Update GenerationDefaults and comments

src/deepsparse/transformers/pipelines/text_generation.py

@@ -62,16 +62,21 @@
 __all__ = ["TextGenerationPipeline"]
 
 
+# Based off of https://huggingface.co/docs/transformers/main_classes/text_generation#transformers.GenerationConfig # noqa E501
 class GenerationDefaults:
-    num_return_sequences = 1
+    # Parameters that control the length of the output
     max_length = None
     max_new_tokens = 100
-    output_scores = False
-    top_k = 0
-    top_p = 0.0
-    repetition_penalty = 0.0
+    # Parameters that control the generation strategy used
     do_sample = False
+    # Parameters for manipulation of the model output logits
     temperature = 1.0
+    top_k = 50
+    top_p = 1.0
+    repetition_penalty = 1.0
+    # Parameters that define the outputs
+    num_return_sequences = 1
+    output_scores = False
 
 
 class FinishReason(Enum):