Willian7004
Willian7004
I got a similar issue when running the demo with `tools/replay/replay --demo` on WSL2 . The demo works well but I got a similar error when starting the ui. ```...
> I fix the seed and try three different sampling procedure: > > 1. diffusers pipeline 2. diffusers model + self dpm-solver 3. self model self dpm-solver > I got...
I got a similar issue when using Siliconflow as OpenAI Compatible provider, it didn't happen when using Openrouter and Deepseek.
I ran on huggingface's hub to run through the process without using the GPU and had the same problem. The same problem occurs when I start training after the dataset...
SANA1.5 4.8b also works on 8g VRAM with ` device_map="balanced" `
> 音频回放还没加。8gen1 下跑音频输出应该是有点慢的。 音频输入可以再试试，录个视频看下。 再次测试了，音频输入应该是长度问题，剪到20秒的话能正常输入。为了处理较长的音频，可以考虑添加滑动窗口等优化方法。 
> 您好，ArkInfer将于近期开源，敬请期待！ 已经过一个月了，有正式消息吗？目前llama.cpp实现不了MoE加速，非cuda设备只能靠Arkinfer了。
I tried llama.cpp and got the same error, I think the model architecture is actually changed. I asked about llama.cpp support in the model repo before but didn't get a...
I got a reply of another developer in huggingface recently, the gguf version was created on mradermacher/Cosmos-1.0-Prompt-Upsampler-12B-Text2World-hf-GGUF.
I have a similar issue, and the model I use is not supported in v0.3.6.