Hugging Face's logo

mmoz-root
/
kernelvision

Object Detection
ultralytics
ONNX
TensorRT
yolov8
fp16
gpu-optimization
computer-vision
Model card Files
xet
 Community
mmoz-root commited on
Commit
5fe5a4c
·
verified ·
1 Parent(s): 80597a6

Clarify artifact size units

Browse files
Files changed (1) hide show
  1. README.md +2 -2
README.md CHANGED
@@ -27,8 +27,8 @@ new set of learned weights.
27
 
28
  | File | Graph precision | Input | Raw output | Size |
29
  |---|---|---|---|---:|
30
- | `yolov8n_fp32.onnx` | FP32 | FP32 `[1, 3, 640, 640]` | FP32 `[1, 84, 8400]` | 12.2 MB |
31
- | `yolov8n_mixed_fp16.onnx` | ModelOpt mixed FP16/FP32 | FP16 `[1, 3, 640, 640]` | FP16 `[1, 84, 8400]` | 6.2 MB |
32
 
33
  Both graphs use ONNX opset 17 and a static batch size and image shape. The raw
34
  output contains pre-NMS box and class predictions; confidence filtering,
 
27
 
28
  | File | Graph precision | Input | Raw output | Size |
29
  |---|---|---|---|---:|
30
+ | `yolov8n_fp32.onnx` | FP32 | FP32 `[1, 3, 640, 640]` | FP32 `[1, 84, 8400]` | 12.2 MiB |
31
+ | `yolov8n_mixed_fp16.onnx` | ModelOpt mixed FP16/FP32 | FP16 `[1, 3, 640, 640]` | FP16 `[1, 84, 8400]` | 6.2 MiB |
32
 
33
  Both graphs use ONNX opset 17 and a static batch size and image shape. The raw
34
  output contains pre-NMS box and class predictions; confidence filtering,