[11:11:54:265809]:[Metadata] AssetAttributesManager.cpp(110)::AssetAttributesManager : Built default primitive asset templates : 12
[11:11:54:266126]:[Metadata] SceneDatasetAttributesManager.cpp(308)::readDatasetJSONCell : "stages.default_attributes" set in Attributes Manager from JSON.
[11:11:54:266182]:[Metadata] AttributesManagerBase.h(357)::buildAttrSrcPathsFromJSONAndLoad : No Glob path result for /root/sound-spaces/data/scene_da tasets/data/versioned_data/hm3d-1.0/hm3d/example/00861-GLAQ4DNUx5U/.basis.glb
[11:11:54:266212]:[Metadata] AttributesManagerBase.h(357)::buildAttrSrcPathsFromJSONAndLoad : No Glob path result for /root/sound-spaces/data/scene_da tasets/data/versioned_data/hm3d-1.0/hm3d/minival/00800-TEEsavR23oF/.basis.glb
[11:11:54:266240]:[Metadata] AttributesManagerBase.h(357)::buildAttrSrcPathsFromJSONAndLoad : No Glob path result for /root/sound-spaces/data/scene_da tasets/data/versioned_data/hm3d-1.0/hm3d/minival/00802-wcojb4TFT35/*.basis.glb
Hello, I trained the model the slfm_pretext and then slfm_geometric. However, when I evaluate the geometric model, the graph shows that only from 200-1000 are accurately predicted. However, from 0-200 and 1000-1200, are every inaccurately predicted. I was wondering if this is something wrong with the evaluate_angle code. Please look at the attachment
After running the multi-preprocess, I met thist problems:
Platform::WindowlessEglApplication::tryCreateContext(): cannot get default EGL display: EGL_BAD_PARAMETER
WindowlessContext: Unable to create windowless context
Hello, I was trying the preset training with resnet-50. However, the error shows that there's a key not found error, and a size mismatch error. The only thing I changed was in slfm-pretext.py. I've attached the screenshots below. Please take a look. Thank you so much!