added test epoch to training model

configs/_base_/datasets/imagenet_bs128_poolformer_small_224_custom.py

@@ -85,6 +85,7 @@ val_dataloader = dict(
     sampler=dict(type='DefaultSampler', shuffle=False),
 )
 
+# If you want standard test, please manually configure the test dataset
 test_dataloader = dict(
     batch_size=32,
     num_workers=5,
@@ -100,7 +101,9 @@ val_evaluator = [
     dict(type='AveragePrecision', prefix='val'),
     dict(type='SingleLabelMetric', prefix='val'),
 ]
+test_evaluator = [
+    dict(type='Accuracy', topk=(1,), prefix='test'),
+    dict(type='AveragePrecision', prefix='test'),
+    dict(type='SingleLabelMetric', prefix='test'),
+]
 
-# If you want standard test, please manually configure the test dataset
-test_dataloader = test_dataloader
-test_evaluator = val_evaluator

docker/Dockerfile

@@ -21,8 +21,8 @@ RUN pip install openmim
 
 # Install MMPretrain
 RUN conda clean --all
-RUN git clone https://github.com/logivations/mmpretrain.git
-WORKDIR ./mmpretrain
+RUN git clone https://github.com/logivations/mmpretrain.git /mmpretrain
+WORKDIR /mmpretrain
 
 # Worked version mim list -> mmpretrain  1.0.0rc7   /workspace/mmpretrain
 RUN mim install --no-cache-dir -e .

tools/train.py

@@ -3,7 +3,7 @@ import argparse
 import os
 import os.path as osp
 from copy import deepcopy
-
+import mmengine
 from mmengine.config import Config, ConfigDict, DictAction
 from mmengine.runner import Runner
 from mmengine.utils import digit_version
@@ -155,6 +155,8 @@ def main():
 
     # start training
     runner.train()
+    metrics = runner.test()
+    mmengine.dump(metrics, os.path.join(args.work_dir, "metrics.json"))
 
 
 if __name__ == '__main__':