fix accuracy computation

dvc.lock

@@ -32,8 +32,8 @@ stages:
       nfiles: 60001
     - path: train.py
       hash: md5
-      md5: aabaf1a407badf48c97b14a69b0072ea
-      size: 3420
+      md5: d32feb4bad10a201fdde2b6424238d16
+      size: 3401
     params:
       params.yaml:
         train:
@@ -44,7 +44,7 @@ stages:
     outs:
     - path: model.pt
       hash: md5
-      md5: 8ead2a7cd52d70b359d3cdc3df5e43e3
+      md5: a0823af18fbf58ff562d804a02dca7d2
       size: 102592994
   evaluate:
     cmd: python evaluate.py
@@ -56,11 +56,11 @@ stages:
       nfiles: 60001
     - path: evaluate.py
       hash: md5
-      md5: 8a9a2e95a6b64e632a4f2feac62d294b
-      size: 1473
+      md5: 8eb9acfdc80a5ca6b7bd782643a0997b
+      size: 1483
     - path: model.pt
       hash: md5
-      md5: 8ead2a7cd52d70b359d3cdc3df5e43e3
+      md5: a0823af18fbf58ff562d804a02dca7d2
       size: 102592994
     params:
       params.yaml:

dvclive/eval/metrics.json

@@ -1,3 +1,3 @@
 {
-    "test_acc": 0.7336809039115906
+    "test_acc": 0.20589999854564667
 }

dvclive/eval/plots/metrics/test_acc.tsv

@@ -1,2 +1,2 @@
 step	test_acc
-0	0.7336809039115906
+0	0.20589999854564667

dvclive/eval/report.md

@@ -10,6 +10,6 @@ metrics.json
 
 |   test_acc |
 |------------|
-|   0.733681 |
+|     0.2059 |
 
 ![static/test_acc](static/test_acc.png)

dvclive/train/metrics.json

@@ -1,7 +1,7 @@
 {
-    "train_loss": 2.2422571182250977,
-    "train_acc": 0.7347080707550049,
-    "valid_loss": 2.3184266090393066,
-    "valid_acc": 0.7381500005722046,
+    "train_loss": 2.2407546043395996,
+    "train_acc": 0.22537143528461456,
+    "valid_loss": 2.324389696121216,
+    "valid_acc": 0.20746666193008423,
     "step": 4
 }

dvclive/train/plots/metrics/train_acc.tsv

@@ -1,6 +1,6 @@
 step	train_acc
-0	0.6712241768836975
-1	0.6976224184036255
-2	0.7157850861549377
-3	0.7277812957763672
-4	0.7347080707550049
+0	0.0997999981045723
+1	0.13202856481075287
+2	0.16345714032649994
+3	0.19614285230636597
+4	0.22537143528461456

dvclive/train/plots/metrics/train_loss.tsv

@@ -1,6 +1,6 @@
 step	train_loss
-0	3.0726168155670166
-1	2.7409346103668213
-2	2.5224294662475586
-3	2.364570140838623
-4	2.2422571182250977
+0	3.096755027770996
+1	2.749246120452881
+2	2.522728681564331
+3	2.3631479740142822
+4	2.2407546043395996

dvclive/train/plots/metrics/valid_acc.tsv

@@ -1,6 +1,6 @@
 step	valid_acc
-0	0.6918894052505493
-1	0.7131190896034241
-2	0.7261338233947754
-3	0.7339118123054504
-4	0.7381500005722046
+0	0.10953333228826523
+1	0.13466666638851166
+2	0.15913332998752594
+3	0.18406666815280914
+4	0.20746666193008423

dvclive/train/plots/metrics/valid_loss.tsv

@@ -1,6 +1,6 @@
 step	valid_loss
-0	2.890321969985962
-1	2.669679880142212
-2	2.5183584690093994
-3	2.4061686992645264
-4	2.3184266090393066
+0	2.919710636138916
+1	2.68220853805542
+2	2.524806022644043
+3	2.4115779399871826
+4	2.324389696121216

dvclive/train/report.md

@@ -10,7 +10,7 @@ metrics.json
 
 |   train_loss |   train_acc |   valid_loss |   valid_acc |   step |
 |--------------|-------------|--------------|-------------|--------|
-|      2.24226 |    0.734708 |      2.31843 |     0.73815 |      4 |
+|      2.24075 |    0.225371 |      2.32439 |    0.207467 |      4 |
 
 ![static/valid_loss](static/valid_loss.png)
 

evaluate.py

@@ -8,7 +8,7 @@ import yaml
 
 import torch
 from torch.utils.data import DataLoader
-from torchmetrics.classification import MulticlassAccuracy
+from torchmetrics.classification import Accuracy
 from dvclive import Live
 
 from utils.dataset import ProcessedDataset
@@ -33,7 +33,7 @@ def evaluate(params_path: str = 'params.yaml') -> None:
     net.to(device)
     net.eval()
 
-    metric = MulticlassAccuracy(num_classes=10, top_k=1, average='weighted')
+    metric = Accuracy(task='multiclass', num_classes=10)
     metric.to(device)
 
     with Live(dir='dvclive/eval', report='md') as live:
@@ -43,7 +43,7 @@ def evaluate(params_path: str = 'params.yaml') -> None:
             for data in test_dataloader:
                 inputs, labels = data[0].to(device), data[1].to(device)
                 outputs = net(inputs)
-                _ = metric(outputs, labels)
+                _ = metric(outputs.topk(k=1, dim=1)[1], labels.topk(k=1, dim=1)[1])
             test_acc = metric.compute()
 
         print(f'test_acc:{test_acc}')

train.py

@@ -10,7 +10,7 @@ import torch
 from rich.progress import track
 from torch.utils.data import DataLoader
 from torchvision.models import resnet50
-from torchmetrics.classification import MulticlassAccuracy
+from torchmetrics.classification import Accuracy
 from dvclive import Live
 
 from utils.dataset import ProcessedDataset
@@ -45,7 +45,7 @@ def train(params_path: str = 'params.yaml') -> None:
 
     criterion = torch.nn.CrossEntropyLoss()
     optimizer = torch.optim.SGD(net.parameters(), lr=learning_rate)
-    metric = MulticlassAccuracy(num_classes=10, top_k=1, average='weighted')
+    metric = Accuracy(task='multiclass', num_classes=10)
     metric.to(device)
 
     with Live(dir='dvclive/train', report='md') as live:
@@ -63,7 +63,7 @@ def train(params_path: str = 'params.yaml') -> None:
                 train_loss += loss
                 loss.backward()
                 optimizer.step()
-                _ = metric(outputs, labels)
+                _ = metric(outputs.topk(k=1, dim=1)[1], labels.topk(k=1, dim=1)[1])
             train_loss /= len(train_dataloader)
             train_acc = metric.compute()
             metric.reset()
@@ -76,7 +76,7 @@ def train(params_path: str = 'params.yaml') -> None:
                     outputs = net(inputs)
                     loss = criterion(outputs, labels)
                     valid_loss += loss
-                    _ = metric(outputs, labels)
+                    _ = metric(outputs.topk(k=1, dim=1)[1], labels.topk(k=1, dim=1)[1])
                 valid_loss /= len(valid_dataloader)
                 valid_acc = metric.compute()
                 metric.reset()
@@ -93,7 +93,6 @@ def train(params_path: str = 'params.yaml') -> None:
             live.next_step()
 
         torch.save(net, 'model.pt')
-        live.log_artifact('model.pt', type='model', name='resnet50')
 
 
 if __name__ == '__main__':