mbkgh
commented
3 years ago With another model, compiled on the same machine and using the same docker version, it works without any issue. The difference is the model itself and how the parameter are passed to tensorflow/serving:
This works fine: cd \tmp\tfserving set-variable -Name "TESTDATA" -Value "$(pwd)/serving/tensorflow_serving/servables/tensorflow/testdata" docker run -t --rm -p 8501:8501 -v "$TESTDATA/wisconsin:/models/wisconsin" -e MODEL_NAME=wisconsin tensorflow/serving
curl -d "{\"instances\": [[17.99,10.38,122.8,1001,0.1184,0.2776,0.3001,0.1471,0.2419,0.07871]]}" -X POST http://localhost:8501/v1/models/wisconsin:predict returns { "predictions": [[1.0] ] }
As requested by @mohantym, issue posted in addition here:
Original issue: (https://github.com/tensorflow/tensorflow/issues/52536)
I am using tensorflow-serving which runs with the build-in examples well and without issues: cd c:\tmp\tfserving PS C:\tmp\tfserving> C:\programme\git\bin\git clone https://github.com/tensorflow/serving set-variable -Name "TESTDATA" -Value "$(pwd)/serving/tensorflow_serving/servables/tensorflow/testdata" docker run -t --rm -p 8501:8501 -v "$TESTDATA/saved_model_half_plus_two_cpu:/models/half_plus_two" -e MODEL_NAME=half_plus_two tensorflow/serving curl -d "{"instances": [1.0, 2.0, 5.0]}" -X POST http://localhost:8501/v1/models/half_plus_two:predict
However, by trying to run the following model in a recent docker container it causes: "Invalid reduction dimension (1 for input with 1 dimension(s)\
Steps to verify: I used the example code ... https://keras.io/examples/structured_data/structured_data_classification_from_scratch/ and at the end saved the model model.save('my-model.tf')
and put it into the serving directory where it was recognized: C:\tmp\tfserving\serving\tensorflow_serving\servables\tensorflow\testdata cd \tmp\tfserving set-variable -Name "TESTDATA" -Value "$(pwd)/serving/tensorflow_serving/servables/tensorflow/testdata" docker run -t --rm -p 8501:8501 -v "$TESTDATA/my_model:/models/my_model" -e MODEL_NAME=my_model tensorflow/serving
curl http://localhost:8501/v1/models/my_model { "model_version_status": [ { "version": "1", "state": "AVAILABLE", "status": { "error_code": "OK", [...]
However, trying to run a prediction
curl -d "{ "instances": [ {"age": 50,"sex": 1,"cp": 1,"trestbps": 145,"chol": 133,"fbs": 1,"restecg": 2,"thalach": 150,"exang": 0,"oldpeak": 2.3,"slope": 3,"ca": 0,"thal": "fixed" } ]}" -X POST http://localhost:8501/v1/models/my_model:predict
... returns:
"error": "Invalid reduction dimension (1 for input with 1 dimension(s)\n\t [[{{node model/integer_lookup_5/bincount/Max}}]]"
System information
Running on Windows with no GPU-Support. No further changes were made on the example code OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Windows 10 Mobile device (e.g. iPhone 8, Pixel 2, Samsung Galaxy) if the issue happens on mobile device: n.a. TensorFlow installed from (source or binary): binary TensorFlow version (use command below): v2.6.0-rc2-32-g919f693420e 2.6.0 Python version: 3.8.11 (default, Aug 6 2021, 09:57:55) [MSC v.1916 64 bit (AMD64)] Bazel version (if compiling from source): have also tried to compile on linux with bazel but with the same result GCC/Compiler version (if compiling from source): n.a. CUDA/cuDNN version: n.a. GPU model and memory: n.a.