Poetry generation is an interesting research topic in the field of text\ngeneration. As one of the most valuable literary and cultural heritages of\nChina, Chinese classical poetry is very familiar and loved by Chinese people\nfrom generation to generation. It has many particular characteristics in its\nlanguage structure, ranging from form, sound to meaning, thus is regarded as an\nideal testing task for text generation. In this paper, we propose a GPT-2 based\nuniformed framework for generating major types of Chinese classical poems. We\ndefine a unified format for formulating all types of training samples by\nintegrating detailed form information, then present a simple form-stressed\nweighting method in GPT-2 to strengthen the control to the form of the\ngenerated poems, with special emphasis on those forms with longer body length.\nPreliminary experimental results show this enhanced model can generate Chinese\nclassical poems of major types with high quality in both form and content,\nvalidating the effectiveness of the proposed strategy. The model has been\nincorporated into Jiuge, the most influential Chinese classical poetry\ngeneration system developed by Tsinghua University (Guo et al., 2019).\n