Yang Yong(雍洋)

您好，请问在xnornet代码中，看到了alpha，没有看到特征对应的K，请问这个在哪里实现的？

谢谢回复！ 我在xnor-net++论文( https://bmvc2019.org/wp-content/uploads/papers/0121-paper.pdf )里面，发现了这个表述： “Note, that as the calculation of K is relatively expensive due to the fact that it is recomputed at each forward pass, it is common to...

You can set save_fp in llmc to True. Then you can use trt-llm ammo to convert a naive quant engine.

[general_custom_data_examples.zip](https://github.com/user-attachments/files/18433608/general_custom_data_examples.zip) @aptsunny

Thanks for reaching out and for your interest in LightX2V! At the moment, we don’t have concrete plans or a committed timeline to add native support for AMD GPUs (such...

@baihei-1122 You can try tiny vae. Please refer : English Doc: https://lightx2v-en.readthedocs.io/en/latest/deploy_guides/for_low_latency.html Chinese Doc: https://lightx2v-zhcn.readthedocs.io/zh-cn/latest/deploy_guides/for_low_latency.html download taew2_1.pth from https://github.com/madebyollin/taehv/raw/refs/heads/main/taew2_1.pth { "use_tiny_vae": true, "tiny_vae_path": "/path to taew2_1.pth" } Note: tiny vae...

可以结合使用，直接在你的offload的config中，增加： ``` "parallel": { "cfg_p_size": 2 } ``` 并且启动脚本改成 ``` torchrun --nproc_per_node=2 -m lightx2v.infer ```

Not support vace now. We are going to support it soon.

Please show me your bash script file and your config file. Thanks! @yogesh030

https://github.com/ModelTC/LightX2V/blob/main/scripts/dist_infer/run_wan_t2v_dist_cfg_ulysses.sh You can try this script. I do not see any bug. @yogesh030 You should give me your `default script`, so that I can help you.