============================================ Activated conda environment: da3 CUDA_HOME: /home/ywan0794/miniconda3/envs/da3 ============================================ === GPU Info === Wed Jan 14 15:19:23 2026 +---------------------------------------------------------------------------------------+ | NVIDIA-SMI 535.274.02 Driver Version: 535.274.02 CUDA Version: 12.2 | |-----------------------------------------+----------------------+----------------------+ | GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. | | | | MIG M. | |=========================================+======================+======================| | 0 NVIDIA A40 Off | 00000000:CA:00.0 Off | 0 | | 0% 52C P0 90W / 300W | 13MiB / 46068MiB | 0% Default | | | | N/A | +-----------------------------------------+----------------------+----------------------+ +---------------------------------------------------------------------------------------+ | Processes: | | GPU GI CI PID Type Process name GPU Memory | | ID ID Usage | |=======================================================================================| | 0 N/A N/A 2845 G /usr/lib/xorg/Xorg 4MiB | +---------------------------------------------------------------------------------------+ CUDA available: True GPU count: 1 GPU name: NVIDIA A40 ============================================ Starting MoGe Evaluation for DA3+SDT at Wed Jan 14 03:19:26 PM AEDT 2026 Decoder: sdt Checkpoint: /home/ywan0794/Depth-Anything-3/training/exp/da3_sdt_vitl_both/epoch_010.pth Output: eval_output/da3_sdt.json ============================================ [INFO ] using MLP layer as FFN /home/ywan0794/MoGe/baselines/da3_custom.py:70: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature. ckpt = torch.load(checkpoint, map_location='cpu') Loaded checkpoint from /home/ywan0794/Depth-Anything-3/training/exp/da3_sdt_vitl_both/epoch_010.pth Benchmarks: 0%| | 0/10 [00:00