运行scalar_encoder.test()结果输出非法值

[张华卿]

您好，我通过test.py运行scalar_encoder.test()函数，打印embedded_scalar，结果如下

(base) zhq@Ubuntu20:~/Doctor/RL_Project/mini-AlphaStar$ python test.py 
pygame 2.0.1 (SDL 2.0.14, Python 3.8.5)
Hello from the pygame community. https://www.pygame.org/contribute.html
test init
embedded_scalar: tensor([[nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan],
        [nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan,
         nan, nan, nan, nan, nan, nan, nan, nan]], grad_fn=<ReluBackward0>)
embedded_scalar.shape: torch.Size([2, 128])
test over

我运行您代码库中 rl_vs_computer_wo_replay.test()，运行出错，我在想：是不是scalar_encoder.test()中打印embedded_scalar出现的错误的原因也是导致 rl_vs_computer_wo_replay.test()出错的原因呢？

[liuruoze]

可以逐一调试，看看是哪行的代码首次导致了nan的出现。

torch有一个语句可以方便的监测向量中是否有nan的： "torch.isnan"

在delay_head中的checkNaNandInf函数也可以或许给你参考

[张华卿]

多谢啦，我再调试下