Fixed a couple of typos.

This fix fixes a couple of typos:
```
s/unneccessary/unnecessary/
s/accidently/accidentally/
```

Author: yongtang

tensorflow/tools/graph_transforms/README.md

@@ -169,7 +169,7 @@ then you'll need to make local modifications to the build files to include the
 right .cc file that defines it. In a lot of cases the op is just a vestigial
 remnant from the training process though, and if that's true then you can run
 the [strip_unused_nodes](#strip_unused_nodes), specifying the inputs and outputs
-of your inference usage, to remove those unneccessary nodes:
+of your inference usage, to remove those unnecessary nodes:
 
 ```bash
 bazel build tensorflow/tools/graph_transforms:transform_graph

tensorflow/tools/tfprof/README.md

@@ -439,7 +439,7 @@ with gfile.Open(os.path.join(output_dir, "run_meta"), "w") as f:
 <b>--op_log_path:</b>
 tensorflow::tfprof::OpLog. A proto used to provide extra op information
 for ops. By giving a group of ops a type name, users can easily aggregate the
-statistics for those ops without accidently missing or including extra ops.
+statistics for those ops without accidentally missing or including extra ops.
 tfprof exposes the following Python API to add op information and logging.
 
 ```python