release v0.3.6

README.md

@@ -42,28 +42,40 @@ To use lightning do 2 things:
 ```python
 import pytorch_lightning as ptl
 import torch
+from torch.nn import functional as F
+from torch.utils.data import DataLoader
+from torchvision.datasets import MNIST
 
 class CoolModel(ptl.LightningModule):
 
     def __init(self):
-        self.l1 = torch.nn.Linear(28*28, 10)
-        
+        super(CoolModel, self).__init__()
+        # not the best model...
+        self.l1 = torch.nn.Linear(28 * 28, 10)
+
     def forward(self, x):
-        return self.l1(x)
-    
+        return torch.relu(self.l1(x))
+
+    def my_loss(self, y_hat, y):
+        return F.cross_entropy(y_hat, y)
+
     def training_step(self, batch, batch_nb):
         x, y = batch
         y_hat = self.forward(x)
-        return {'tng_loss': some_loss(y_hat, y)}
-    
+        return {'tng_loss': self.my_loss(y_hat, y)}
+
     def validation_step(self, batch, batch_nb):
         x, y = batch
         y_hat = self.forward(x)
-        return {'val_loss': some_loss(y_hat, y)}
-        
+        return {'val_loss': self.my_loss(y_hat, y)}
+
+    def validation_end(self, outputs):
+        avg_loss = torch.stack([x for x in outputs['val_loss']]).mean()
+        return avg_loss
+
     def configure_optimizers(self):
-        return [optim.Adam(self.parameters(), lr=0.02)]
-    
+        return [torch.optim.Adam(self.parameters(), lr=0.02)]
+
     @ptl.data_loader
     def tng_dataloader(self):
         return DataLoader(MNIST('path/to/save', train=True), batch_size=32)
@@ -71,11 +83,10 @@ class CoolModel(ptl.LightningModule):
     @ptl.data_loader
     def val_dataloader(self):
         return DataLoader(MNIST('path/to/save', train=False), batch_size=32)
-        
+
     @ptl.data_loader
     def test_dataloader(self):
-        return DataLoader(MNIST('path/to/save', train=False), batch_size=32)   
-        
+        return DataLoader(MNIST('path/to/save', train=False), batch_size=32)
 ```
 
 2. Fit with a [trainer](https://williamfalcon.github.io/pytorch-lightning/Trainer/)    

docs/LightningModule/RequiredTrainerInterface.md

@@ -26,6 +26,58 @@ Otherwise, to Define a Lightning Module, implement the following methods:
 - [update_tng_log_metrics](RequiredTrainerInterface.md#update_tng_log_metrics)
 - [add_model_specific_args](RequiredTrainerInterface.md#add_model_specific_args)
 
+---
+**Minimal example**
+```python
+import pytorch_lightning as ptl
+import torch
+from torch.nn import functional as F
+from torch.utils.data import DataLoader
+from torchvision.datasets import MNIST
+
+class CoolModel(ptl.LightningModule):
+
+    def __init(self):
+        super(CoolModel, self).__init__()
+        # not the best model...
+        self.l1 = torch.nn.Linear(28 * 28, 10)
+
+    def forward(self, x):
+        return torch.relu(self.l1(x))
+
+    def my_loss(self, y_hat, y):
+        return F.cross_entropy(y_hat, y)
+
+    def training_step(self, batch, batch_nb):
+        x, y = batch
+        y_hat = self.forward(x)
+        return {'tng_loss': self.my_loss(y_hat, y)}
+
+    def validation_step(self, batch, batch_nb):
+        x, y = batch
+        y_hat = self.forward(x)
+        return {'val_loss': self.my_loss(y_hat, y)}
+
+    def validation_end(self, outputs):
+        avg_loss = torch.stack([x for x in outputs['val_loss']]).mean()
+        return avg_loss
+
+    def configure_optimizers(self):
+        return [torch.optim.Adam(self.parameters(), lr=0.02)]
+
+    @ptl.data_loader
+    def tng_dataloader(self):
+        return DataLoader(MNIST('path/to/save', train=True), batch_size=32)
+
+    @ptl.data_loader
+    def val_dataloader(self):
+        return DataLoader(MNIST('path/to/save', train=False), batch_size=32)
+
+    @ptl.data_loader
+    def test_dataloader(self):
+        return DataLoader(MNIST('path/to/save', train=False), batch_size=32)
+```
+
 ---
 
 ### training_step

docs/index.md

@@ -1,10 +1,10 @@
 ###### New project Quick Start    
 To start a new project define these two files.   
 
-1. [Define a LightningModule](/LightningModule/RequiredTrainerInterface/#template-model-definition)  
+1. [Define a LightningModule](/LightningModule/RequiredTrainerInterface/) 
 2. Pick a trainer      
-    - [Basic CPU Trainer](https://github.com/williamFalcon/pytorch-lightning/blob/master/examples/new_project_templates/trainer_cpu_template.py) 
-    - [GPU cluster Trainer](https://github.com/williamFalcon/pytorch-lightning/blob/master/examples/new_project_templates/trainer_gpu_cluster_template.py)
+    - [Basic CPU Trainer](https://github.com/williamFalcon/pytorch-lightning/blob/master/pytorch_lightning/examples/new_project_templates/single_cpu_template.py) 
+    - [GPU cluster Trainer](https://github.com/williamFalcon/pytorch-lightning/blob/master/pytorch_lightning/examples/new_project_templates/multi_node_cluster_template.py)
 
 ###### Docs shortcuts
 - [LightningModule](LightningModule/RequiredTrainerInterface/)  

pytorch_lightning/examples/new_project_templates/lightning_module_template.py

@@ -25,7 +25,8 @@ class LightningTemplateModel(LightningModule):
         :param hparams:
         """
         # init superclass
-        super(LightningTemplateModel, self).__init__(hparams)
+        super(LightningTemplateModel, self).__init__()
+        self.hparams = hparams
 
         self.batch_size = hparams.batch_size
 

pytorch_lightning/root_module/root_module.py

@@ -8,9 +8,8 @@ from pytorch_lightning.root_module.decorators import data_loader
 
 class LightningModule(GradInformation, ModelIO, ModelHooks):
 
-    def __init__(self, hparams):
-        super(LightningModule, self).__init__()
-        self.hparams = hparams
+    def __init__(self, *args, **kwargs):
+        super(LightningModule, self).__init__(*args, **kwargs)
 
         self.dtype = torch.FloatTensor
         self.exp_save_path = None
@@ -64,26 +63,6 @@ class LightningModule(GradInformation, ModelIO, ModelHooks):
         """
         raise NotImplementedError
 
-    def loss(self, *args, **kwargs):
-        """
-        Expand model_out into your components
-        :param model_out:
-        :return:
-        """
-        raise NotImplementedError
-
-    def summarize(self):
-        model_summary = ModelSummary(self)
-        print(model_summary)
-
-    def freeze(self):
-        for param in self.parameters():
-            param.requires_grad = False
-
-    def unfreeze(self):
-        for param in self.parameters():
-            param.requires_grad = True
-
     @data_loader
     def tng_dataloader(self):
         """
@@ -136,5 +115,17 @@ class LightningModule(GradInformation, ModelIO, ModelHooks):
         model.load_state_dict(checkpoint['state_dict'], strict=False)
         return model
 
+    def summarize(self):
+        model_summary = ModelSummary(self)
+        print(model_summary)
+
+    def freeze(self):
+        for param in self.parameters():
+            param.requires_grad = False
+
+    def unfreeze(self):
+        for param in self.parameters():
+            param.requires_grad = True
+
 
 

pytorch_lightning/testing_models/lm_test_module.py

@@ -25,7 +25,8 @@ class LightningTestModel(LightningModule):
         :param hparams:
         """
         # init superclass
-        super(LightningTestModel, self).__init__(hparams)
+        super(LightningTestModel, self).__init__()
+        self.hparams = hparams
 
         self.batch_size = hparams.batch_size
 

setup.py

@@ -7,7 +7,7 @@ from setuptools import setup, find_packages
 # http://blog.ionelmc.ro/2014/05/25/python-packaging/
 setup(
     name="pytorch-lightning",
-    version='0.3.51',
+    version='0.3.6',
     description="The Keras for ML researchers using PyTorch",
     author="William Falcon",
     author_email="waf2107@columbia.edu",

tests/debug.py

@@ -11,6 +11,55 @@ import os
 import shutil
 import pdb
 
+import pytorch_lightning as ptl
+import torch
+from torch.nn import functional as F
+from torch.utils.data import DataLoader
+from torchvision.datasets import MNIST
+
+
+class CoolModel(ptl.LightningModule):
+
+    def __init(self):
+        super(CoolModel, self).__init__()
+        # not the best model...
+        self.l1 = torch.nn.Linear(28 * 28, 10)
+
+    def forward(self, x):
+        return torch.relu(self.l1(x))
+
+    def my_loss(self, y_hat, y):
+        return F.cross_entropy(y_hat, y)
+
+    def training_step(self, batch, batch_nb):
+        x, y = batch
+        y_hat = self.forward(x)
+        return {'tng_loss': self.my_loss(y_hat, y)}
+
+    def validation_step(self, batch, batch_nb):
+        x, y = batch
+        y_hat = self.forward(x)
+        return {'val_loss': self.my_loss(y_hat, y)}
+
+    def validation_end(self, outputs):
+        avg_loss = torch.stack([x for x in outputs['val_loss']]).mean()
+        return avg_loss
+
+    def configure_optimizers(self):
+        return [torch.optim.Adam(self.parameters(), lr=0.02)]
+
+    @ptl.data_loader
+    def tng_dataloader(self):
+        return DataLoader(MNIST('path/to/save', train=True), batch_size=32)
+
+    @ptl.data_loader
+    def val_dataloader(self):
+        return DataLoader(MNIST('path/to/save', train=False), batch_size=32)
+
+    @ptl.data_loader
+    def test_dataloader(self):
+        return DataLoader(MNIST('path/to/save', train=False), batch_size=32)
+
 
 def get_model():
     # set up model with these hyperparams
@@ -94,11 +143,9 @@ def run_prediction(dataloader, trained_model):
 def main():
 
     save_dir = init_save_dir()
-    model, hparams = get_model()
 
     # exp file to get meta
     exp = get_exp(False)
-    exp.argparse(hparams)
     exp.save()
 
     # exp file to get weights
@@ -113,6 +160,8 @@ def main():
         distributed_backend='dp',
     )
 
+    model = CoolModel()
+
     result = trainer.fit(model)
 
     # correct result and ok accuracy