Fix docs formatting (#188)

Author: JiahaoYao

README.md

@@ -36,11 +36,11 @@ Or to install master:
 Here are the supported PyTorch Lightning versions:
 
 | Ray Lightning | PyTorch Lightning |
-|---|---|
-| 0.1 | 1.4 |
-| 0.2 | 1.5 |
-| 0.3 | 1.6 |
-| master | 1.6 |
+|---------------|-------------------|
+| 0.1           | 1.4               |
+| 0.2           | 1.5               |
+| 0.3           | 1.6               |
+| master        | 1.6               |
 
 
 ## PyTorch Distributed Data Parallel Strategy on Ray
@@ -194,7 +194,7 @@ As discussed [here](https://github.com/pytorch/pytorch/issues/51688#issuecomment
 Neither of these should be an issue with the `RayStrategy` due to Ray's serialization mechanisms. The only thing to keep in mind is that when using this strategy, your model does have to be serializable/pickleable.
 
 > Horovod installation issue
-please see [details](./docs/horovod_faq.md)
+please see [details](https://github.com/ray-project/ray_lightning/blob/main/docs/horovod_faq.md)
 
 <!--$UNCOMMENT## API Reference
 

ray_lightning/ray_ddp.py

@@ -23,6 +23,7 @@
 @PublicAPI(stability="beta")
 class RayStrategy(DDPSpawnStrategy):
     """Pytorch Lightning strategy for DDP training on a Ray cluster.
+
     This strategy is used to manage distributed training using DDP and
     Ray for process launching. Internally, the specified number of
     Ray actors are launched in the cluster and are registered as part of a
@@ -35,6 +36,7 @@ class RayStrategy(DDPSpawnStrategy):
     script: ``python train.py``, and only on the head node if running in a
     distributed Ray cluster. There is no need to run this script on every
     single node.
+
     Args:
         num_workers (int): Number of training workers to use.
         num_cpus_per_worker (int): Number of CPUs per worker.
@@ -51,6 +53,7 @@ class RayStrategy(DDPSpawnStrategy):
             ``DistributedDataParallel`` initialization
     Example:
         .. code-block:: python
+
             import pytorch_lightning as ptl
             from ray_lightning import RayAccelerator
             ptl_model = MNISTClassifier(...)