search

microsoft / table-transformer

Table Transformer (TATR) is a deep learning model for extracting tables from unstructured documents (PDFs and images). This is also the official repository for the PubTables-1M dataset and GriTS evaluation metric.
MIT License
2.01k stars 231 forks source link

Can I resume the training if I stop it? #131

Open linkstatic12 opened 10 months ago

linkstatic12 commented 10 months ago

If i run my PC for the epoch 0 training, stop the training and then change the epoch value to 1 in the json and start the training again does it take the weights from the epoch 0?

bsmock commented 10 months ago

Hi,

The model weights and current optimizer state is saved after every epoch. If epoch N completes and then training is interrupted after that, you can resume training at epoch N+1 by starting training with the flag --model_load_path /path/to/model.pth.

Let me know if that answers your question.

Best, Brandon

linkstatic12 commented 9 months ago

@bsmock It did. I was training the fintable dataset but you guys already did it.