continue to fix Benchmark CI issue #968 (#979)

# Pull Request

## What problem does this PR solve?

Issue Number: Fixed #968

## Possible side effects?

- Performance:

- Backward compatibility:

.circleci/benchmark-config.yml

@@ -44,10 +44,12 @@ jobs:
             mv spu .spu
             # TODO: run all stax_nn models with different optimizers
             docker run --rm --mount type=bind,source="$(pwd)",target=/home/admin/dev/ --network nn-benchmark secretflow/ubuntu-base-ci:latest \
-                "python3 -m pip install -U pip && \
+                "python3 -m pip install --upgrade pip && \
                    python3 -m pip install spu && \
                    cd /home/admin/dev/ && \
-                   python3 -m pip install -r requirements-dev.txt && \
+                   python3 -m pip install -r requirements.txt && \
+                   python3 -m pip install scikit-learn tensorflow keras tensorflow_datasets && \
+                   python3 -m pip install protobuf==5.27.3 && \
                    export PYTHONPATH="${PWD}:$PYTHONPATH" && \
                    bash .circleci/run-nn.sh" | tee benchmark_results.log
       - run:

examples/python/ml/stax_nn/stax_nn.py

@@ -34,7 +34,7 @@
 from keras.datasets import cifar10
 from sklearn.metrics import accuracy_score
 
-import examples.python.ml.stax_nn.models as models
+import models
 import spu.utils.distributed as ppd
 
 parser = argparse.ArgumentParser(description='distributed driver.')