Apply suggestions from code review

Co-authored-by: Dalton Bohning <[email protected]>
Signed-off-by: 0xE0F <[email protected]>

src/client/pydaos/torch/Readme.md

@@ -12,17 +12,17 @@ To implement map style dataset only two methods are required: `__len__()` and `_
 
 During dataset creation the connection to container will be established and its namespace will be scanned to build
 a list of files in container with their size. The number of items in that list will be used to implement `__len__()` method.
-`__getitem__()` implementation consist of looking up the object by its absolute path and reading its content into the buffer
-create on the python side.
+`__getitem__()` implementation consists of looking up the object by its absolute path and reading its content into the buffer
+created on the python side.
 
-The `__getitems__()` method allow to request multiple samples at once making this a good case to use DAOS event queue to send and wait on batch items.
+The `__getitems__()` method allows requesting multiple samples at once, making this a good case to use DAOS event queues to send and wait on batch items.
 
-By default Dataset is single threaded (more like single process in python), `__getitem__()` and `__getitems__()` are regular blocking calls.
-If multiprocessing is enabled, Dataset provides the `worker_init` method, which worker processes are calling upon their startup,
-during this setup the global connection should be reused and the new event queue should be created for calling worker process.
+By default `Dataset` is single threaded (more like single process in python), `__getitem__()` and `__getitems__()` are regular blocking calls.
+If multiprocessing is enabled, `Dataset` provides the `worker_init` method, which worker processes are calling upon their startup.
+During this setup the global connection should be reused and the new event queue should be created for calling worker processes.
 
 There's no internal multithreading inside the shim module - it's driven on outside by `torch.utils.DataLoader`.
-If DataLoader is configured to have 8 readers then 8 event queues are going to be created per each worker process so the performance of individual worker should not be affected by others.
+If `DataLoader` is configured to have 8 readers then 8 event queues are going to be created per each worker process so the performance of individual worker should not be affected by others.
 
 
 Implementation of `torch.utils.data.IterableDataset` requires to implement `__iter__()` protocol, which can be fully implemented on python side,
@@ -35,7 +35,7 @@ Configured and running DAOS agent on the node(s) and correctly set ACLs - the us
 
 
 
-### Example of usage Map style Dataset
+### Example usage of Map style Dataset
 
 ```python
 import numpy as np

src/client/pydaos/torch/__init__.py

@@ -1,4 +1,4 @@
-# (C) Copyright 2019-2024 Intel Corporation.
+# (C) Copyright 2024 Intel Corporation.
 # (C) Copyright 2024 Google LLC
 # (C) Copyright 2024 Enakta Labs Ltd
 #

src/client/pydaos/torch/torch_api.py

@@ -316,7 +316,7 @@ def parallel_list(self, path=None,
                       readdir_batch_size=READDIR_BATCH_SIZE,
                       workers=PARALLEL_SCAN_WORKERS):
         """
-        Parallel list tires to leverage DAOS ability to read dir in parallel
+        Parallel list tries to leverage DAOS ability to read dir in parallel
         by splitting across multiple engines.
 
         To fully use this feature the container should be configured with directory object classes