release v0.4.3

* Update trainer.py

* Update trainer.py

* Update trainer.py

* Update trainer.py

* Update trainer.py

* Update trainer.py

* Update trainer.py

.github/ISSUE_TEMPLATE/bug_report.md

@@ -7,6 +7,10 @@ assignees: ''
 
 ---
 
+### Common bugs:
+1. Tensorboard not showing in Jupyter-notebook see [issue 79](https://github.com/williamFalcon/pytorch-lightning/issues/79).    
+2. PyTorch 1.1.0 vs 1.2.0 support [see FAQ](https://github.com/williamFalcon/pytorch-lightning#faq)    
+
 **Describe the bug**
 A clear and concise description of what the bug is.
 
@@ -28,11 +32,5 @@ If applicable, add screenshots to help explain your problem.
  - Browser [e.g. chrome, safari]
  - Version [e.g. 22]
 
-**Smartphone (please complete the following information):**
- - Device: [e.g. iPhone6]
- - OS: [e.g. iOS8.1]
- - Browser [e.g. stock browser, safari]
- - Version [e.g. 22]
-
 **Additional context**
 Add any other context about the problem here.

README.md

@@ -370,6 +370,10 @@ Nope.
 **Are there plans to support virtualenv?**    
 Nope. Please use anaconda or miniconda.    
 
+**Which PyTorch versions do you support?**    
+Lightning 0.4.2+ supports PyTorch 1.2.0.    
+For PyTorch 1.1.0 install Lightning 0.4.0 with test-tube=0.6.7.6.    
+
 ## Bleeding edge
 If you can't wait for the next release, install the most up to date code with:  
 ```bash

docs/Trainer/Checkpointing.md

@@ -1,11 +1,11 @@
-Lightning can automate saving and loading checkpoints.
+i Lightning can automate saving and loading checkpoints.
 
 ---
 ### Model saving
 To enable checkpointing, define the checkpoint callback and give it to the trainer.
 
 ``` {.python}
-from pytorch_lightning.utils.pt_callbacks import ModelCheckpoint
+from pytorch_lightning.callbacks import ModelCheckpoint
 
 checkpoint_callback = ModelCheckpoint(
     filepath='/path/to/store/weights.ckpt',
@@ -65,4 +65,4 @@ for scheduler, lrs_state in zip(self.lr_schedulers, lr_schedulers):
 
 # uses the model you passed into trainer        
 model.load_state_dict(checkpoint['state_dict'])
-```    
+```    

docs/Trainer/Training Loop.md

@@ -28,15 +28,18 @@ trainer = Trainer(enable_early_stop=True)
 ```
 
 ---
-#### Gradient Clipping 
-Use this to turn off early stopping and run training to the [max_epoch](#force-training-for-min-or-max-epochs)
+#### Gradient Clipping
+Gradient clipping may be enabled to avoid exploding gradients.
+Specifically, this will [clip the gradient norm computed over all model parameters *together*](https://pytorch.org/docs/stable/nn.html#torch.nn.utils.clip_grad_norm_).
+
 ``` {.python}
 # DEFAULT (ie: don't clip)
 trainer = Trainer(gradient_clip=0)
+
+# clip gradients with norm above 0.5
+trainer = Trainer(gradient_clip=0.5)
 ```
 
-
-
 ---
 #### Inspect gradient norms
 Looking at grad norms can help you figure out where training might be going wrong.

pytorch_lightning/models/trainer.py

@@ -899,6 +899,9 @@ We recommend you switch to ddp if you want to use amp
 
         self.__add_tqdm_metrics(model_specific_tqdm_metrics_dic)
 
+        # accumulate loss (if accumulate_grad_batches = 1 no effect)
+        loss = loss / self.accumulate_grad_batches
+
         # backward pass
         if self.use_amp:
             # scale loss when using amp
@@ -918,12 +921,11 @@ We recommend you switch to ddp if you want to use amp
             for param in model.parameters():
                 print(param.grad.float().sum())
 
-        # avoid memory leaks
+        # track total loss for logging (avoid mem leaks)
         self.batch_loss_value += loss.item()
 
         # gradient update with accumulated gradients
         if (self.batch_nb + 1) % self.accumulate_grad_batches == 0:
-
             # clip gradients
             if self.gradient_clip > 0:
                 model = self.__get_model()
@@ -941,11 +943,7 @@ We recommend you switch to ddp if you want to use amp
                 # clear gradients
                 optimizer.zero_grad()
 
-            # queuing loss across batches blows it up proportionally...
-            #  divide out the number accumulated
-            self.batch_loss_value = self.batch_loss_value / self.accumulate_grad_batches
-
-            # track loss
+            # calculate running loss for display
             self.running_loss.append(self.batch_loss_value)
             self.batch_loss_value = 0
             self.avg_loss = np.mean(self.running_loss[-100:])

setup.py

@@ -9,12 +9,12 @@ from setuptools import setup, find_packages
 
 # https://packaging.python.org/discussions/install-requires-vs-requirements /
 # keep the meta-data here for simplicity in reading this file... it's not obvious
-# what happens and to non-engineers they won't know to look in init...
+# what happens and to non-engineers they won't know to look in init ...
 # the goal of the project is simplicity for researchers, don't want to add too much
 # engineer specific practices
 setup(
     name='pytorch-lightning',
-    version='0.4.0',
+    version='0.4.3',
     description='The Keras for ML researchers using PyTorch',
     author='William Falcon',
     author_email='waf2107@columbia.edu',
@@ -29,9 +29,9 @@ setup(
     keywords=['deep learning', 'pytorch', 'AI'],
     python_requires='>=3.6',
     install_requires=[
-        'torch==1.1.0',
+        'torch==1.2.0',
         'tqdm',
-        'test-tube>=0.6.7.6',
+        'test-tube==0.6.8',
         'pandas>=0.20.3',
     ],
     classifiers=[