Search results

因為顯存不夠，只有16GB，想詢問是否能使用v1.5的版本？

Status: Open.

#335 In bytedance/LatentSync;

· HaoHaohedgehog opened on Sep 15, 2025

hello，重训Unet后infer画质会下降，人脸有一点抖动，这个怎么解决

Status: Open.

#334 In bytedance/LatentSync;

· gangxu822 opened on Sep 9, 2025

hello，训练syncnet一般需要多少数据？训练时候也不是从0开始训吧，要把如下的resume_ckpt_path设置成作者发布的pretrain model（checkpoints/stable_syncnet.pt）?

Status: Open.

#333 In bytedance/LatentSync;

· gangxu822 opened on Sep 6, 2025

Use SageAttention2?

Status: Open.

#332 In bytedance/LatentSync;

· btakita opened on Sep 5, 2025

什么情况下需要重训syncnet网络？

Status: Open.

#331 In bytedance/LatentSync;

· gangxu822 opened on Sep 4, 2025

train_unet时候这个报错什么问题？

Status: Open.

#330 In bytedance/LatentSync;

· gangxu822 opened on Sep 4, 2025

请教train_unet的问题？

Status: Open.

#329 In bytedance/LatentSync;

· gangxu822 opened on Sep 4, 2025

Mask blurry in the face also with everything at maximum

Status: Open.

#328 In bytedance/LatentSync;

· thejack999 opened on Aug 23, 2025

How to train on 1 (or 2) RTX 5090's?

Status: Open.

#327 In bytedance/LatentSync;

· etale-cohomology opened on Aug 20, 2025

How to run inference on longer video?

Status: Open.

#326 In bytedance/LatentSync;

· etale-cohomology opened on Aug 20, 2025

Data preprocessing really shrinks the data size.

Status: Open.

#325 In bytedance/LatentSync;

· sjk0825 opened on Aug 17, 2025

A100推理生成爆显存，如何实时释放资源

Status: Open.

#321 In bytedance/LatentSync;

· turkeymz opened on Aug 4, 2025