微调出错

[tututututurana]

hydra.errors.InstantiationException: Error locating target 'training.trainer.Trainer', set env var HYDRA_FULL_ERROR=1 to see chained exception.
full_key: trainer

[bhack]

It is pytorch/tensordict#1048

[zsz00]

I use pytorch 2.4.1 and tensordict 0.5, I meet a same error, Do I need to upgrade to tensordict==0.6 and pytorch==2.5 ?
@bhack

[bhack]

#402

[zsz00]

I saw that, I upgrade tensordict to 0.6, it need pytorch to 2.5, To solve this problem, pytorch must be upgraded to 2.5 ??

[bhack]

They have upgraded the minimal pytorch version in the last tensordict release.

You can manually backport this pytorch/tensordict#1049 if you want to work with pytorch 2.4

[zsz00]

Well, I can't do that, so I'll just upgrade pytorch

[zsz00]

I upgraded to tensordict==0.6 and pytorch==2.5, but meet this error :
[rank0]: hydra.errors.InstantiationException: Error in call to target 'training.trainer.Trainer':
[rank0]: InstantiationException("Error locating target 'training.dataset.vos_dataset.VOSDataset', set env var HYDRA_FULL_ERROR=1 to see chained exception.\nfull_key: data.train.datasets[0].dataset.datasets0")
[rank0]: full_key: trainer