CatDroid
CatDroid
**Describe the bug** run ShaderConductorTest ,and crash on LinkTest **To Reproduce** 1. vs2019 on Win10 Debug and x64 build 2. commit id: 30a77c78d24fa08f4fe5fc4428f10dbfc92717a6 branch: maset 3. start up with ShaderConductorTest...
**Environment information:** - AGI version: Version: 3.3.1 - Host OS: Windows If tracing on Android: - Device model: Google Pixel 6 - Android Version: Android 12 **Bug description** 1. In...
[AGI document](https://developer.android.com/agi/sys-trace/long) For example, [gpu-counters-adreno](https://developer.android.com/agi/sys-trace/long#gpu-counters-adreno) "Figure 20. The Vertex Instructions / Second track. " but the picutre above this text has nothing about "vertex instructions/Second" information. The other pictures of...
``` Real-time Segmenting Human Portrait at Anywhere Ruifeng Yuan, Yuhao Cheng♯ , Yiqiang Yan, Haiyan Liu Lenovo Research Buidling1, No.10 Courtyard Xibeiwang East Road, Beijing, China ``` I found it...
follow "https://github.com/SHI-Labs/VMFormer/blob/main/INSTALL.md" Step 5: Install CUDA version of MultiScaleDeformableAttention cd ./models/ops sh ./make.sh python test.py test.py will error “ "error in ms_deformable_col2im_cuda: too many resources requested for launch\n",” 
参考 https://forums.imgtec.com/t/does-pvrshadereditor-support-gl-lastfragdata/3317/3
``` vertex_data = defaultdict(OrderedDict) ``` the order is 'insert' order , not base on IDX from 0 to MAX
in utils/util.py ``` if "ce" in loss_type or "kd" in loss_type: return 0 if epoch
in KLDiscretLoss (classification_loss.py) log_pt's or labels's shape is [B*K, width] or [B*K, height] width or height is ‘number of categories' ``` def criterion(self, dec_outs, labels): ...... loss = torch.mean(self.kl_loss(log_pt, labels),...
branch: onnx DWPose\mmpose\mmpose\models\backbones no file or backbone is CSPNeXt ？ but config file 'DWPose\mmpose\configs\wholebody_2d_keypoint\rtmpose\ubody\rtmpose-l_8xb64-270e_coco-ubody-wholebody-256x192.py' use CSPNeXt as backbone ?