0

我必须在 pipeline.config 中设置一个类,我将类数更改为 1,然后我开始训练它显示错误

InvalidArgumentError (see above for traceback): Restoring from checkpoint failed. 

这很可能是由于当前图与检查点的图不匹配造成的。

请确保您没有根据检查点更改预期的图表。原始错误:

Assign requires shapes of both tensors to match. lhs shape= [1,1,256,12] rhs shape= [1,1,256,546]
[[node save/Assign_86 (defined at /home/drive/Downloads/models- 
master/research/object_detection/legacy/trainer.py:378) @bmabey @alextp 
@moonboots @ewilderj @jtoy @tensorflowbutler

还有如何设置微调检查点?

4

0 回答 0