LMC63 comments

Repositories
Issues
Comments

Results 4 comments of


                                            LMC63

why specify `ignore_index=0` in the NLLLoss function in BERTTrainer?

see #32 change `self.criterion = nn.NLLLoss(ignore_index=0)` to `self.criterion = nn.NLLLoss()`

RuntimeError: you must first build vocabulary before training the model

An importance step before running code is to unzip NTCIR-12 dataset and extract all tarballs under the sub-directories by `for x in *.tar.bz2; do echo $x; tar xjf $x; done`

Bounding boxes are displaced from math regions

NOTE: If you find the bounding boxes are displaced from math regions, it is because the document image that you have rendered is of different size than the one used...

Visualization of a 12-layer BERT for sentence encoding/embedding

Amazing work about visualization!