Deployed 2425ef0 with MkDocs version: 1.0.4

Trainer/Checkpointing/index.html

@@ -572,10 +572,39 @@ However, the dataloaders will start from the first batch again (if you shuffled
 <pre><code class="python">from test_tube import Experiment
 
 exp = Experiment(version=a_previous_version_with_a_saved_checkpoint)
-Trainer(experiment=exp)
+trainer = Trainer(experiment=exp)
 
-trainer = Trainer(checkpoint_callback=checkpoint_callback)
-# the trainer is now restored
+# this fit call loads model weights and trainer state
+# the trainer continues seamlessly from where you left off
+# without having to do anything else.
+trainer.fit(model)
+</code></pre>
+
+<p>The trainer restores:<br />
+- global_step  <br />
+- current_epoch  <br />
+- All optimizers  <br />
+- All lr_schedulers  <br />
+- Model weights</p>
+<p>You can even change the logic of your model as long as the weights and "architecture" of 
+the system isn't different. If you add a layer, for instance, it might not work.   </p>
+<p>At a rough level, here's <a href="https://github.com/williamFalcon/pytorch-lightning/blob/master/pytorch_lightning/root_module/model_saving.py#L63">what happens inside Trainer</a>:   </p>
+<pre><code class="python">
+self.global_step = checkpoint['global_step']
+self.current_epoch = checkpoint['epoch']
+
+# restore the optimizers
+optimizer_states = checkpoint['optimizer_states']
+for optimizer, opt_state in zip(self.optimizers, optimizer_states):
+    optimizer.load_state_dict(opt_state)
+
+# restore the lr schedulers
+lr_schedulers = checkpoint['lr_schedulers']
+for scheduler, lrs_state in zip(self.lr_schedulers, lr_schedulers):
+    scheduler.load_state_dict(lrs_state)
+
+# uses the model you passed into trainer        
+model.load_state_dict(checkpoint['state_dict'])
 </code></pre>