shensheng272

> > add `--master_port 30001` or other value in start command, for example: > > ``` > > python -m torch.distributed.launch --nproc_per_node 8 --master_port 30002 tools/train.py ... > > ```...

For now, `python tools/eval.py` only support pt weights. To run onnx as infer, you might export onnx with `--end2end `, then try onnxruntime or other supported [onnx](https://github.com/onnx/onnx) backend. If you...

Hi, thanks for reproduce exp. You might try --batch 128 for n and s. 0.2~0.5% gap is normal because of random seed.

你好，日志看起来使用的命令是 "python tools/train.py" 吗？ 应在训练命中中指定对应的数据集参数 例如 python tools/train.py --batch 256 --conf configs/yolov6s_finetune.py --data data/data.yaml --device 0 可以参考 https://github.com/meituan/YOLOv6/blob/main/docs/tutorial_voc.ipynb 以及 https://github.com/meituan/YOLOv6/blob/main/docs/Train_custom_data.md

Hi, thanks for demo. How about adding the links to the Third-party resources in README ? Like https://github.com/meituan/YOLOv6/pull/288

https://github.com/meituan/YOLOv6/issues/98 the same problem, might try pull the lastest code

https://github.com/meituan/YOLOv6/pull/309 fix this, plz try again

> ``` > pred = anno.loadRes(pred_json) > cocoEval = COCOeval(anno, pred, 'bbox') > ``` > > 这个地方会报错，不知道作者能不能弄一个验证的或者MAP的数据格式 @the-cat-crying 拉取新代码，删掉原本生成的 *_instance.json 试试

Log has show iou_loss is nan, the train might failed. Your dataset is quite small, did you use finetune. Could you show your train script?

any description?