Hi I try to train question generating model, but the use_qag keeps on set to False although use_qag: True in textvqa_TAG_training.yml
The print-out on the terminal:
"return_info": True, "use_ocr": True, "use_ocr_info": True, "use_qag": False
I have tried to override using following code, but it is the same. python -m torch.distributed.launch --nproc_per_node 2 TAG/tools/run.py --tasks vqa --datasets m4c_textvqa --model m4c_split --seed 0 \ --config $config_dir/textvqa_TAG_training.yml \ --save_dir $save_dir \ training_parameters.distributed True \ dataset_attributes.m4c_textvqa.use_qag True
Hi I try to train question generating model, but the
use_qag
keeps on set to False althoughuse_qag: True
intextvqa_TAG_training.yml
The print-out on the terminal:
"return_info": True, "use_ocr": True, "use_ocr_info": True, "use_qag": False
I have tried to override using following code, but it is the same.
python -m torch.distributed.launch --nproc_per_node 2 TAG/tools/run.py --tasks vqa --datasets m4c_textvqa --model m4c_split --seed 0 \ --config $config_dir/textvqa_TAG_training.yml \ --save_dir $save_dir \ training_parameters.distributed True \ dataset_attributes.m4c_textvqa.use_qag True