Run dvc.yaml pipeline

dvc.lock

@@ -0,0 +1,111 @@
+schema: '2.0'
+stages:
+  data_split:
+    cmd: python src/data_split.py
+    deps:
+    - path: data/pool_data
+      hash: md5
+      md5: 14d187e749ee5614e105741c719fa185.dir
+      size: 18999874
+      nfiles: 183
+    - path: src/data_split.py
+      hash: md5
+      md5: 280fa1684c5496fb9f76ff8c96bd2561
+      size: 1035
+    params:
+      params.yaml:
+        base:
+          random_seed: 42
+        data_split:
+          test_regions:
+          - REGION_1
+    outs:
+    - path: data/test_data
+      hash: md5
+      md5: 1bb16eb1219b47a8bf711ade27c476e4.dir
+      size: 2087761
+      nfiles: 24
+    - path: data/train_data
+      hash: md5
+      md5: a28a7e4d342c27c1d7ad3c17ec1dfa7a.dir
+      size: 16905965
+      nfiles: 158
+  train:
+    cmd: python src/train.py
+    deps:
+    - path: data/train_data
+      hash: md5
+      md5: a28a7e4d342c27c1d7ad3c17ec1dfa7a.dir
+      size: 16905965
+      nfiles: 158
+    - path: src/train.py
+      hash: md5
+      md5: 9db72f1631f53eecb232bc48992c425a
+      size: 2507
+    params:
+      params.yaml:
+        base:
+          random_seed: 42
+        train:
+          valid_pct: 0.1
+          arch: shufflenet_v2_x2_0
+          img_size: 256
+          batch_size: 8
+          fine_tune_args:
+            epochs: 8
+            base_lr: 0.01
+    outs:
+    - path: models/model.pkl
+      hash: md5
+      md5: 63cb30df484bfa1d6ea2cdebac74876c
+      size: 201725
+    - path: models/model.pth
+      hash: md5
+      md5: 07b113fe1ab01de2d8a5453dccacdd3e
+      size: 165147
+    - path: results/train
+      hash: md5
+      md5: a26429ff680d9b01c7e92043821bf41c.dir
+      size: 955
+      nfiles: 5
+  evaluate:
+    cmd: python src/evaluate.py
+    deps:
+    - path: data/test_data
+      hash: md5
+      md5: 1bb16eb1219b47a8bf711ade27c476e4.dir
+      size: 2087761
+      nfiles: 24
+    - path: models/model.pkl
+      hash: md5
+      md5: 63cb30df484bfa1d6ea2cdebac74876c
+      size: 201725
+    - path: src/evaluate.py
+      hash: md5
+      md5: 84d2fd3b371546730396a763a51527a0
+      size: 3322
+    params:
+      params.yaml:
+        base:
+          random_seed: 42
+        evaluate:
+          n_samples_to_save: 10
+    outs:
+    - path: results/evaluate
+      hash: md5
+      md5: 34985c391291e22ac0cfdf6e83ec7268.dir
+      size: 1257936
+      nfiles: 11
+  sagemaker:
+    cmd: cp models/model.pth sagemaker/code/model.pth && cd sagemaker && tar -cpzf
+      model.tar.gz code/ && cd .. && mv sagemaker/model.tar.gz .  && rm sagemaker/code/model.pth
+    deps:
+    - path: models/model.pth
+      hash: md5
+      md5: 07b113fe1ab01de2d8a5453dccacdd3e
+      size: 165147
+    outs:
+    - path: model.tar.gz
+      hash: md5
+      md5: c615b812da343d71fcda080597a24525
+      size: 145778

dvc.yaml

@@ -6,6 +6,7 @@ metrics:
 plots:
 - results/train/plots/metrics:
     x: step
+- results/evaluate/plots/images
 artifacts:
   pool-segmentation:
     path: models/model.pkl
@@ -16,7 +17,7 @@ artifacts:
     - cv
     - segmentation
     - satellite-images
-    - unet
+    - shufflenet_v2_x2_0
 stages:
   data_split:
     cmd: python src/data_split.py