Checkout the example colab notebook in the repo that shows how to finetune the model on the ChartQA Dataset. The training code is optimized such that you can train it on a GPU with 24 GB of memory.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results