Wonseok Hwang comments

Results 5 comments of


                                            Wonseok Hwang

Annotation Issues [Please report any annotation errors here, thanks!]

Hi! I would like to inform you small inconsistency `db_id = formula_1` table in `tables.json`, The order of tables presented in "table_names" & "table_names_original" fields are different making in difficult...

Share FUNSD trained model weights

Hi @AruniRC Unfortunately, as I have moved to a new position, I'm no longer able to access the trained model. Please use [`funsd.1.5layers.train.yaml`](https://github.com/clovaai/spade/blob/master/configs/funsd.1.5layers.train.yaml) to train the model. Wonseok

Training on FUNSD: Cuda out of Memory on GPU with 12Gb memory.

Hi @asidharth019 You may turn off `relative_attention` (see [this comment](https://github.com/clovaai/spade/issues/2#issuecomment-915036284) ) Or, you may use smaller encoder, for example, bert-base-multilingual-cased-(3 or 4)layers (please refer to [this comment](https://github.com/clovaai/spade/issues/13#issuecomment-1007823265)) Good luck!

Training on FUNSD: Cuda out of Memory on GPU with 12Gb memory.

- The reason why the FUNSD output format is different compared to CORD is because of the difference in the depth of information. Please refer to the [Table 2](https://arxiv.org/pdf/2005.00642.pdf). Also,...

Training on FUNSD: Cuda out of Memory on GPU with 12Gb memory.

Hi @asidharth019 Sorry for being late in reply. You may increase the number of training epochs? As far as I remember, you should get near 100% accuracy on training set....