qinb

Could you release the smoothnet training code?

3

Thanks for your releasing the SmoothNet inference code and pre-trained model. And I want to ask when you plan to release the training code? Thanks again @yl-1993

Train smoothnet using 3D pose data in mmpose

1

Thanks for your open-sourcing smoothnet! I already read the smoothnet paper, and find it inputs the pose data produced by SPIN/VIBE. So, I want to how to retrain smoothnet in...

running run_selfcontact_optimization.py in SMPL model

2

Hi, I successfully run the run_selfcontact_optimization.py in model_type='smplx', when I change model_type from smplx to smpl, and modify some codes to be compatible with it, the mismatch still occurs. Finally,...

Thank you for sharing your interesting job. Could you offer an evaluation code on the dataset you mentioned in your paper?

joint3d's z coordinate change largely

1

Hi, Thank you for sharing such excellent research. I try to infer my [own video](https://github.com/davrempe/contact-human-dynamics/blob/main/data/example_data/dance1/dance1.mp4) and obtain model_joints after the run-fitting phase, unfortunately, it's strange that model joints' z coordinate...

Thanks for sharing so fast smplx codes. As your examples: ``` seq = SequenceAMASS("./RightTurn01_poses.npz") ``` Actually, I want to input numpy array of Nx24x3 or Nx24x3x3, could you offer some...

benchncnn gpu速度慢于cpu

5

您好，我在利用benchncnn测试模型遇到一些速度问题 测试平台： linux服务器、骁龙845芯片 ncn版本： 20200727 编译链接：https://github.com/Tencent/ncnn/wiki/how-to-build#build-for-linux-x86 https://github.com/Tencent/ncnn/wiki/how-to-build#build-for-android 编译了vulkan和ncnn linux服务器 cpu利用benchncnn测试模型速度如下： ./benchncnn 4 8 loop_count = 4 num_threads = 8 powersave = 0 gpu_device = -1 cooling_down = 1 squeezenet min...

Haruhi-Zero数据组织问题

4

@LC1332 感谢作者项目的无私分享，我在学习您的数据生成时发现一些疑惑，想请教一下： 具体参考的数据 https://huggingface.co/datasets/silk-road/Haruhi-Zero/tree/main 1、对于RoleLLM-sample.jsonl，Haruhi52_sample.jsonl等，**我大致总结了一下数据组织方式，_问题 1：_ 不知是否正确？** a. 通过embedding相似度获取context，并以\n###\n分割； b. 数据组织成from,value这种shareGPT格式； c. 每个对话使用「和」括住； **疑惑：** 我跟进了您最新上传的novel_50_xx-sample.jsonl以及tranlsated_and_split_PIPPA_sample.jsonl，\ 发现似乎并不按照上述方式组织，【**似乎没有了相似度获取的context，以及每个对话使用「和」括住；**】\ **问题2：** 类似多种数据组织方式对于训练的影响大吗？