fix: loader, etc, etc

.github/workflows/dl_cpu.yml

@@ -109,9 +109,20 @@ jobs:
       - name: check examples
         env:
           REQUIREMENTS: ${{ matrix.requirements }}
-        run: |
-          pytest .
-          pip install -e .
-          cd examples && bash mnist_stages/run_config.sh && cd ..
-          cd examples && bash mnist_stages/run_hydra.sh && cd ..
-          cd examples && bash mnist_stages/run_tune.sh && cd ..
+        uses: nick-invision/retry@v2
+        with:
+          timeout_minutes: 15
+          max_attempts: 4
+          retry_on: timeout
+          command: |
+            pytest .
+            pip install -e .
+            cd examples && bash mnist_stages/run_config.sh && cd ..
+            cd examples && bash mnist_stages/run_hydra.sh && cd ..
+            cd examples && bash mnist_stages/run_tune.sh && cd ..
+#        run: |
+#          pytest .
+#          pip install -e .
+#          cd examples && bash mnist_stages/run_config.sh && cd ..
+#          cd examples && bash mnist_stages/run_hydra.sh && cd ..
+#          cd examples && bash mnist_stages/run_tune.sh && cd ..

CHANGELOG.md

@@ -12,26 +12,32 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Nifti Reader (NiftiReader) ([#1151](https://github.com/catalyst-team/catalyst/pull/1151))
 - CMC score and callback for ReID task (ReidCMCMetric and ReidCMCScoreCallback) ([#1170](https://github.com/catalyst-team/catalyst/pull/1170))
 - Market1501 metric learning datasets (Market1501MLDataset and Market1501QGDataset) ([#1170](https://github.com/catalyst-team/catalyst/pull/1170))
+- extra kwargs support for Engines ([#1156](https://github.com/catalyst-team/catalyst/pull/1156))
+- engines exception for unknown model type ([#1174](https://github.com/catalyst-team/catalyst/issues/1174))
+- a few docs to the supported loggers ([#1174](https://github.com/catalyst-team/catalyst/issues/1174))
 
 ### Changed
 
--
+- ``TensorboardLogger`` switched from ``global_batch_step`` counter to ``global_sample_step`` one ([#1174](https://github.com/catalyst-team/catalyst/issues/1174))
+- ``TensorboardLogger`` logs loader metric ``on_loader_end`` rather than ``on_epoch_end`` ([#1174](https://github.com/catalyst-team/catalyst/issues/1174))
+- ``prefix`` renamed to ``metric_key`` for ``MetricAggregationCallback`` ([#1174](https://github.com/catalyst-team/catalyst/issues/1174))
+- ``micro``, ``macro`` and ``weighted`` aggregations renamed to ``_micro``, ``_macro`` and ``_weighted`` ([#1174](https://github.com/catalyst-team/catalyst/issues/1174))
 
 ### Removed
 
--
+- auto ``torch.sigmoid`` usage for ``metrics.AUCMetric`` and ``metrics.auc`` ([#1174](https://github.com/catalyst-team/catalyst/issues/1174))
 
 ### Fixed
 
--
+- hitrate calculation issue ([#1155](https://github.com/catalyst-team/catalyst/issues/1155))
+- ILoader wrapper usage issue with Runner ([#1174](https://github.com/catalyst-team/catalyst/issues/1174))
 
 ## [21.03.2] - 2021-03-29
 
 ### Fixed
 
 - minimal requirements issue ([#1147](https://github.com/catalyst-team/catalyst/issues/1147))
-- nested dicts in `loaders_params`/`samplers_params` overriding fixed ([#1150](https://github.com/catalyst-team/catalyst/pull/1150))
-- fixed hitrate calculation issue ([#1155]) (https://github.com/catalyst-team/catalyst/issues/1155)
+- nested dicts in `loaders_params`/`samplers_params` overriding ([#1150](https://github.com/catalyst-team/catalyst/pull/1150))
 
 ## [21.03.1] - 2021-03-28
 
@@ -46,7 +52,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - FactorizedLinear to contrib ([1142](https://github.com/catalyst-team/catalyst/pull/1142))
 - Extra init params for ``ConsoleLogger`` ([1142](https://github.com/catalyst-team/catalyst/pull/1142))
 - Tracing, Quantization, Onnx, Pruninng Callbacks ([1127](https://github.com/catalyst-team/catalyst/pull/1127))
-- `_key_value` for schedulers in case of multiple optimizers fixed ([#1146](https://github.com/catalyst-team/catalyst/pull/1146))
+
 
 ### Changed
 
@@ -63,7 +69,8 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - BatchLimitLoaderWrapper logic for loaders with shuffle flag ([#1136](https://github.com/catalyst-team/catalyst/issues/1136))
 - config description in the examples ([1142](https://github.com/catalyst-team/catalyst/pull/1142))
 - Config API deprecated parsings logic ([1142](https://github.com/catalyst-team/catalyst/pull/1142)) ([1138](https://github.com/catalyst-team/catalyst/pull/1138))
-- RecSys metrics Top_k calculations ([#1140] (https://github.com/catalyst-team/catalyst/pull/1140))
+- RecSys metrics Top_k calculations ([#1140](https://github.com/catalyst-team/catalyst/pull/1140))
+- `_key_value` for schedulers in case of multiple optimizers ([#1146](https://github.com/catalyst-team/catalyst/pull/1146))
 
 ## [21.03] - 2021-03-13 ([#1095](https://github.com/catalyst-team/catalyst/issues/1095))
 

README.md

@@ -181,6 +181,90 @@ Tested on Ubuntu 16.04/18.04/20.04, macOS 10.15, Windows 10, and Windows Subsyst
 
 ### Minimal Examples
 
+<details>
+<summary>CustomRunner – PyTorch for-loop decomposition</summary>
+<p>
+
+```python
+import os
+from torch import nn, optim
+from torch.nn import functional as F
+from torch.utils.data import DataLoader
+from catalyst import dl, metrics
+from catalyst.data.transforms import ToTensor
+from catalyst.contrib.datasets import MNIST
+
+model = nn.Sequential(nn.Flatten(), nn.Linear(28 * 28, 10))
+optimizer = optim.Adam(model.parameters(), lr=0.02)
+
+loaders = {
+    "train": DataLoader(
+        MNIST(os.getcwd(), train=True, download=True, transform=ToTensor()), batch_size=32
+    ),
+    "valid": DataLoader(
+        MNIST(os.getcwd(), train=False, download=True, transform=ToTensor()), batch_size=32
+    ),
+}
+
+class CustomRunner(dl.Runner):
+    def predict_batch(self, batch):
+        # model inference step
+        return self.model(batch[0].to(self.device))
+
+    def on_loader_start(self, runner):
+        super().on_loader_start(runner)
+        self.meters = {
+            key: metrics.AdditiveValueMetric(compute_on_call=False)
+            for key in ["loss", "accuracy01", "accuracy03"]
+        }
+
+    def handle_batch(self, batch):
+        # model train/valid step
+        # unpack the batch
+        x, y = batch
+        # run model forward pass
+        logits = self.model(x)
+        # compute the loss
+        loss = F.cross_entropy(logits, y)
+        # compute other metrics of interest
+        accuracy01, accuracy03 = metrics.accuracy(logits, y, topk=(1, 3))
+        # log metrics
+        self.batch_metrics.update(
+            {"loss": loss, "accuracy01": accuracy01, "accuracy03": accuracy03}
+        )
+        for key in ["loss", "accuracy01", "accuracy03"]:
+            self.meters[key].update(self.batch_metrics[key].item(), self.batch_size)
+        # run model backward pass
+        if self.is_train_loader:
+            loss.backward()
+            self.optimizer.step()
+            self.optimizer.zero_grad()
+
+    def on_loader_end(self, runner):
+        for key in ["loss", "accuracy01", "accuracy03"]:
+            self.loader_metrics[key] = self.meters[key].compute()[0]
+        super().on_loader_end(runner)
+
+runner = CustomRunner()
+# model training
+runner.train(
+    model=model,
+    optimizer=optimizer,
+    loaders=loaders,
+    logdir="./logs",
+    num_epochs=5,
+    verbose=True,
+    valid_loader="valid",
+    valid_metric="loss",
+    minimize_valid_metric=True,
+)
+# model inference
+for logits in runner.predict_loader(loader=loaders["valid"]):
+    assert logits.detach().cpu().numpy().shape[-1] == 10
+```
+</p>
+</details>
+
 <details>
 <summary>ML - linear regression</summary>
 <p>
@@ -493,88 +577,6 @@ runner.train(
 </details>
 
 
-<details>
-<summary>Custom batch-metrics logging</summary>
-<p>
-
-```python
-import os
-from torch import nn, optim
-from torch.nn import functional as F
-from torch.utils.data import DataLoader
-from catalyst import dl, metrics
-from catalyst.data.transforms import ToTensor
-from catalyst.contrib.datasets import MNIST
-
-model = nn.Sequential(nn.Flatten(), nn.Linear(28 * 28, 10))
-optimizer = optim.Adam(model.parameters(), lr=0.02)
-
-loaders = {
-    "train": DataLoader(
-        MNIST(os.getcwd(), train=True, download=True, transform=ToTensor()), batch_size=32
-    ),
-    "valid": DataLoader(
-        MNIST(os.getcwd(), train=False, download=True, transform=ToTensor()), batch_size=32
-    ),
-}
-
-class CustomRunner(dl.Runner):
-    def predict_batch(self, batch):
-        # model inference step
-        return self.model(batch[0].to(self.device))
-
-    def on_loader_start(self, runner):
-        super().on_loader_start(runner)
-        self.meters = {
-            key: metrics.AdditiveValueMetric(compute_on_call=False)
-            for key in ["loss", "accuracy01", "accuracy03"]
-        }
-
-    def handle_batch(self, batch):
-        # model train/valid step
-        # unpack the batch
-        x, y = batch
-        # run model forward pass
-        logits = self.model(x)
-        # compute the loss
-        loss = F.cross_entropy(logits, y)
-        # compute other metrics of interest
-        accuracy01, accuracy03 = metrics.accuracy(logits, y, topk=(1, 3))
-        # log metrics
-        self.batch_metrics.update(
-            {"loss": loss, "accuracy01": accuracy01, "accuracy03": accuracy03}
-        )
-        for key in ["loss", "accuracy01", "accuracy03"]:
-            self.meters[key].update(self.batch_metrics[key].item(), self.batch_size)
-        # run model backward pass
-        if self.is_train_loader:
-            loss.backward()
-            self.optimizer.step()
-            self.optimizer.zero_grad()
-
-    def on_loader_end(self, runner):
-        for key in ["loss", "accuracy01", "accuracy03"]:
-            self.loader_metrics[key] = self.meters[key].compute()[0]
-        super().on_loader_end(runner)
-
-runner = CustomRunner()
-# model training
-runner.train(
-    model=model,
-    optimizer=optimizer,
-    loaders=loaders,
-    logdir="./logs",
-    num_epochs=5,
-    verbose=True,
-    valid_loader="valid",
-    valid_metric="loss",
-    minimize_valid_metric=True,
-)
-```
-</p>
-</details>
-
-
 <details>
 <summary>CV - MNIST classification</summary>
 <p>

catalyst/__version__.py

@@ -1 +1 @@
-__version__ = "21.03.2"
+__version__ = "21.04"