onedotone-wei issues

Results 5 issues of


                                            onedotone-wei

FileNotFoundError: [Errno 2] No such file or directory: '/ps/scratch/face2d3d/texture_in_the_wild_code/VGGFace2_cleaning_codes/ringnetpp_training_lists/second_cleaning/vggface2_train_list_max_normal_100_ring_5_1_serial.npy'

FileNotFoundError: [Errno 2] No such file or directory: '/ps/scratch/face2d3d/texture_in_the_wild_code/VGGFace2_cleaning_codes/ringnetpp_training_lists/second_cleaning/vggface2_train_list_max_normal_100_ring_5_1_serial.npy' I found this error when I ran the example, Where can I get or download this file?

手势始终存在

使用了最新代码，使用命令python -m scripts.audio2vid --config ./configs/prompts/animation_audio.yaml -W 512 -H 512 -acc时，手势还是会有 https://github.com/Zejun-Yang/AniPortrait/assets/51695272/0789d65f-4f9f-4db7-ba51-54b698eb2376 我上传了测试视频，从20秒开始 ![1](https://github.com/Zejun-Yang/AniPortrait/assets/51695272/3a0d23d5-2ae5-4511-9fd3-67d7c2245289) ![2](https://github.com/Zejun-Yang/AniPortrait/assets/51695272/1261f1d9-00af-4040-92e7-62427613d314) ![3](https://github.com/Zejun-Yang/AniPortrait/assets/51695272/50a10d17-795e-4401-b2ea-34d47fc9c55c)

作者大大，可以加微有偿学习一下吗？

V:leavesofautumn

请问作者大大，encoder如何训练？

看了知乎链接的教程，尝试训练encoder ![Screenshot 2024-04-29 184437](https://github.com/babysor/MockingBird/assets/51695272/901047ec-0d6a-4dd8-842a-6d813eae230c) 练了半天，这结果似乎没什么变化 ![wei_umap_038600](https://github.com/babysor/MockingBird/assets/51695272/b924f344-9ab0-429e-bfaa-ea6dde7c791f) ![wei_umap_038700](https://github.com/babysor/MockingBird/assets/51695272/7d0b5ba1-9105-487b-8be5-d50ae50bd88d) 数据自建的，有2个多G 问题1：这种情况是正常的吗？如果不正常是什么原因造成的？问题2：根据知乎上的说法“**实测了一次训练synthesizer时，4000左右step就能attention收敛，22k step的时候loss就到0.35了，可以很快进行finetune，算是超越预期。**”，训练synthesizer时，如何把encoder加入？

请问，这是练丹练坏了吗？

https://github.com/Zejun-Yang/AniPortrait/assets/51695272/1fe11dac-3a5a-48d4-9387-49a68c92b913 作者大大，这是怎么回事？数据问题？训练次数不够？