'can I use Pytorch to weighted sample form datasets? I am going to train T5-model with multitask
I meet a problem when coding with Pytorch, I'd like to fine-tune a T5-modal using multitask. Now, I have two datasets with different size, I want to use temperature strategy to sample these two dataset, how can I achieve it with pytorch.
Any answers will be appreciated!Thanks~
the pytorch cat function just concatenate datasets without sampling which can't meet the requirement Can i use WeightedRandomSampler to achieve it?
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|