Instructions to use mlx-community/sam-3d-objects-bf16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use mlx-community/sam-3d-objects-bf16 with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir sam-3d-objects-bf16 mlx-community/sam-3d-objects-bf16
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Atomic Chat
File size: 2,828 Bytes
b846e48 26f21f6 b846e48 26f21f6 b846e48 26f21f6 b846e48 26f21f6 b846e48 26f21f6 b846e48 26f21f6 b846e48 26f21f6 b846e48 26f21f6 b846e48 26f21f6 b846e48 26f21f6 b846e48 26f21f6 b846e48 26f21f6 b846e48 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 | {
"mlx_version": "0.32.2",
"tensor_count": 3895,
"parameter_elements": 3530447862,
"weight_bytes": 7801464522,
"dtypes": {
"sam": {
"bfloat16": 3288
},
"depth_model": {
"float32": 607
}
},
"strict_load": {
"missing": 0,
"extra": 0,
"shape_mismatches": 0
},
"trainable_tensors": 0,
"regression_tests": {
"class": "TestSAM3DObjects",
"passed": 17
},
"reference_comparison": {
"precision": "float32",
"mlx_device": "CPU",
"reference": "Original PyTorch source; CPU adapter for the unavailable spconv backend",
"maximum_absolute_errors": {
"attention_False": 1.1920928955078125e-07,
"attention_True": 1.1920928955078125e-07,
"resize_False_False": 2.384185791015625e-07,
"resize_False_True": 1.430511474609375e-06,
"resize_True_False": 3.5762786865234375e-07,
"resize_True_True": 9.5367431640625e-07,
"structure_6drotation_normalized": 1.1920928955078125e-07,
"structure_scale": 4.76837158203125e-07,
"structure_shape": 4.76837158203125e-07,
"structure_translation": 1.1920928955078125e-07,
"structure_translation_scale": 2.9802322387695312e-08,
"occupancy_decoder": 1.6391277313232422e-06,
"sparse_latent_flow": 2.942979335784912e-07,
"flexicubes_vertices": 2.9802322387695312e-08,
"flexicubes_colors": 1.1920928955078125e-07,
"flexicubes_faces": 0.0
}
},
"depth_model": {
"source": "mlx-community/moge-3-vitl-mlx-fp32",
"base": "Ruicheng/moge-3-vitl",
"validation": "MoGe-3 MLX port validated against the PyTorch reference in mlx-vlm (CPU float32): identical valid-pixel masks and under 0.2% median relative depth error"
},
"benchmark": {
"hardware": "Apple M5 Max, 128 GiB unified memory",
"dtype": "bfloat16 (SAM), float32 (MoGe-3)",
"steps": [
25,
25
],
"input_size": [
448,
672
],
"input": "id3_shutterstock_WildAnimal_Waterhole_2010559391/image.png and 0.png resized to 672x448",
"formats": [
"gaussian",
"gaussian_4",
"mesh"
],
"samples": [
{
"seconds": 10.577361750009004,
"peak_memory_gb": 9.539739473
},
{
"seconds": 10.955263582989573,
"peak_memory_gb": 9.578930137
},
{
"seconds": 11.333075082977302,
"peak_memory_gb": 9.518227537
}
],
"median_seconds": 10.955263582989573,
"peak_memory_gb": 9.578930137,
"gaussians": 298112,
"mesh_vertices": 330848,
"mesh_faces": 655702,
"pointmap_conditioned": true,
"identical_latents_across_runs": true,
"async_validation": {
"request_ids": [
"0",
"1"
],
"event_loop_heartbeats": 3955,
"latents_match_synchronous": true
}
}
}
|