Image Segmentation
Transformers
Safetensors
rf_detr
instance-segmentation
rf-detr
roboflow
coco
quantized
nvfp4
fp4
experimental
Instructions to use Reza2kn/rf-detr-segmentation-NVFP4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Reza2kn/rf-detr-segmentation-NVFP4 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("image-segmentation", model="Reza2kn/rf-detr-segmentation-NVFP4")# Load model directly from transformers import AutoImageProcessor, RfDetrForInstanceSegmentation processor = AutoImageProcessor.from_pretrained("Reza2kn/rf-detr-segmentation-NVFP4") model = RfDetrForInstanceSegmentation.from_pretrained("Reza2kn/rf-detr-segmentation-NVFP4") - Notebooks
- Google Colab
- Kaggle
| { | |
| "base": "/home/rezo/rfdetr-nvfp4/rf-detr-segmentation", | |
| "quant": "/home/rezo/rfdetr-nvfp4/rf-detr-segmentation-nvfp4-heads", | |
| "device": "cuda", | |
| "dtype": "bfloat16", | |
| "metrics": { | |
| "logits": { | |
| "shape": [ | |
| 1, | |
| 200, | |
| 91 | |
| ], | |
| "rel_l2": 0.12226187437772751, | |
| "cosine": 0.9925256371498108, | |
| "max_abs": 4.203125 | |
| }, | |
| "pred_boxes": { | |
| "shape": [ | |
| 1, | |
| 200, | |
| 4 | |
| ], | |
| "rel_l2": 0.5620495676994324, | |
| "cosine": 0.8393532633781433, | |
| "max_abs": 0.95947265625 | |
| }, | |
| "pred_masks": { | |
| "shape": [ | |
| 1, | |
| 200, | |
| 108, | |
| 108 | |
| ], | |
| "rel_l2": 0.9607120752334595, | |
| "cosine": 0.5827363729476929, | |
| "max_abs": 119.4375 | |
| } | |
| } | |
| } | |