learning rate adjustment
nitba opened this issue · 1 comments
nitba commented
Hi @bearpaw
the plot from log file shows that network is trained for 220
epoch and we have adjustment on 150th
epoch,
but in code , I did not find any adjustment , would you please share adjustment for learning rate to get your results?
bearpaw commented
You can use arguments to set the epoch, lr, and lr decay strategy
- for training epoch, see https://github.com/bearpaw/pytorch-pose/blob/master/example/main.py#L383
- for lr decay, see: https://github.com/bearpaw/pytorch-pose/blob/master/example/main.py#L397