search

poloclub / tsr-convstem

High-Performance Transformers for Table Structure Recognition Need Early Convolutions
https://arxiv.org/abs/2311.05565
MIT License
39 stars 2 forks source link

Could not find 'dataset/mini_pubtabnet #3

Open liJay010 opened 6 months ago

liJay010 commented 6 months ago

Why does running the configuration "make experiments/r18_e2_d4_adamw/.done_train_structure" not result in an error, but running configurations within the convolutional stem, such as "EXP_cs_c384_e4_d4_nhead8_adamw", produces the error "In 'main': Could not find 'dataset/mini_pubtabnet' "?

Available options in 'dataset':
        pubtabnet
Config search path:
        provider=hydra, path=pkg://hydra.conf
        provider=main, path=file:///rxhui/lxj/workspace/table_rec/tsr-convstem-main/configs
        provider=hydra-colorlog, path=pkg://hydra_plugins.hydra_colorlog.conf
        provider=schema, path=structured://

Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace.
In 'main': Could not find 'dataset/mini_pubtabnet'
ShengYun-Peng commented 6 months ago

Thanks, @liJay010! Our latest work, UniTable: Towards a Unified Table Foundation Model, has been fully released at https://github.com/poloclub/unitable/tree/main. I highly recommend testing on UniTable as it achieves SOTA performance on four of the largest table datasets. If you are looking for mini_pubtabnet, which contains 20 samples extracted from PubTabNet, it's also available in UniTable's repo at https://github.com/poloclub/unitable/tree/main/dataset/mini_pubtabnet.