📂 Original input file copied to: /tmp/unirig_prediction/input_copy/tmp8ao85pghtripo_pbr_model_8b2ddca8-9dbc-4310-b30c-f8c5f1d1deb6.glb ℹ️ Created datalist for run.py at /app/repo/dataset_inference_clean/inference_datalist.txt pointing to 'cog_predict_data_dir' --- 📝 Stage 1a: Data Extraction (raw_data.npz) --- 🛠️ Creating NPZ for /tmp/unirig_prediction/input_copy/tmp8ao85pghtripo_pbr_model_8b2ddca8-9dbc-4310-b30c-f8c5f1d1deb6.glb at /app/repo/dataset_inference_clean/cog_predict_data_dir/raw_data.npz Mesh has 267738 faces, simplifying to 50000... ✅ NPZ file created at: /app/repo/dataset_inference_clean/cog_predict_data_dir/raw_data.npz --- 💀 Stage 1b: Skeleton Prediction --- 🚀 Running command: python /app/repo/run.py --task /app/repo/configs/task/quick_inference_skeleton_articulationxl_ar_256.yaml in /app/repo /root/.pyenv/versions/3.11.10/lib/python3.11/site-packages/timm/models/layers/__init__.py:48: FutureWarning: Importing from timm.models.layers is deprecated, please import via timm.layers warnings.warn(f"Importing from {__name__} is deprecated, please import via timm.layers", FutureWarning) Seed set to 123 [92mload task config: /app/repo/configs/task/quick_inference_skeleton_articulationxl_ar_256.yaml[0m [92mload data config: configs/data/quick_inference.yaml[0m [92mload transform config: configs/transform/inference_ar_transform.yaml[0m [92mload tokenizer config: configs/tokenizer/tokenizer_parts_articulationxl_256.yaml[0m [92mload model config: configs/model/unirig_ar_350m_1024_81920_float32.yaml[0m Flash Attention 2.0 only supports torch.float16 and torch.bfloat16 dtypes, but the current dype in OPTForCausalLM is torch.float32. You should run training or inference using Automatic Mixed-Precision via the `with torch.autocast(device_type='torch_device'):` decorator, or load the model with the `torch_dtype` argument. Example: `model = AutoModel.from_pretrained("openai/whisper-tiny", attn_implementation="flash_attention_2", torch_dtype=torch.float16)` [92mload system config: configs/system/ar_inference_articulationxl.yaml[0m Using bfloat16 Automatic Mixed Precision (AMP) Using default `ModelCheckpoint`. Consider installing `litmodels` package to enable `LitModelCheckpoint` for automatic upload to the Lightning model registry. GPU available: True (cuda), used: True TPU available: False, using: 0 TPU cores HPU available: False, using: 0 HPUs /root/.pyenv/versions/3.11.10/lib/python3.11/site-packages/lightning/pytorch/trainer/connectors/logger_connector/logger_connector.py:76: Starting from v1.9.0, `tensorboardX` has been removed as a dependency of the `lightning.pytorch` package, due to potential conflicts with other packages in the ML ecosystem. For this reason, `logger=True` will use `CSVLogger` as the default logger, unless the `tensorboard` or `tensorboardX` packages are found. Please `pip install lightning[extra]` or one of them to enable TensorBoard support by default Restoring states from the checkpoint path at /root/.cache/huggingface/hub/models--VAST-AI--UniRig/snapshots/e53ad6237db5cea261727c5d50b67d140a4b9571/skeleton/articulation-xl_quantization_256/model.ckpt LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0] Loaded model weights from the checkpoint at /root/.cache/huggingface/hub/models--VAST-AI--UniRig/snapshots/e53ad6237db5cea261727c5d50b67d140a4b9571/skeleton/articulation-xl_quantization_256/model.ckpt /root/.pyenv/versions/3.11.10/lib/python3.11/site-packages/lightning/pytorch/trainer/connectors/data_connector.py:425: The 'predict_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` to `num_workers=9` in the `DataLoader` to improve performance. Predicting: | | 0/? [00:00