* adaption for CogVideoX1.5 * add patch_size_t in full finetuning of T2V and lora finetuning of I2V * add I2V sft and add ofs to adapt cogvideox 1.5