[BUG]: Chat: about load_state_dict of critic and reward_model

[HaomiaoPan]

🐛 Describe the bug

[train_prompts.py](https://github.com/hpcaitech/ColossalAI/blob/main/applications/Chat/examples/train_prompts.py

        if rm_model_name == 'gpt2':
            reward_model = GPTRM(pretrained=args.rm_pretrain)
        elif rm_model_name == 'bloom':
            reward_model = BLOOMRM(pretrained=args.rm_pretrain)
        elif rm_model_name == 'opt':
            reward_model = OPTRM(pretrained=args.rm_pretrain)
        elif rm_model_name == 'llama':
            reward_model = LlamaRM(pretrained=args.rm_pretrain)
        elif rm_model_name == 'roberta':
            reward_model = RoBERTaRM(pretrained=args.rm_pretrain)
        else:
            raise ValueError(f'Unsupported reward model "{rm_model_name}"')

        if args.rm_path is not None:
            reward_model.load_state_dict(state_dict)

        if rm_model_name == 'gpt2':
            critic = GPTCritic(pretrained=args.rm_pretrain, lora_rank=args.lora_rank, use_action_mask=True)
        elif rm_model_name == 'bloom':
            critic = BLOOMCritic(pretrained=args.rm_pretrain, lora_rank=args.lora_rank, use_action_mask=True)
        elif rm_model_name == 'opt':
            critic = OPTCritic(pretrained=args.rm_pretrain, lora_rank=args.lora_rank, use_action_mask=True)
        elif rm_model_name == 'llama':
            critic = LlamaCritic(pretrained=args.rm_pretrain, lora_rank=args.lora_rank, use_action_mask=True)
        elif rm_model_name == 'roberta':
            critic = RoBERTaCritic(pretrained=args.rm_pretrain, lora_rank=args.lora_rank, use_action_mask=True)
        else:
            raise ValueError(f'Unsupported reward model "{rm_model_name}"')

        if args.rm_path is not None:
            critic.load_state_dict(state_dict)
            del state_dict

critic and reward_model have the same function to init, but have different parameter of lora_rank. when I run the train_prompts.sh with --lora_rank and --rm_path, I meet errors about loading state_dict: Missing keys in state_dict : "model.h.0.self_attention.query_key_value.lora_A", "model.h.0.self_attention.query_key_value.lora_B", .....

Environment

No response