Lei Li

tinybert task pred_distil dev data loader is overwritten by mnli-mm

3

According to your [code](https://github.com/huawei-noah/Pretrained-Language-Model/blob/a8a705e9c8c952e078b45d1091d3f0ed161483d8/TinyBERT/task_distill.py#L1069) in `task_distill.py`, the evaluation on MNLI-mm dataset will overwrite the original dev loader of MNLI-m, thus the best checkpoints are selected based on the MNLI-mm instead...

Hi, thanks for the nice paper and neat code, well done. After read the paper and check the code, I was confused about several lines in `model_task.py`, from 306 to...

Question about the Qformer training

2

Hi, big thanks for your great work and open-sourced code & weights. I am fine-tuning/continuing training the checkpoints and hope you can kindly share some knowledge about the design &...

how many training instances are used?

1

Hi, thanks for your great project! I am wondering how many training dataset instances you are used, such as COCO, OCR-VQA and A-OKVQA, did you just transform the original dataset...