LongxingTan

@hieuth133 The last dimension of y_true has 6 values. - the first four value is the box coordinates, like xywh or xyxy - the 5th value is the confidence of...

Hi @Darkhunter9 , Thanks for your carefully check about the code. you found so many bug and kindly give solutions, I will check that. Thanks

@tak-s Hey, if loss is nan and the training steps are more than one epoch and less than the warmup epochs, you can try decreasing the learning rate. As for...

@tak-s yes, that's right. 1. run get_voc.sh first 2. python dataset/prepare_data.py 3. python train.py but I checked my result, I guess i only run 20 epochs for that result due...

@gongkecun 你好，loss还是出现NAN的话，建议按一下常见步骤查看： - 确保输入数据没有nan、没有越界、格式与处理时相符，如果nan出现在第一个epoch可能是数据的原因，也可能是学习率过大的原因 - 数据没问题的前提下，如果nan出现在大于1epoch，小于warmup-epochs阶段，则降低warmup阶段最大学习率

Hi @kalikhademi , The anchor issue I will check the code. If the total loss is 200, it looks too big. You can check each item's loss (box_loss, confidence_loss, class_loss)...

@lee-liew 这个看上去是预测时数据输入范围的问题。 比如训练时这个数据是从0-11的，embedding只到类别11。但是预测时，你的输入有12，embed查找时越界了。

嗯，你这部分理解的是没错的。 模型的输入除了时空这部分外，还有月份的。(时间比较久了，我记得也不一定清楚) 但这个12应该是来自月份的embedding，可以参考一下[inference script](https://github.com/LongxingTan/Data-competitions/blob/master/tianchi-enso-prediction/code/final/inference.py) 里对于月份的输入。

@plyu3 空间信息主要是手动提取了一些业务特征，除了待预测的NINO3.4以外，其他还有一些NINO12，NINO3，NINO4这样的定义，[可参考相关代码和注释](https://github.com/LongxingTan/Data-competitions/blob/40af4f620fcc145a8cb3cd0cbcade5a4277e21eb/tianchi-enso-prediction/code/final/train.py#L28-L109)

> Hey, > > this error came up when i start to train any model. I also tried it on my other pc, its probably a configuration problem. > Thanks...