Updated doc.

docs/examples/DetailedMNIST.md

@@ -163,8 +163,8 @@ from leanai.training.losses import SparseCrossEntropyLossFromLogits, Loss
 
 
 class MyLoss(Loss):
-    def __init__(self, parent: Experiment):
-        super().__init__(parent)
+    def __init__(self):
+        super().__init__()
         self.loss = SparseCrossEntropyLossFromLogits()
 
     def forward(self, y_pred, y_true):
@@ -181,6 +181,7 @@ Example:
 ```python
 import torch
 from torch.optim import SGD
+from leanai.core.config import DictLike
 from leanai.core.experiment import Experiment, set_seeds
 
 
@@ -191,14 +192,14 @@ experiment = Experiment(
     example_input=torch.zeros((2, 28, 28, 1), dtype=torch.float32),
 )
 experiment.run_training(
-    load_dataset=dict(
+    load_dataset=DictLike(
         type=FashionMNISTDataset,
         data_path="outputs",
     ),
-    build_loss=dict(
+    build_loss=DictLike(
         type=MyLoss
     ),
-    build_optimizer=dict(
+    build_optimizer=DictLike(
         type=SGD,
         lr=1e-3,
     ),
@@ -227,8 +228,6 @@ LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0,1]
 
 ```
 
-> **Note**: Unfortunately the progress bar does not work properly in jupyter notebooks.
-
 ## Wrap-Up
 
 That is it for the tutorial. You might want to have a look at tensorboard though. Here you go.

docs/examples/SimpleMNIST.md

@@ -25,6 +25,7 @@ Example:
 import torch
 from torch.optim import SGD
 
+from leanai.core.config import DictLike
 from leanai.core.experiment import Experiment, set_seeds
 from leanai.data.datasets import FashionMNISTDataset
 from leanai.training.losses import SparseCrossEntropyLossFromLogits
@@ -76,14 +77,14 @@ Lastly, do not forget to specify your `batch_size` and `epochs`, which are techn
 Example:
 ```python
 experiment.run_training(
-    load_dataset=dict(
+    load_dataset=DictLike(
         type=FashionMNISTDataset,
         data_path="outputs",
     ),
-    build_loss=dict(
+    build_loss=DictLike(
         type=SparseCrossEntropyLossFromLogits,
     ),
-    build_optimizer=dict(
+    build_optimizer=DictLike(
         type=SGD,
         lr=1e-3,
     ),
@@ -112,8 +113,6 @@ LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0,1]
 
 ```
 
-> **Note**: Unfortunately the progress bar does not work properly in jupyter notebooks.
-
 ## Wrap-Up
 
 That is it for the tutorial. You might want to have a look at tensorboard though. Here you go.
@@ -123,10 +122,4 @@ Example:
 %load_ext tensorboard
 %tensorboard --logdir outputs
 ```
-Output:
-```
-The tensorboard extension is already loaded. To reload it, use:
-  %reload_ext tensorboard
-
-```
 

docs/leanai/core/experiment.md

@@ -70,11 +70,6 @@ This is required, if you execute the code from within the backup in the checkpoi
 (Default: True)
 
 
----
-### *def* **configure_gradient_clipping**(*self*, optimizer: Optimizer, optimizer_idx: int, gradient_clip_val: Optional[Union[int, float]] = None, gradient_clip_algorithm: Optional[str] = None)
-
-*(no documentation found)*
-
 ---
 ### *def* **run_training**
 

docs/leanai/data/dataset.md

@@ -170,6 +170,8 @@ parser or consider using and inheriting from the SimpleDataset.
 
 * **file_provider_iterable**: The iterable file provider providing samples to the parser.
 * **parser**: The parser converting samples into a usable format.
+:transformers: Transformers that are applied on the dataset to convert the format to what the model requires. (Default: [])
+:test_mode: A parameter that is passed to the constructor of the transformers (Default: False).
 
 
 ---
@@ -187,6 +189,8 @@ parser or consider using and inheriting from the SimpleDataset.
 
 * **file_provider_sequence**: The sequence file provider providing samples to the parser.
 * **parser**: The parser converting samples into a usable format.
+:transformers: Transformers that are applied on the dataset to convert the format to what the model requires. (Default: [])
+:test_mode: A parameter that is passed to the constructor of the transformers (Default: False).
 
 
 ---
@@ -217,6 +221,8 @@ If that is not the case, your code may break somewhere.
 * **InputType**: A definition of a named tuple that defines the input of the neural network.
 * **OutputType**: A definition of a named tuple that defines the output of the neural network.
 * **ignore_file_not_found**: If a file is missing return None instead of an exception.  (Default: False).
+:transformers: Transformers that are applied on the dataset to convert the format to what the model requires. (Default: [])
+:test_mode: A parameter that is passed to the constructor of the transformers (Default: False).
 
 
 ---

docs/leanai/training/losses/loss.md

@@ -11,14 +11,6 @@
 ---
 ## *class* **Loss**(Module)
 
-Create a loss.
-
-* **parent**: The parent is required for logging. Parent must be of type Loss or Experiment.
-
-
----
-### *def* **set_parent**(*self*, parent)
-
 *(no documentation found)*
 
 ---

docs/leanai/training/losses/multiloss.md

@@ -29,7 +29,7 @@ Compute the multiloss using the provided losses.
 
 
 ---
-### *def* **MultiLossV2**(parent, **losses) -> SumLoss
+### *def* **MultiLossV2**(**losses) -> SumLoss
 
 Normalizes the losses by variance estimation and then sums them.