Update README.md

NeonHo · Jul 5, 2022 · bbb0a7e · bbb0a7e
1 parent 7fc9297
commit bbb0a7e
Showing 1 changed file with 2 additions and 2 deletions.
diff --git a/configs/repopt/README.md b/configs/repopt/README.md
@@ -13,13 +13,13 @@ The training of V6-RepOpt can be divided into two stages, hyperparameter search
   ```
   python tools/train.py --batch 32 --conf configs/repopt/yolov6s_hs.py --data data/coco.yaml --device 0
   ```
-  Or you can directly use the [pretrained scale](/assert/yolov6s_scale.pt) we provided and omit this stage.
+  Or you can directly use the [pretrained scale](https://github.com/xingyueye/YOLOv6/releases/download/0.1.0/yolov6s_scale.pt) we provided and omit this stage.
 2. Training. Add the flag of `training_mode='repopt'` and pretraind model `pretrained='../run/exp/weights/last_ckpt.pt',` in your config
   ```
   python tools/train.py --batch 32 --conf configs/repopt/yolov6s_opt.py --data data/coco.yaml --device 0
   ```
 ## Evaluation
-Reproduce mAP on COCO val2017 dataset, you can directly test our [pretrained model](/assert/yolov6s_opt.pt).
+Reproduce mAP on COCO val2017 dataset, you can directly test our [pretrained model](https://github.com/xingyueye/YOLOv6/releases/download/0.1.0/yolov6s_opt.pt).
   ```
   python tools/eval.py --data data/coco.yaml --batch 32 --weights yolov6s_opt.pt --task val
   ```