大佬 我在训练的时候遇到这个问题pytorch_lightning.utilities.exceptions.MisconfigurationException: DDPShardedPlugin requires fairscale to be installed. Install it by running pip install fairscale.，但是我明明已经安装好了还是报相同的错，这是啥原因啊，另外想问一下cpu可以跑通么

[lichengyang666]

No description provided.

[renmada]

试试改个参数，不用fairscale

--plugins ddp

[satisic]

试试改个参数，不用fairscale

--plugins ddp

大佬我也遇到了同款问题，改成ddp后报了这个错误
File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 86, in dev_collate
return self.train_collate(batch)
File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 53, in train_collate
src = [b['src'] for b in batch]
File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 53, in
src = [b['src'] for b in batch]
KeyError: 'src'

[lichengyang666]

试试改个参数，不用fairscale

--plugins ddp

大佬我也遇到了同款问题，改成ddp后报了这个错误 File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 86, in dev_collate return self.train_collate(batch) File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 53, in train_collate src = [b['src'] for b in batch] File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 53, in src = [b['src'] for b in batch] KeyError: 'src'

你的数据集格式有问题，参考下旧版本里面的example中的数据格式

[satisic]

试试改个参数，不用fairscale

--plugins ddp

大佬我也遇到了同款问题，改成ddp后报了这个错误 File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 86, in dev_collate return self.train_collate(batch) File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 53, in train_collate src = [b['src'] for b in batch] File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 53, in src = [b['src'] for b in batch] KeyError: 'src'

你的数据集格式有问题，参考下旧版本里面的example中的数据格式

感谢回复，我更改key名称后数据加载条到一半多线程又报错了
File "C:\ProgramData\anaconda3\lib\multiprocessing\reduction.py", line 60, in dump
ForkingPickler(file, protocol).dump(obj)
TypeError: cannot pickle '_thread.RLock' object

[lichengyang666]

试试改个参数，不用fairscale

--plugins ddp

大佬我也遇到了同款问题，改成ddp后报了这个错误 File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 86, in dev_collate return self.train_collate(batch) File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 53, in train_collate src = [b['src'] for b in batch] File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 53, in src = [b['src'] for b in batch] KeyError: 'src'

你的数据集格式有问题，参考下旧版本里面的example中的数据格式

感谢回复，我更改key名称后数据加载条到一半多线程又报错了 File "C:\ProgramData\anaconda3\lib\multiprocessing\reduction.py", line 60, in dump ForkingPickler(file, protocol).dump(obj) TypeError: cannot pickle '_thread.RLock' object

num_workers改成0试试

[satisic]

试试改个参数，不用fairscale

--plugins ddp

大佬我也遇到了同款问题，改成ddp后报了这个错误 File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 86, in dev_collate return self.train_collate(batch) File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 53, in train_collate src = [b['src'] for b in batch] File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 53, in src = [b['src'] for b in batch] KeyError: 'src'

你的数据集格式有问题，参考下旧版本里面的example中的数据格式

感谢回复，我更改key名称后数据加载条到一半多线程又报错了 File "C:\ProgramData\anaconda3\lib\multiprocessing\reduction.py", line 60, in dump ForkingPickler(file, protocol).dump(obj) TypeError: cannot pickle '_thread.RLock' object

num_workers改成0试试

okk了，主机用cpu模式太慢，pycham识别cuda.count()是有卡的，但pycharm连git运行sh文件就认不到卡pytorch_lightning.utilities.exceptions.MisconfigurationException: You requested GPUs: [0]
But your machine only has: []（1080ti不支持半精度的原因？还是说git没有认到anaconda的环境？）。

后面转到linux用4090没修改argument文件就能跑通

[WSChange]

试试改个参数，不用fairscale

--plugins ddp

大佬我也遇到了同款问题，改成ddp后报了这个错误 File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 86, in dev_collate return self.train_collate(batch) File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 53, in train_collate src = [b['src'] for b in batch] File "C:\Users\16771\OneDrive\桌面\t5-pegasus-pytorch-main\utils.py", line 53, in src = [b['src'] for b in batch] KeyError: 'src'

你的数据集格式有问题，参考下旧版本里面的example中的数据格式

感谢回复，我更改key名称后数据加载条到一半多线程又报错了 File "C:\ProgramData\anaconda3\lib\multiprocessing\reduction.py", line 60, in dump ForkingPickler(file, protocol).dump(obj) TypeError: cannot pickle '_thread.RLock' object

请问如何修改key的名称，是全部改成旧版本的示例数据的格式吗？