Hey guys, I already use TabRecSet to fine tune table transformer successful！
I find Table transformer didn't performed well in my task, then I found that there many images in my task are take photo by phone camera.We know table transformer was trained in PubTabs-1M, I think I need to use other datasets to fine tune it. So I tried and Table Transformer performed well in my task afted tuning by TabRecSet now, because there are many scence table images in this dataset.
There are also some conclusion in my practice for you to refer to：
- Just like @bsmock said in issue#108.I used 9k sample for fine tuing and model performed well
In some reason , I am not allow to upload my checkpoint. So I record my method in my Github table-transformer-finetuning
Hey guys, I already use TabRecSet to fine tune table transformer successful！
I find Table transformer didn't performed well in my task, then I found that there many images in my task are take photo by phone camera.We know table transformer was trained in PubTabs-1M, I think I need to use other datasets to fine tune it. So I tried and Table Transformer performed well in my task afted tuning by TabRecSet now, because there are many scence table images in this dataset.
There are also some conclusion in my practice for you to refer to：
In some reason , I am not allow to upload my checkpoint. So I record my method in my Github table-transformer-finetuning