RuntimeError: stack expects each tensor to be equal size, but got [76, 4] at entry 0 and [91, 4] at entry 1

[Valder-Field]

According to the instructions, I have installed docker, and starting training,the error occurs when training to 249/15000 rounds, as shown in the follow figure:

After query, it is said that the dimensions are inconsistent when the Dataloader merge data...
How could I fix it? Thanks.

[zxll0106]

I have the same problem. Have you solved the problem yet?

[mzgmzg]

@zxll0106 I noticed your question in Issue #2. Have you resolved this issue? I also have the same problem. Thank you.