search

airsplay / lxmert

PyTorch code for EMNLP 2019 paper "LXMERT: Learning Cross-Modality Encoder Representations from Transformers".
MIT License
923 stars 157 forks source link

Error in fine-tuning GQA for large datasets #118

Closed zilxv closed 8 months ago

zilxv commented 8 months ago

root@autodl-container-a54d4c9bab-1dc3f41d:~/autodl-tmp/lxmert# bash run/nlvr2_finetune.bash 0 nlvr2_lxr955 Load 86373 data from split(s) train. Start to load Faster-RCNN detected objects from data/nlvr2_imgfeat/train_obj36.tsv Traceback (most recent call last): File "src/tasks/nlvr2.py", line 150, in nlvr2 = NLVR2() File "src/tasks/nlvr2.py", line 34, in init self.train_tuple = get_tuple( File "src/tasks/nlvr2.py", line 21, in get_tuple tset = NLVR2TorchDataset(dset) File "/root/autodl-tmp/lxmert/src/tasks/nlvr2_data.py", line 73, in init img_data.extend(load_obj_tsv('data/nlvr2_imgfeat/train_obj36.tsv', topk=topk)) File "/root/autodl-tmp/lxmert/src/utils.py", line 45, in load_obj_tsv item[key] = np.frombuffer(base64.b64decode(item[key]), dtype=dtype) File "/root/miniconda3/lib/python3.8/base64.py", line 87, in b64decode return binascii.a2b_base64(s) binascii.Error: Incorrect padding root@autodl-container-a54d4c9bab-1dc3f60d:~/au