Even
Even
【台湾车牌】
是否支持台湾车牌生成?
Thanks for the excellent work! May i ask whether this method could handle variable number of lane lines?
Hi, thanks for the excellent work! I found b0 "cityscape" semantic image size is only 1/8 of the input image size? May i ask where the problem is?
请问LLaMA-Factory的SFT训练(假设用的是QWen2.5-VL基座系列),是固定分辨率还是动态分辨率? 因为LLaMA-Factory的配置文件中有分辨率方面的参数设置: 、、、 ### model model_name_or_path: Qwen/Qwen2.5-VL-3B-Instruct image_resolution: 262144 video_resolution: 16384 trust_remote_code: true 、、、 如果是固定分辨率,需要提前(训练之前)把图像处理成方形,且为28倍数吗? 如果是动态分辨,需要提前(训练之前)再尽量保持宽高比的前提下,将图像处理成28倍数吗?