Update reward_main.py (PaddlePaddle#7880)

Author: wawltor

examples/RLHF/reward_main.py

@@ -32,14 +32,6 @@
 from paddlenlp.transformers import AutoConfig, AutoTokenizer, LlamaTokenizer
 from paddlenlp.utils.log import logger
 
-# launch would unset http_proxy
-# export https_proxy=http://172.19.57.45:3128
-# os.environ["http_proxy"] = "http://172.19.57.45:3128"
-# os.environ["https_proxy"] = "http://172.19.57.45:3128"
-os.environ["http_proxy"] = "http://10.162.37.16:8128"
-os.environ["https_proxy"] = "http://10.162.37.16:8128"
-
-
 @dataclass
 class TrainingArguments(TrainingArguments):
     loss_type: Literal["token-wise", "sequence-wise"] = field(