| model: |
| name: hrnet |
| display_name: HRNet-W48 |
| upstream: open-mmlab/mmpretrain |
| source: |
| kind: github |
| id: open-mmlab/mmpretrain |
| url: https://github.com/open-mmlab/mmpretrain/blob/main/configs/hrnet/hrnet-w48_4xb32_in1k.py |
|
|
| architecture: |
| family: hrnet |
| variant: w48 |
| dataset: imagenet-1k |
| input_resolution: 224x224 |
| num_classes: 1000 |
| modality: |
| - vision |
| parameters: 77.4M |
| parameters_source: exact |
| parameters_note: >- |
| computed directly from the shipped fp32 ONNX graph's initializer |
| tensors (sum of onnx.numpy_helper.to_array(t).size for each |
| graph.initializer), not a cited published figure. |
| |
| format: |
| type: onnx |
| opset: 12 |
|
|
| tasks: |
| - image-classification |
|
|
