Dear Authors,
Impressive work! Thank you for your open source code！
I have test your released code for OpenSora on a single NVIDIA 4090, where the Baseline inference speed is approximately 60 Sec, and the inference speed for NaviCache-Fast, NaviCache-Mid is 22.77 Sec and 26.53 Sec on my server. The result above does not align with the results reported in your paper, could you please help me confirm whether it is a bug from my side, or there may be some mistakes in the paper?
Best,
Yuhan
我已在一台 NVIDIA 4090 显卡上对 OpenSora 的已发布代码进行了测试，其基准推理速度约为 60 秒。而在我的服务器上，NaviCache-Fast 和 NaviCache-Mid 的推理速度分别为 22.77 秒和 26.53 秒。上述结果与您论文中所报告的结果不一致，能否请您帮我确认一下这到底是我的问题，还是论文中可能存在错误？
Dear Authors,
Impressive work! Thank you for your open source code！
I have test your released code for OpenSora on a single NVIDIA 4090, where the Baseline inference speed is approximately 60 Sec, and the inference speed for NaviCache-Fast, NaviCache-Mid is 22.77 Sec and 26.53 Sec on my server. The result above does not align with the results reported in your paper, could you please help me confirm whether it is a bug from my side, or there may be some mistakes in the paper?
Best,
Yuhan
我已在一台 NVIDIA 4090 显卡上对 OpenSora 的已发布代码进行了测试，其基准推理速度约为 60 秒。而在我的服务器上，NaviCache-Fast 和 NaviCache-Mid 的推理速度分别为 22.77 秒和 26.53 秒。上述结果与您论文中所报告的结果不一致，能否请您帮我确认一下这到底是我的问题，还是论文中可能存在错误？