codingfun2022
Results
3
comments of
codingfun2022
Using estimator = ms.PyTorchEstimator(..., updater=ms.AdamTensorUpdater(adam_learning_rate), ...) is better.
You probably forgot to call model.half(). I encountered the same error, after I added the model.half() invocation, the error disappeared. The example code in README.md contains the model.half() invocation, though...
Linux 下可能可以这样解决,参考: Support loading cuda libraries from nvidia package. https://github.com/OpenBMB/cpm_kernels/pull/8