Why did you loss 0.01 after 1 epoch when you started training from zero？

wbenbihi / hourglasstensorflow

Tensorflow implementation of Stacked Hourglass Networks for Human Pose Estimation

MIT License

480 stars 179 forks source link

Why did you loss 0.01 after 1 epoch when you started training from zero？ #68

Closed xiaoxin05 closed 2 years ago

xiaoxin05 commented 3 years ago

hi，First of all, thank you for sharing 1）I see that I'm using the cross-entropy loss function here？ 2）The training data was 20948，but An epoch appears later，loss is 0.01？