Hu Xu

the memory consumption is impractical if not use fp16. sorry, the code is not well tested on fp32. make sure gpu is volta or ampire or rtx. On Wed, Aug...

it’s recommended to plug-in a standard trainer hugging face or pytorch lightning. fp32 are not well tested and some fp16 feature may not be fully disabled. On Thu, Aug 26,...

You need to fine tune on laptop aspect extraction data. On Thu, Mar 18, 2021 at 2:23 AM Vishnu ***@***.***> wrote: > After downloading "post-trained Laptop weights" (fine-tuned model). How...

it's on laptop domain but not on a particular task such as AE. On Fri, Mar 19, 2021 at 11:28 PM Vishnu ***@***.***> wrote: > You need to fine tune...

Make sure you can run the CMD in eval code from terminal first. Then the reason could be the semeval jar cannot be executed on new versions of java. There's...

put model predictions into a XML file that jar can only read. On Mon, Mar 22, 2021 at 3:26 AM Vishnu ***@***.***> wrote: > Thanks a lot for the quick...

https://www.yelp.com/dataset/download On Thu, Mar 18, 2021 at 6:58 AM logout ***@***.***> wrote: > Could you please provide another method to download the yelp dataset, as > the link provided in...

1. there's a mixed domain version [here](https://github.com/howardhsu/BERT-for-RRC-ABSA/blob/master/amazon_yelp.md), if you wish to have more domains. 2. post-training took one day of training on the original paper. I aim for single gpu...

https://github.com/howardhsu/BERT-for-RRC-ABSA/blob/master/pytorch-pretrained-bert/src/gen_pt_review.py Getting the raw text should be straightforward and we didn't release the code. On Thu, Aug 19, 2021 at 1:56 AM zjj0266 ***@***.***> wrote: > Thanks for your code！Could...

we didn’t release a fine-tuned model on a specific task. but you can easily tune on using semieval data. best, hu On Tue, Feb 23, 2021 at 5:56 AM Prasad...