* Add chatglm3, qwen15-7b and baichuan-7b QLoRA alpaca example * Remove unnecessary tokenization setting.