are not supported by the native TFLite runtime, you can enable TF kernels fallback using TF Select. See instructions: https://www.tensorflow.org/lite/guide/ops_select
TF Select ops: Cosh
Details:
tf.Cosh(tensor<?xf32>) -> (tensor<?xf32>) : {device = ""}
2023-11-07 21:48:16.310019: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:378] Ignored output_format.
2023-11-07 21:48:16.310058: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:381] Ignored drop_control_dependency.
Summary on the non-converted ops:
---------------------------------
* Accepted dialects: tfl, builtin, func
* Non-Converted Ops: 1, Total Ops 4, % non-converted = 25.00 %
* 1 TF ops
- tf.Cosh: 1 occurrences (f32: 1)
loc(fused["Cosh:", callsite("Cosh"("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py":695:1) at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":178:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":283:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":310:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/framework/func_graph.py":1059:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py":598:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/autograph_util.py":41:1 at callsite("/tmpfs/tmp/ipykernel_34429/885400331.py":5:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/weak_tensor_ops.py":88:1 at "/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/gen_math_ops.py":2612:1)))))))))]): error: 'tf.Cosh' op is neither a custom op nor a flex op
error: failed while converting: 'main':
Some ops are not supported by the native TFLite runtime, you can enable TF kernels fallback using TF Select. See instructions: https://www.tensorflow.org/lite/guide/ops_select
TF Select ops: Cosh
Details:
tf.Cosh(tensor<?xf32>) -> (tensor<?xf32>) : {device = ""}
简单的目标感知创作用法
我们引入了 Authoring API 来检测模型创作期间的 TensorFlow Lite 兼容性问题。
您只需添加 @tf.lite.experimental.authoring.compatible 修饰器来封装您的 tf.function 模型，以检查 TFLite 兼容性。
之后，当您评估模型时，将自动检查兼容性。
@tf.lite.experimental.authoring.compatible
@tf.function(input_signature=[
tf.TensorSpec(shape=[None], dtype=tf.float32)
])
def f(x):
return tf.cosh(x)
# Evaluate the tf.function
result = f(tf.constant([0.0]))
print (f"result = {result}")
COMPATIBILITY WARNING: op 'tf.Cosh' require(s) "Select TF Ops" for model conversion for TensorFlow Lite. https://www.tensorflow.org/lite/guide/ops_select
Op: tf.Cosh
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/gen_math_ops.py:2612
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/weak_tensor_ops.py:88
- /tmpfs/tmp/ipykernel_34429/885400331.py:5
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/autograph_util.py:41
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:598
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/framework/func_graph.py:1059
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:310
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:283
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:178
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:695
result = [1.]
2023-11-07 21:48:16.415556: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:378] Ignored output_format.
2023-11-07 21:48:16.415597: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:381] Ignored drop_control_dependency.
Summary on the non-converted ops:
---------------------------------
* Accepted dialects: tfl, builtin, func
* Non-Converted Ops: 1, Total Ops 4, % non-converted = 25.00 %
* 1 TF ops
- tf.Cosh: 1 occurrences (f32: 1)
loc(fused["Cosh:", callsite("Cosh"("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py":695:1) at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":178:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":283:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":310:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/framework/func_graph.py":1059:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py":598:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/autograph_util.py":41:1 at callsite("/tmpfs/tmp/ipykernel_34429/885400331.py":5:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/weak_tensor_ops.py":88:1 at "/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/gen_math_ops.py":2612:1)))))))))]): error: 'tf.Cosh' op is neither a custom op nor a flex op
error: failed while converting: 'main':
Some ops are not supported by the native TFLite runtime, you can enable TF kernels fallback using TF Select. See instructions: https://www.tensorflow.org/lite/guide/ops_select
TF Select ops: Cosh
Details:
tf.Cosh(tensor<?xf32>) -> (tensor<?xf32>) : {device = ""}
如果发现任何 TensorFlow Lite 兼容性问题，它将显示 COMPATIBILITY WARNING 或 COMPATIBILITY ERROR 以及有问题的运算的确切位置。在本例中，它显示了 tf.unction 模型中 tf.Cosh 运算的位置。
您还可以使用 <function_name>.get_compatibility_log() 方法检查兼容性日志。
compatibility_log = '\n'.join(f.get_compatibility_log())
print (f"compatibility_log = {compatibility_log}")
compatibility_log = COMPATIBILITY WARNING: op 'tf.Cosh' require(s) "Select TF Ops" for model conversion for TensorFlow Lite. https://www.tensorflow.org/lite/guide/ops_select Op: tf.Cosh - /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/gen_math_ops.py:2612 - /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/weak_tensor_ops.py:88 - /tmpfs/tmp/ipykernel_34429/885400331.py:5 - /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/autograph_util.py:41 - /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:598 - /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/framework/func_graph.py:1059 - /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:310 - /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:283 - /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:178 - /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:695
引发不兼容性异常
您可以为 @tf.lite.experimental.authoring.compatible 修饰器提供一个选项。当您尝试评估经过修饰的模型时，raise_exception 选项会引发异常。
@tf.lite.experimental.authoring.compatible(raise_exception=True)
@tf.function(input_signature=[
tf.TensorSpec(shape=[None], dtype=tf.float32)
])
def f(x):
return tf.cosh(x)
# Evaluate the tf.function
try:
result = f(tf.constant([0.0]))
print (f"result = {result}")
except Exception as e:
print(f"Got an exception: {e}")
COMPATIBILITY WARNING: op 'tf.Cosh' require(s) "Select TF Ops" for model conversion for TensorFlow Lite. https://www.tensorflow.org/lite/guide/ops_select
Op: tf.Cosh
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/gen_math_ops.py:2612
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/weak_tensor_ops.py:88
- /tmpfs/tmp/ipykernel_34429/885400331.py:5
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/autograph_util.py:41
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:598
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/framework/func_graph.py:1059
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:310
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:283
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:178
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:695
Got an exception: CompatibilityException at <tensorflow.python.eager.polymorphic_function.polymorphic_function.Function object at 0x7f919cfbfb50>
2023-11-07 21:48:16.539202: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:378] Ignored output_format.
2023-11-07 21:48:16.539249: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:381] Ignored drop_control_dependency.
Summary on the non-converted ops:
---------------------------------
* Accepted dialects: tfl, builtin, func
* Non-Converted Ops: 1, Total Ops 4, % non-converted = 25.00 %
* 1 TF ops
- tf.Cosh: 1 occurrences (f32: 1)
loc(fused["Cosh:", callsite("Cosh"("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py":695:1) at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":178:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":283:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":310:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/framework/func_graph.py":1059:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py":598:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/autograph_util.py":41:1 at callsite("/tmpfs/tmp/ipykernel_34429/885400331.py":5:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/weak_tensor_ops.py":88:1 at "/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/gen_math_ops.py":2612:1)))))))))]): error: 'tf.Cosh' op is neither a custom op nor a flex op
error: failed while converting: 'main':
Some ops are not supported by the native TFLite runtime, you can enable TF kernels fallback using TF Select. See instructions: https://www.tensorflow.org/lite/guide/ops_select
TF Select ops: Cosh
Details:
tf.Cosh(tensor<?xf32>) -> (tensor<?xf32>) : {device = ""}
指定 "Select TF ops" 用法
如果您已了解 Select Tf op 用法，可以通过设置 converter_target_spec 将其用于 Authoring API。它与您将在 tf.lite.TFLiteConverter API 中使用的对象相同。
target_spec = tf.lite.TargetSpec()
target_spec.supported_ops = [
tf.lite.OpsSet.TFLITE_BUILTINS,
tf.lite.OpsSet.SELECT_TF_OPS,
]
@tf.lite.experimental.authoring.compatible(converter_target_spec=target_spec, raise_exception=True)
@tf.function(input_signature=[
tf.TensorSpec(shape=[None], dtype=tf.float32)
])
def f(x):
return tf.cosh(x)
# Evaluate the tf.function
result = f(tf.constant([0.0]))
print (f"result = {result}")
result = [1.]
2023-11-07 21:48:16.665435: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:378] Ignored output_format.
2023-11-07 21:48:16.665473: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:381] Ignored drop_control_dependency.
Summary on the non-converted ops:
---------------------------------
* Accepted dialects: tfl, builtin, func
* Non-Converted Ops: 1, Total Ops 4, % non-converted = 25.00 %
* 1 TF ops
- tf.Cosh: 1 occurrences (f32: 1)
2023-11-07 21:48:16.680080: W tensorflow/compiler/mlir/lite/flatbuffer_export.cc:2921] TFLite interpreter needs to link Flex delegate in order to run the model since it contains the following Select TFop(s):
Flex ops: FlexCosh
Details:
tf.Cosh(tensor<?xf32>) -> (tensor<?xf32>) : {device = ""}
See instructions: https://www.tensorflow.org/lite/guide/ops_select
检查 GPU 兼容性
如果您希望确保您的模型与 TensorFlow Lite 的 GPU 委托兼容，您可以设置 tf.lite.TargetSpec 的 experimental_supported_backends。
以下示例展示如何确保您的模型的 GPU 委托兼容性。请注意，此模型存在兼容性问题，因为它使用带有 tf.slice 算子和不受支持的 tf.cosh 算子的二维张量。您将看到两个带有位置信息的 COMPATIBILITY WARNING。
target_spec = tf.lite.TargetSpec()
target_spec.supported_ops = [
tf.lite.OpsSet.TFLITE_BUILTINS,
tf.lite.OpsSet.SELECT_TF_OPS,
]
target_spec.experimental_supported_backends = ["GPU"]
@tf.lite.experimental.authoring.compatible(converter_target_spec=target_spec)
@tf.function(input_signature=[
tf.TensorSpec(shape=[4, 4], dtype=tf.float32)
])
def func(x):
y = tf.cosh(x)
return y + tf.slice(x, [1, 1], [1, 1])
result = func(tf.ones(shape=(4,4), dtype=tf.float32))
'tfl.slice' op is not GPU compatible: SLICE supports for 3 or 4 dimensional tensors only, but node has 2 dimensional tensors.
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py:1260
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/util/traceback_utils.py:150
- /tmpfs/tmp/ipykernel_34429/3833138856.py:13
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/autograph_util.py:41
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:598
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/framework/func_graph.py:1059
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:310
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:283
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:178
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:695
'tf.Cosh' op is not GPU compatible: Not supported custom op FlexCosh
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/gen_math_ops.py:2612
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/weak_tensor_ops.py:88
- /tmpfs/tmp/ipykernel_34429/3833138856.py:12
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/autograph_util.py:41
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:598
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/framework/func_graph.py:1059
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:310
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:283
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py:178
- /tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py:695
COMPATIBILITY WARNING: op 'tf.Cosh, tfl.slice' aren't compatible with TensorFlow Lite GPU delegate. https://www.tensorflow.org/lite/performance/gpu
2023-11-07 21:48:16.820800: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:378] Ignored output_format.
2023-11-07 21:48:16.820846: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:381] Ignored drop_control_dependency.
Summary on the non-converted ops:
---------------------------------
* Accepted dialects: tfl, builtin, func
* Non-Converted Ops: 2, Total Ops 7, % non-converted = 28.57 %
* 1 ARITH ops, 1 TF ops
- arith.constant: 1 occurrences (i32: 1)
- tf.Cosh: 1 occurrences (f32: 1)
(f32: 1)
(f32: 1)
2023-11-07 21:48:16.837705: W tensorflow/compiler/mlir/lite/flatbuffer_export.cc:2921] TFLite interpreter needs to link Flex delegate in order to run the model since it contains the following Select TFop(s):
Flex ops: FlexCosh
Details:
tf.Cosh(tensor<4x4xf32>) -> (tensor<4x4xf32>) : {device = ""}
See instructions: https://www.tensorflow.org/lite/guide/ops_select
loc(fused["Cosh:", callsite("Cosh"("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py":695:1) at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":178:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":283:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":310:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/framework/func_graph.py":1059:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py":598:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/autograph_util.py":41:1 at callsite("/tmpfs/tmp/ipykernel_34429/3833138856.py":12:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/weak_tensor_ops.py":88:1 at "/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/ops/gen_math_ops.py":2612:1)))))))))]): error: 'tf.Cosh' op is not GPU compatible: Not supported custom op FlexCosh
loc(fused["Slice:", callsite("Slice"("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py":695:1) at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":178:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":283:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/tracing_compilation.py":310:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/framework/func_graph.py":1059:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/polymorphic_function.py":598:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/eager/polymorphic_function/autograph_util.py":41:1 at callsite("/tmpfs/tmp/ipykernel_34429/3833138856.py":13:1 at callsite("/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/util/traceback_utils.py":150:1 at "/tmpfs/src/tf_docs_env/lib/python3.9/site-packages/tensorflow/python/util/dispatch.py":1260:1)))))))))]): error: 'tfl.slice' op is not GPU compatible: SLICE supports for 3 or 4 dimensional tensors only, but node has 2 dimensional tensors.
阅读更多
有关更多信息，请参阅：