Paddle之路

【1】

Traceback (most recent call last):
  File "run_robust.py", line 28, in 
    from paddlenlp.data import Pad, Stack, Tuple, Dict
  File "/home/qxn/py3_qxn/lib/python3.6/site-packages/paddlenlp/__init__.py", line 24, in 
    from . import data
  File "/home/qxn/py3_qxn/lib/python3.6/site-packages/paddlenlp/data/__init__.py", line 16, in 
    from .data_collator import *
  File "/home/qxn/py3_qxn/lib/python3.6/site-packages/paddlenlp/data/data_collator.py", line 23, in 
    from ..transformers.tokenizer_utils_base import BatchEncoding, PretrainedTokenizerBase, PaddingStrategy
  File "/home/qxn/py3_qxn/lib/python3.6/site-packages/paddlenlp/transformers/__init__.py", line 72, in 
    from .megatronbert.modeling import *
  File "/home/qxn/py3_qxn/lib/python3.6/site-packages/paddlenlp/transformers/megatronbert/modeling.py", line 20, in 
    from paddle import einsum
ImportError: cannot import name 'einsum'

解决：重装

【2】

Traceback (most recent call last):
  File "run_robust.py", line 485, in 
    run(args)
  File "run_robust.py", line 113, in run
    paddle.set_device(args.device)
  File "/home/qxn/py3_qxn/lib/python3.6/site-packages/paddle/device/__init__.py", line 313, in set_device
    place = _convert_to_place(device)
  File "/home/qxn/py3_qxn/lib/python3.6/site-packages/paddle/device/__init__.py", line 202, in _convert_to_place
    raise ValueError("The device should not be 'gpu', "
ValueError: The device should not be 'gpu', since PaddlePaddle is not compiled with CUDA

解决：官网安装GPU版本Paddle，注意Cuda版本，Cuda版本取决于nvcc -V